Biologists like to debate whether life can be reduced to chemicals or, conversely, whether chemicals can be used to assemble a new organism from scratch. In his 1911 book, The Mechanism of Life, Stéphane Leduc (who coined the term “synthetic biology”) argued that life can be fully explained by its molecules and interactions between them:

Life is not a substance but a mechanical phenomenon; it is a dynamic and kinetic transference of energy determined by physico-chemical reactions.

A few years later, naturalist John Burroughs, seeking to refute Leduc, argued the opposite: that living organisms exhibit additional “organizing tendencies” that chemicals alone cannot explain. Writing in The Breath of Life, published in 1915, Burroughs asserted that:

We can analyze the material elements of an organism, but we cannot synthesize them and produce the least spark of living matter.

Who is correct? My gut nudges me to side with Leduc, but it’s also been more than 100 years since Leduc posed his argument, and nobody has made life by mixing together inert molecules.1

The truth is, cells are ridiculously complicated things. Even the simplest ones have millions of proteins, metabolites, RNA and DNA, all tightly packed and jostling around within a lipid membrane. Each protein collides with trillions of water molecules per second, and glucose molecules ping-pong around at 460 miles per hour. Many cellular reactions are also stochastic, or “noisy,” meaning that chance encounters between molecules cause their timing and outcomes to vary even in identical cells.

In one experiment, published in 2002, researchers studied genetically identical Escherichia coli cells carrying genes for two fluorescent proteins, one cyan and one yellow. Each was placed next to an identical promoter to which the LacI repressor could bind, and thus “shut down” the gene’s expression. The researchers photographed the cells under a microscope and measured the fluorescence from each protein separately.

Despite being genetically identical, the cells differed in brightness and color balance. Some cells made more cyan than yellow, and others made more yellow than cyan. These differences were often substantial, with a standard deviation of 40% of the mean.

Yet, despite the “noise” of biochemistry, life persists. It has found a way to flourish within this complexity and, for billions of years, cells have spawned new cells by turning simple molecules, such as sugars and amino acids, into more complex assemblages, beating back the tides of entropy. Or, as the Austrian physicist Erwin Schrödinger wrote in his classic 1944 book, What Is Life?, cells refuse to succumb to chaos “by eating, drinking, breathing and (in the case of plants) assimilating.”2

Leduc may have been right when he said that life can be explained purely by chemistry and physics, but it’s also clear that we can’t yet recreate the reactions necessary for life purely from synthetic chemicals. Perhaps the reason is that we don’t understand how life adapts and persists despite its noise and complexity. We may be able to collect all the ingredients needed to make a cell, but we are as blind as amateur cooks without a recipe when it comes to assembling them.

To create any given cell, we must first figure out what it is made from. How many atoms—and how much energy—does it take to build a new one? By tallying the parts, it becomes clear why humans can disassemble cells and understand each piece in isolation but struggle to put them back together and make them roar to life.

What Cells Are Made From

Let’s build an inventory for E. coli, the little bacterium that is the subject of more than 100,000 published research papers. Scientists have studied many types of E. coli, from “flesh-eating” variants to engineered strains that act as living “camera film.”3 For our purposes, let’s take a median example of this species and pick it apart. What is it made from?

About 70% of its mass is water. The other 30%—the “dry mass”—is everything else: DNA, RNA, proteins, lipids and so on. DNA makes up just 3% of dry mass, yet encodes all the information needed for the cell to grow, divide, adapt and evolve.4 The bulk of this dry mass, 55%, is protein, the large molecules that catalyze reactions and help control which genes switch “on” or “off.”5

Proteins make up the bulk of the dry mass for a few reasons. First, there are lots of them, and each protein is heavy. A single E. coli cell has about two million proteins floating around, each built from hundreds of amino acids. If the cell were blown up to the size of a tennis court, each protein would be roughly the size of a tennis ball.

Why are proteins so big? For one, large proteins fold faster than small ones due to cooperativity, meaning, basically, that one part of a protein coaxes another part to snap into place.6 Enzymes—proteins that speed up chemical reactions—need the extra bulk to build pockets, or cavities shaped in such a way that only certain molecules can enter and reach the active site. The enzyme that attaches tyrosine to its tRNA, for example, prefers tyrosine over phenylalanine—a molecule that differs by just one oxygen atom—by a ratio of more than 100,000-to-one, simply because its “pocket” has evolved into a structure ideally shaped for tyrosine.

Although cells have millions of proteins floating around, there are only a few thousand types. Some proteins are abundant, with millions of copies, whereas others are rare; for example, there is only a handful of RepA proteins, essential for DNA replication, present in the cell at any one time. The distribution of protein counts is also long-tailed, meaning that the 100 most abundant protein types account for more than half of all protein molecules in a cell.7

In the 1970s, most biologists thought EF-Tu, a protein that delivers tRNAs to ribosomes during protein synthesis, was the most abundant. This made intuitive sense because every E. colihas tens of thousands of ribosomes, and EF-Tu helps to quickly “feed” amino acids to them.

But in 1979, a study found that the most abundant protein is actually Braun’s lipoprotein, or Lpp, with more than 700,000 copies per cell. Lpp is a small protein that staples the cell’s outer membrane to its peptidoglycan layer, thus preventing it from blistering and leaking out viscous cytoplasm.8

After proteins, about 20% of the cell’s dry mass is RNA, of which a majority is ribosomal RNA.9 Each ribosome is two-thirds RNA by mass; these big machines have long threads of RNA inside, about 4,500 nucleotides in length. If you were to take the strands of RNA from a single ribosome and stretch them out, end to end, they would measure about 2.1 micrometers in length, longer than the entire cell.

Finally, lipids make up about 10% of a cell’s dry weight. There are about 30 million lipid molecules in each E. coli, beautifully packed into a double layer that stretches around the entire cell. If this membrane were pressed into a flat sheet, it would cover about 6 µm² of space. It would take 10,000 of these sheets, neatly sewn together, to cover the period at the end of this sentence.

Clearly, then, a cell has many molecules. How much energy is needed to build them?

One might assume the answer is “a lot”! After all, a cell has millions of proteins, each assembled from hundreds of amino acids.10 For currency, cells pay for this work using ATP, a small molecule that, when broken, releases energy to power chemical reactions. Adding just one amino acid to a protein costs about four ATPs. Energy is also burned to build DNA, RNA, lipids and every other molecule.11 Surprisingly, though, the total amount of energy needed to make a cell is very small (at least from our human-sized perspective).

Scientists have used biochemical calculations to estimate the energy required. By measuring how much food is eaten and how many new cells are created by bacteria in a flask, for example, they can then build models, based on known metabolic pathways, to predict how much ATP the bacteria have spent. And for an E. coli weighing about 0.3 picograms, producing another cell’s worth of material costs about 10 billion ATPs. These ATP molecules (and many macromolecules in the cell itself), in turn, are made from two to three billion glucose molecules. (A standard cube of sugar weighs four grams, and would be enough to make something like five trillion cells.) Lifting a small apple one meter off the ground is roughly the energy needed to build one billion cells.12 The creation of life is highly efficient.

These are just estimates, of course, and the numbers get fuzzy based on assumptions.13 Some energy is always lost as heat, for instance, and the actual ATP cost goes up or down based on cell size, nutrients, oxygen supply and growth rates.14

In the laboratory, we already have synthetic tools to build parts of cells. It is easy to assemble amino acid chains using solid-phase peptide synthesis, for example, even though long proteins are still difficult to make efficiently by chemical synthesis. The laboratory process is also woefully inefficient. Chemists first attach an amino acid to a little bead, chemically join the next amino acid, wash away the reagents, and then repeat this hundreds of times. Cells, for comparison, do all of this in minutes, for orders of magnitude less energy, while building thousands of different proteins using solely atoms taken from their environments.

But now that we have a nearly complete list of the molecules in a cell, along with estimates of the energy needed to assemble them, we can finally ponder ways to arrange them and judge how close we might be to the possibility of “creating life.” The reason nobody has succeeded so far, I suspect, is that we haven’t yet mastered the fault tolerance or exquisite regulation required to do so.

Woeful Dynamism

It’s a short walk from my Berkeley apartment to a grocery store, where I can find the same ingredients the chefs at The French Laundry and other Michelin-starred restaurants cook with. Even so, I can’t assemble these ingredients into a prize-winning dish. I don’t know how to finesse them, or feel when a sauce has thickened just enough to pull it off the burner. More importantly, if an ingredient goes missing, I have no way to adapt or modify a dish on the fly, using my understanding of saltiness and sweetness, sourness and bitterness alone. A list of ingredients is not a recipe, and a recipe does not imply mastery of cooking.

The most amazing thing about cells, then, is not that they are packed with millions of proteins and other molecules and still manage to work. Rather, it’s that (as in the case of E. coli) they can re-make themselves every 20 minutes; they are producing millions of machines and adapting them on the fly and in the face of enormous uncertainty.15 The most amazing thing about life is that it persists despite the noise.

Consider one small part of a cell: DNA replication. In a 1998 paper in Cell, a pair of MIT biologists did a thought experiment in which they scaled up DNA to a width of one meter. In that scenario, DNA polymerase (the enzyme responsible for copying it) would be the size of a FedEx truck. Copying the E. coli genome would require two of these trucks, each moving in opposite directions at speeds of 375 miles per hour. Over the span of 40 minutes, these trucks would deliver five million packages and make just one error every 106 miles.16

At the same time that FedEx truck is speeding by, thousands of ribosomes are busily making proteins, enzymes are converting millions of molecules each second, and the membrane is adding lipid molecules to expand, preparing to make a daughter cell. Seemingly, this would offer thousands of opportunities for something to go wrong. At this scale, and these speeds, even rare errors can accumulate and kill the cell. And yet life persists. It remakes itself even when things inevitably go wrong. If human builders are ever to rival the complexity and dynamism of a cell, then we not only need to find the ingredients, but also arrange them in ways suited to extreme speeds, forces and noise.

Two factors are especially critical.

First, populations of cells have a high fault tolerance, partly because they contain so many cells and each cell differs from its neighbors. A threat that kills some cells may spare others. These differences, when multiplied across millions of cells in a group, are one way that cells “adapt” to uncertainty and persist.

For a 2020 study, researchers deleted two genes encoding enzymes required for biosynthesis of isoleucine, an essential amino acid used to make proteins. Without it, cells will eventually die. The researchers knocked out these genes, put the sick bacteria into little tubes without isoleucine, and then waited. For the next 70 hours, the bacteria appeared dead. But then, slowly, some of the cells began to divide.

After sequencing these cells, the researchers found that about a dozen mutations had randomly arisen in the population, mainly in a gene encoding MetC, an enzyme responsible for making another amino acid called methionine. The mutations shifted the cell’s metabolism slightly, enabling them to make an isoleucine precursor instead.17 The few mutated cells from that initial population had inadvertently hit upon a solution to their molecular dilemma by rewiring a separate pathway, thus allowing the population to survive.18

Second, cells are robust despite uncertainties because they can, in some cases, sense relative changes while ignoring absolute ones. Such sensitivity proves useful when the environment changes and a cell must adapt without re-assembling molecules.

Consider an E. coli cell foraging for food. Little protein sensors embedded in its cell membrane detect the local concentration of food particles. As the cell spins its flagella and moves forward, the sensors take measurements in real time. If the food concentration dwindles, the cell is more likely to “tumble” and swim in a new, potentially more promising direction. But if the concentration of food goes up, the cell is less likely to tumble, and tends to keep swimming forward longer.

This is called chemotaxis, and the most amazing thing about it is that, within certain concentration ranges, the cell’s response depends on relative rather than absolute concentrations. The cell instead computes fold-changes, easily tuning its response to environments with much more or less food.

Regardless of whether the concentration of food rises from one to three, or from 10 to 30, the cell “processes” either change as simply a three-fold increase. Responding to a ratio, rather than absolute numbers, allows the bacterium to find more food regardless of its internal noise or huge changes in its environment.19 The bacteria are effectively judging changes against a baseline, enabling the same sensors to work across environments with very different concentrations. They are “compressing” their environment.

(Chemotaxis is so sensitive that E. coli can swim toward higher concentrations of an amino acid even when it is as dilute as one milligram dissolved in 2,300 liters of water.)

These examples are relevant to efforts to build a cell. It is not enough merely to understand which ingredients to include at which concentrations, but also to note that these ingredients will not work in all environments, and must be able to adapt to an extreme range of situations. Instead of trying to build “a” cell, then, we ought to build many variants, and then let evolution select the design best suited to life.20 Once a cell can reproduce with a selection mechanism, evolution can improve it.

Humans have a borderline-obsessive desire to define parameters precisely, to understand every knob in a machine and fine-tune them perfectly. But cells do not care. They deliberately vary their parameters and dial knobs up or down to create variation. When it comes to science, humans care about understanding, whereas cells care about survival, and these goals can conflict.

Returning to the Leduc–Burroughs debate, building a viable cell probably doesn’t require a mysterious organizing force or vitalist “spark.” Sensing, adjusting, repairing and reproducing are all tasks carried out by molecules and the physical interactions between them. But it’s also true that our understanding of molecular interactions remains a bit fuzzy, and cannot fully explain cells’ ability to adapt on the fly. Molecules may be necessary to explain life, but they’re also just one small part of the recipe.

Niko McCarty is founding editor of Asimov Press.

Thanks to Noah Olsman, Brady Huggett, Devon Balwit and Merrick Pierson Smela for helpful comments. Header image by Ella Watkins-Dulaney.

Did you know that the modern micropipette was invented by a German postdoc so frustrated with mouth pipetting that he ditched his lab for two days until could return with a prototype? Or that an 18th-century scientist sewed tiny taffeta pants for male frogs to determine whether or not semen was essential for fertilization? Or that an Italian physician spent 30 years weighing himself in a specially designed chair before and after eating, sleeping, and excreting to figure out exactly what accounted for the discrepancies he noticed?

These stories and more, from how a machine using salvaged meat-slicing blades chopped up 6,000 glowing jellyfish per day and led to the discovery of GFP to how grabbing the wrong bottle while studying rooster sperm lead to modern cryopreservation, appear in our third book, Making the Modern Laboratory, now available for pre-order. Over 504-pages, 36 authors explain the origins of our most dependable and insightful research tools, making it a fine gift for both the science-curious and the working scientist.

We announced this book one year ago, and getting it across the finish line has been a labor of love. We’re thrilled by the final result and hope you will be too.

This project couldn’t have happened without Ella Watkins-Dulaney’s initial vision and years-long determination to see it through. The book was typeset by Jake Brussel Faria, a designer in New York City, fact-checked by Samuel Clamons and Nicholas Porter, copyedited by Devon Balwit, and edited by Niko McCarty with contributions by Xander Balwit. Making the Modern Laboratory was printed at Grafiche Veneziane, an artisan press in Italy.

We also thank all the writers who contributed essays: Ulkar Aghayeva, Patrick Almhjell, Amelie Andreas, Xander Balwit, Saad Bhamla, Kevin Blake, Karol Buda, Alfred C. Chin, Evan DeTurk, Or Doni, Alex Federation, Alicia J. Kaestli, Sotirios Karathanasis, Andrew J. Konecny, David Lang, Matt Lubin, Niko McCarty, Roberta McLain, Owen Meadows, Lucas Mergan, Andreas Metousis, Luke Mitchelson, Corrado Nai, Mahlaqua Noor, Merrick Pierson Smela, Sachin Rawat, Logan E. Savidge, Smrithi Sunil, Matthew Bryce Taylor, Alex Telford, Nicolas Tilmans, Nehal Udyavar, Gerrit van Rensburg, Donna Vatnick, Ella Watkins-Dulaney, and Spencer Wright.

Pre-Order

Making the Modern Laboratory is priced at $49 and worth every penny. We’re offering free shipping in the United States. With a hard cloth cover, it is printed in full color on Arena Natural Rough, a textured, uncoated paper made by Fedrigoni. The book is finished with colored endpapers, thread-sewn binding, and a squared spine. Each copy is shrink-wrapped for safe shipping, and deliveries are expected in November.

The book was set in typefaces designed by Charles Mazé for Abyme, by Seb McLauchlan for Dinamo, and by Eurostandard in collaboration with Maximage. Its three typefaces—Mercure, ABC ROM and OCR-X—represent threads of typographic history, with each roughly matching a period described in the text. This means the letters printed throughout the book are close cousins of the fonts that the scientists and inventors being described would have read and printed in as they worked.

We consider international shipping for bulk orders, and are offering paid subscribers to this newsletter 25% off their order. Please email editors@asimov.com for details.

Pre-Order

Contents

The molecular biology laboratory is filled with things that most scientists take for granted. Few people know the history of the centrifuge, GFP, or the model organisms they regularly experiment on. We wrote this book to explore how these objects became staples in the lab, hoping that a closer look at the mundane might reveal something about scientific progress. And indeed, we found that even the simplest tools often hide stories of competition, persistence and chance.

The essays span five sections: Benchtop, Molecular Biology, Measurement & Analysis, Model Organisms and Computation & Automation. Each essay focuses on a single tool, technique or organism. You can read the book cover to cover, or open to an essay at random and quickly learn something unexpected.

The book also says something about the future. Even before we assembled these stories, we noticed that if you compared the molecular biology laboratory of today to those from the 1960s, they look basically the same. As we wrote in our original announcement:

… If you were to revisit photos of Howard Berg’s cramped Harvard lab, where the details of bacterial chemotaxis were first worked out, or Sydney Brenner’s Cambridge lab, where they cracked the genetic code, you’d recognize almost everything you saw …

In fact, when we began the book, we planned to use stories of the past to predict what laboratories might look like in the future. But we quickly abandoned that goal because we realized that just about every technology used today arose from unforeseeable breakthroughs and persistent tinkering with methods and materials already in use.

First, we saw that technologies often advance in leaps. A field slowly climbs a curve of progress, until an unexpected scientific discovery opens an entirely new curve, steeper than the last, resetting what is considered feasible.

Polymerase chain reaction, or PCR, is a classic example. Early versions required researchers to move samples between temperatures and replenish enzymes destroyed by the heating steps. The technique was tedious. Adapting Taq, a heat-stable enzyme from a bacterium first isolated in Yellowstone National Park, made PCR much easier to automate (202–206). CRISPR likewise emerged from basic research into bacterial immune systems and rapidly became a widely used gene-editing tool (452).

Second, we realized that one cannot predict what future laboratories will look like simply by choosing the technologies that seem optimal today.

As we mentioned earlier, Heinrich Schnitger built that prototype of the micropipette in 1957 largely because he was sick of mouth pipetting. But other pipette designs already existed at that time! The widespread adoption of his design depended on practical details as well as invention: while a thumb-operated plunger made it convenient, a deal with Eppendorf helped scale it (17–19). Many promising technologies never spread because they are difficult to use, lack funding or fail to reach people who can manufacture and promote them.

Once a tool becomes familiar, it quickly gains staying power. Conventional thermocyclers, for example, still heat and cool samples in metal blocks. Researchers have developed photonic alternatives that use light to heat samples and can offer faster cycling and lower energy use. Yet conventional machines remain in most laboratories. Scientists have good reasons to value familiar, reliable methods—established protocols, for example—so a faster instrument on its own may not be enough to force change.

Thus, while offering this book as a historical document, we think that these stories can help us better understand how to bring new tools into existence and hasten the long arc of progress.

Clinical microbiologist Frank Koontz’s career bridged two major eras of microbiology, split in 1995 by the sequencing of the first two bacterial genomes. The pre-1995 era was the time of old-school or “basic” microbiology, when microbiologists differentiated between closely related strains by sight (wrinkled or smooth colonies?) and smell (caramel or bread crust?). The second introduced, in Koontz’s words, “genetic frickey frackey methods.”

Koontz was an unabashed adherent to the old school. Shortly before his passing in 2001, he said:

When we’re doing these identifications by genetic methods and genomic changes, they say, ‘Oh wait a minute. This organism is different than that. While they produce the same disease, they look exactly the same on the plate, they produce the same biochemical reactions, but they’re different in [x], so they give it a new name.’ … I’ve just gone totally nuts on this stuff and I quit using the genus names.

For Koontz, small differences in DNA didn’t offer strong enough reason to rename microbes that looked and behaved alike clinically. Koontz was not alone in this belief, either. Taxonomy has long been regarded as an old-fashioned, harmless pursuit more suited to the 18th-century world of Linnaeus than our modern world of high-throughput, supercomputer-powered biology. But the advent of genomic sequencing launched taxonomy into the 21st century. Today, scientists can quantify the evolutionary relatedness of two organisms directly from the sequence of “letters” in their DNA, RNA, and a few proteins — no squinting or sniffing required.

Over the last 25 years, genomic methods have revealed that the tree of life is weirder and more tangled than ever imagined. They have shown, as a frustrated Koontz noted, that thousands of microorganisms thought to belong to one genus actually belong to another. But perhaps the most seismic discovery of the genomics era has been the revelation that horizontal gene transfer (HGT) — in which organisms trade genes with members of other species — is far more widespread than previously assumed. HGT has demonstrated that the branches of the tree of life do something scientists hadn’t thought possible: they criss-cross and merge.

This criss-crossing has blurred the lines separating microbial species and complicated rules built around those boundaries. In the United States, whether an engineered microbe faces regulatory review can depend on whether its added DNA comes from a different genus. This distinction can affect the deployment of microbes to recover valuable metals or detect pollution in the environment. But because microbes swap so much DNA, the boundaries between species are often difficult to untangle.

In his 1774 edition of Systema Naturae, Carl Linnaeus placed microbes into a taxonomic group he called “Chaos”; more than 250 years later, the name seems even more appropriate.

Molecular fingerprints

Great scientists throw off ideas like huskies shed fur. In September 1957, Francis Crick gave a talk at the annual symposium of the Society for Experimental Biology, published a year later as “On Protein Synthesis.” After musing about protein folding and making educated guesses about error rates, Crick revealed one of his most profound ideas:

Biologists should realize that before long we shall have a subject which might be called ‘protein taxonomy’—the study of the amino acid sequences of the proteins of an organism and the comparison of them between species.

Taxonomy is the study of classifying organisms. For centuries, naturalists have organized living creatures into hierarchical, tree-like systems that group species with similar features: mammals on one branch, for example, and birds on another. Charles Darwin’s key insight was that these similarities were due not to the whims of a Creator but to descent from a common ancestor. Closely related species, like closely related people, look similar because they share a relatively recent common ancestor. Taxonomy seeks to determine whether two organisms are “sister-species,” first cousins, or fourth cousins thrice removed.

When it came to deciding where to place a species on the tree of life, those studying animals and plants found plentiful morphological detail to measure and compare; bacterial taxonomists faced a more daunting task. Because microbes are simple morphologically, especially when viewed through the rudimentary microscopes of the 19th and early 20th centuries, taxonomists could often compare little beyond a species’ shape (sphere, rod, or spiral), habitat, and basic metabolism. Early bacterial taxonomists thus created classification systems that were confused, contradictory, and mostly inaccurate.

For example, the 19th-century German botanist Ferdinand Cohn, one of the founders of modern microbiology and the first bacterial taxonomist, was convinced that bacteria belonged to the plant kingdom. He saw similarities in the shape, movement, and division of bacteria and organisms then classified as blue-green algae (now recognized as cyanobacteria). Because of such errors, it became almost routine for new generations of scientists to tear down prior taxonomic systems; some even took an axe to their own work.

In 1941, for example, the microbiologists C.B. van Niel and Roger Stanier published an influential critique of Bergey’s Manual of Determinative Bacteriology — the standard taxonomy book for 20 years —and proposed a new system. Then, 20 years after that, they rebutted their own rebuttal, saying their “elaborate taxonomic proposal” was something “neither of us cares any longer to defend.” One hundred years after Darwin published the Origin, bacterial evolution was still a mystery.

Then came Crick’s idea about “protein taxonomy,” or using the sequenceof proteins at the molecular level, to map relationships between species. Although Crick never actually followed up on this, other biochemists did. In the 1960s, Linus Pauling and Emile Zuckerkandl realized that by using electrophoresis, a method which uses electrical charge to separate molecules by size, they could generate protein “fingerprints” that approximated the information in a complete protein sequence. Pauling and Zuckerkandl hypothesized that since proteins (and the genes encoding them) steadily accumulate mutations over time, the number of differences between the same protein in different species should be proportional to the amount of time having passed since those species diverged. They dubbed this “the molecular clock hypothesis.”

The pair focused on hemoglobin, the protein in red blood cells that carries oxygen. Pauling and Zuckerkandl documented close similarity (meaning a small number of mutations) between the fingerprints of human hemoglobin and chimpanzee hemoglobin, and less similarity (or more mutations) between human hemoglobin and orangutan hemoglobin. These two men also found more similarity among primate hemoglobins than between hemoglobins from cows or pigs. These patterns they uncovered offered a new basis of support for evolutionary trees that had been drawn the conventional way. But they took their discovery one step further, predicting not only that this approach could reinforce existing trees, but also that “branching of molecular phylogenetic trees should in principle be definable in terms of molecular information alone.”

The biophysicist Carl Woese was intrigued by this budding new field called molecular phylogenetics. In 1969, he wrote to Crick about his ambitious goal to “unravel the course of events” leading to the origin of the first single-celled microbes. He aimed to extend the current understanding of evolution “backward in time by a billion years or so … using the cell’s ‘internal fossil record.’” Hemoglobin wouldn’t work for this kind of study because it is not shared by all the organisms Woese wanted to compare. Woese needed something older, more basic, and universal to all forms of life. “The obvious choice of molecules here lies in the translation apparatus,” he told Crick. “What more ancient lineages are there?”

Woese was referring to the ribosome. Like a molecular 3D printer, the ribosome takes in genetic instructions and raw amino acids to build proteins. Woese focused on the ribosome for two reasons: first, it was found in all forms of life; and second, its mixture of protein and structural RNA parts suggested that it was very ancient. Woese settled on a component of bacterial ribosomes called the 16S ribosomal RNA, or rRNA. Animals, plants, and fungi have a close variant called 18S rRNA.

In the 1970s, Woese began assembling the 16S sequences of any microorganism he could get his hands on. The method was laborious and dangerous: Woese first cultured microbes with radioactive P-32, then used highly flammable Varsol (a type of general-purpose solvent) to keep the electrophoresis paper from overheating under the high voltage and amperage needed to separate the tiny ribosomal fragments. After taking X-rays, Woese viewed the film on a floor-to-ceiling lightbox. He then compared those rRNA “fingerprints” much as Pauling and Zuckerkandl had compared the fingerprints for hemoglobin, and discovered that a subset of unusual bacteria living in extreme environments weren’t bacteria after all. Their 16S rRNA sequences were as different from those in bacteria as they were from those in animals and plants — apparently representing a “third kingdom.” That group is now known as the archaea.

Shortly thereafter, geneticist Norman Pace realized that he could use the same sequencing techniques to shed light on the “uncultured majority” of bacteria and archaea unable to be cultured under laboratory conditions. Specifically, Pace sequenced rRNA extracted from the pink agglomerations of microorganisms that were found in the near-boiling waters of Yellowstone National Park’s Octopus Spring but resisted attempts to be grown in the laboratory. The results revealed that these pink agglomerations were actually a community dominated by three species — two bacteria and one archaeon — previously unknown to science. His team’s rRNA sequences alone were enough to place these species on the tree of life.

In the decades since, sequencing the uncultured majority has dramatically expanded the phylogenetic tree. A 2016 study led by microbiologist Jillian Banfield compared ribosomal proteins from more than 3,000 genomes to map diversity across bacteria, archaea, and eukaryotes. The resulting tree showed just how much diversity earlier depictions had missed. In Banfield’s paper, most of the tree is made up of bacterial species, many of which lack even a single isolated representative: scientists know them solely from DNA recovered from mixed microbial communities, rather than from a species grown and studied separately in the laboratory. In a humbling reversal from earlier evolutionary trees, all eukaryotes — from animals and plants to single-celled amoebae — fit on a tiny little twig in the corner, overshadowed by an arborescent bush of microscopic bacteria.

Whereas the earliest bacterial phylogenetic trees, based on microbes’ crude physical features, were imprecise, microbiologists today can determine which species a strain belongs to by using a metric called average nucleotide identity, or ANI, which is the mean percentage of identical nucleotides across every gene shared by two genomes. Microbiologists generally use an ANI of ≥95% as the genomic boundary line between bacterial species. If genomic comparisons reveal that a strain historically thought to belong to one species actually has <95% ANI to that species, then it must be renamed.

Such taxonomic revisions are often contentious, but they are part of the routine work of updating old systems with newer, better data. However, during every one of these conversations about bacterial species, there is an elephant in the room: horizontal gene transfer (HGT).

The tree of life is more like a net

Microbiologists first thought that, apart from minor tweaks, widespread adoption of whole-genome sequencing would validate and reinforce Woese’s three-limbed tree. But it quickly became apparent that something strange was happening: genes from the same genome could create phylogenetic trees with different topologies. It was like discovering that while human hemoglobin was most similar to chimpanzee hemoglobin, human insulin was most closely related to a snake’s. Each gene seemed to have its own evolutionary history. What explains this?

During reproduction, genes are copied and passed down vertically from parents to offspring. As a result, genes from different species are largely isolated from one another. In his 1995 book River Out of Eden, Richard Dawkins uses the metaphor of a river of DNA to describe how genetic information flows through time. Speciation is like a fork in the river, after which genes that once freely intermingled rarely recombine. Dawkins calls this process “the long goodbye.” While that may be true for genes of most multicellular eukaryotes, for microbial genes it is more like a quick “see ya in a bit.”

The first hint that bacteria were doing something different came in 1928, when British bacteriologist Fred Griffith showed that material from dead pneumococci could transform living bacteria. In 1944, Avery, MacLeod, and McCarty identified that material as DNA. Those genes were not transferred vertically from parents to offspring, but horizontally between individuals. Later work by Joshua Lederberg would reveal two more kinds of HGT: conjugation, also known as “bacterial sex,” in which two living microorganisms make direct contact and transfer DNA; and transduction, when a bacteriophage (a virus that infects bacteria) accidentally packages and moves bacterial DNA from one cell to another. For bacteria, Dawkins’ river of DNA is more like a flooded river delta, where tributaries that had separated hundreds of millions of years earlier intermingle once again.

HGT also explains why some bacterial strains carry thousands of genes that other strains from the same species do not, and vice versa.1 A 2015 review looked at more than 2,000 full Escherichia coli genome sequences, and found that each genome contained about 5,000 genes. But of those 5,000 genes, only about 3,000 were widely conserved, or found across nearly all the strains; the other 2,000 genes were more unique and varied a lot between the strains of this same species. Collectively, the 2,000 E. coli strains they looked at encoded about 89,000 different genes.

Thanks to HGT, bacteria can acquire important new functions in one fell swoop. Nowhere is this more apparent than in the spread of antibiotic resistance genes. In 1963, Japanese bacteriologist Tsutomu Watanabe alerted the world to the alarming discovery that genes conferring resistance to multiple antibiotics had spread horizontally between Shigella and E. coli species. He would also document similar cases between other species and genera.

More recently, a study using metagenomic sequencing methods found identical resistance genes with 100 percent nucleotide identity shared between non-pathogenic soil bacteria and human pathogens isolated from hospitalized patients, including Pseudomonas aeruginosa, Acinetobacter baumannii, and Salmonella enterica. If those genes had spread vertically from these species’ common ancestor, then we would expect a degree of difference proportional to the evolutionary distance between them. But identical genes in such divergent species can only be explained by recent HGT. These genes had criss-crossed the tree of life.

For most of the 20th century, HGT was thought to be either confined to special categories of genes (such as antibiotic resistance genes) or too rare to matter much. Scientists building evolutionary trees largely ignored it.2 Woese’s 16S rRNA tree, for example, showed branches splitting apart, but no genes moving between them. With the advent of whole-genome sequencing in the 1990s and the deluge of genomic data that followed, scientists could no longer dismiss HGT.

A 1998 analysis of the newly sequenced E. coli genome found that 755 genes (18% of its genome) had been horizontally transferred from other organisms. In the years since, scientists have documented DNA exchange between bacteria (and archaea) within the same species, across species within a genus, and between much more distant lineages.

A recent study of over 93,000 microbial genomes found cross-genus gene transfers in roughly one-third of species, and cross-phylum swaps in at least 8% of species. Bifidobacteria in the human gut can exchange DNA segments up to 247,000 nucleotides long, for example, replacing up to 30% of a chromosome in one step. For reference, most genetic engineers insert only about 2,000 to 6,000 nucleotides of DNA in a typical experiment. Further, long segments introduced by HGT have been shown to be more receptive to further HGT with the donor species, effectively making one organism a member of two or more otherwise distinct species.

HGT challenges the orthodox view of the tree of life. Some have argued that HGT, not natural selection, is the dominant mechanism of evolutionary change for bacteria and archaea (which is to say, for the majority of living organisms for most of life’s history). As a result, the criss-crossing branches instigated by HGT may mean the tree of life is better represented by a scaffolding of vertical branches webbed by horizontal connections. Perhaps the most famous illustration of this is a sketch by molecular biologist Ford Doolittle, which first appeared in Science magazine in 1999. Whether life as a whole is best represented as a tangled tree or an interconnected network is still a matter of debate, but the general consensus is that the two representations can coexist.

Bacteria trading genes like playing cards, though, raises an awkward question for regulators: how much can a taxonomic boundary tell us about biological novelty? In the United States, the Environmental Protection Agency (EPA) regulates many industrial uses of engineered microbes under the Toxic Substances Control Act (TSCA), a 1976 law designed to control risks from chemicals. Its 1997 biotechnology rule generally treats microorganisms formed by the deliberate combination of DNA from different genera as “new,” triggering notification and review for commercial uses. The EPA’s rationale was that, as organisms from different genera generally have fewer traits in common than close relatives, deliberately combining their DNA could therefore produce new combinations of traits whose effects on human health and the environment would be harder to predict. The agency set the genus boundary as a threshold for identifying “new” microorganisms requiring review.

But as we have seen, genus boundaries are an imperfect proxy for novelty. Strains within one species can carry distinct accessory genes, while distant species can acquire the same genes through HGT. Taxonomic revisions can also move organisms into different genera without changing their DNA. This does not make every engineered microbe safe, but it does weaken the notion that crossing a named boundary is, by itself, a reliable measure of biological difference.

Engineered microbes are not inherently safe, either, but crossing a genus boundary is a poor test of whether they pose a risk. The EPA already considers what a microbe does, where it will be used, and how it might survive and spread when conducting a review. Those same considerations should determine which microbes ought to require review in the first place.

Blurred lines

Throughout his life, Frank Koontz transcended boundaries. In the span of just 15 years, he went from a grocery store clerk in the slums of Depression-era Baltimore to a postdoctoral fellowship at Oxford working among Nobel Prize winners. He directed PhDs and MDs in a hospital microbiology lab, lectured at high schools, and spoke to civic groups about sexually transmitted infections. He spoke of complex ideas with a brash sense of humor: during one speech, a kid asked him, “Is it right, doctor, that once you get clap, you can never get it again?” to which Koontz replied, even knowing the kid’s mother was in the audience, “No, the only time you can’t get gonorrhea is when you have gonorrhea.”

Genomic sequencing has shown us that bacteria, too, can transcend boundaries. We may seek order and structure in the universe, but Nature isn’t obliged to provide it for us. The boundaries between kinds of things (branches of the evolutionary tree, species, genomes) are often fuzzy, with one kind grading into another. HGT and the mosaicism of bacterial species confuse and frustrate us only so long as we insist that everything fit neatly into the boxes we’ve constructed. Yet for all that, species names are practically useful.

Without common terminology to anchor our conversations, communication about this microbe and not that one would be much harder. Regulatory rules should therefore set clear thresholds based on potential harm, exposure, and uncertainty, while remaining flexible enough to accommodate the biological variation that species names cannot capture.

Kevin Blake is a scientific editor at Washington University in the Division of Laboratory and Genomic Medicine. He writes about microbiology, bioinformatics, and evolution.

Cite: Blake, K. “Species Have Blurry Boundaries.” Asimov Press (2026). DOI: 10.62211/12hq-96nb

Header by Ella Watkins-Dulaney

Nick has engineered these petunias to include an inducible gene. When he lightly touches their buds with a clear molecule called dexamethasone, the molecules seep into cells, switch on the gene, and cause them to make a pigment. The result is a paintable flower, where molecules literally control which colors show up in particular places. Nick’s plan is ultimately to engineer these petunias to express three separate inducible genes, or biosynthetic pathways, for cyan, magenta, and yellow. For a prior project, Nick turned Arabidopsis, a small (and quite ugly) plant that researchers commonly use in research, into little pink “micro-roses.”

We filmed this episode surrounded by flowers. Nick patiently explained to me how each one gets its shape and pigmentation; how he combines breeding and genetic engineering to make new flowers; and how he searches scientific papers for mutations to figure out which plant crosses to make next. His ultimate dream is to build a botanical garden full of weird and wonderful flowers that exist nowhere else.

Watch on YouTube. Listen on Apple Podcasts or Spotify.

Sponsor: Plasmidsaurus offers whole-plasmid, amplicon, genome, and RNA sequencing. Place an order online, drop your sample in a drop box, and receive results with analysis the next morning. They handle the complete sequencing workflow so you can focus on the science. Sign up today at plasmidsaurus.com.

Timestamps

1:28 Paintunia
7:46 Tools for flower design
18:47 Petunia patterns
33:16 Engineering new traits
40:38 Turning Arabidopsis into pink micro-roses
54:26 Getting genes into plants
1:06:02 Mutant botanical gardens
1:14:24 Beauty is sufficient
1:20:30 Flower design studio

Transcript

Desnoyer: I would like to create a botanical garden that you walk into and you don’t just appreciate the flowers, but you actually understand them a bit deeper through a scientific lens. I want these flowers to only exist there. I want them to be weird and wonderful. When you walk through this garden, you get that feeling of magic in that when you walk through the garden, you’ll be forced to take a closer look at what’s going on. That once you leave this garden, you want to take a closer look at nature.

McCarty: I’m extremely excited for today’s interview, which is with Nick Desnoyer. He’s a plant biologist in the United Kingdom; a postdoctoral fellow at the Sainsbury Laboratory in Norwich, England, which is where we’re filming today. But more famously online, he’s known as a flower designer in that he designs and then creates his own types of flowers using breeding, mutations, and genetic engineering. If you can watch this interview on YouTube, that’s probably the best option because we have a lot of props. There’s a lot of flowers that we’re going to talk about. And let’s just get into it. Welcome, Nick.

Desnoyer: Thanks for having me. Very excited to be here.

McCarty: Let’s just start with the announcement. You’re working on a new project. You haven’t revealed it yet. Tell me about that.

Desnoyer: I haven’t actually talked about this project that I’ve been working on for a bit. It’s my next big project after the micro-rose. It’s called the Paintunia. The whole vision of the Paintunia is that it’s a living canvas. In contrast to a normal canvas that you would paint onto that’s made of dead plants, fiber and cotton, this is a living canvas that you actually direct it to paint itself. And I think there’s a lot of magic in that, which we can maybe get into.

McCarty: How does it actually work? It’s an engineered line of petunia, and tell me about it. We have it on camera.

Desnoyer: We have it here. This is the first version. There’s going to be many iterations. The final concept is that you would add three different inducer molecules. These are small molecules that you can treat to the floral bud at different stages and different concentrations. And based on those treatments, you direct transcription factors that lead to the expression of pigment biosynthesis genes that become turned on after about 16 hours. And so in a way, you’re orchestrating the gene expression of the flower and you get some feedback from the flower after you’ve treated it to see how the molecules diffuse through the flower. How is it responding to different pigments being produced in the same place? Things like that.

McCarty: This is one of the flowers here. And so this is not actually a flower that you made, right? It’s from a 20-year-old science paper, research paper.

Desnoyer: Yeah.

McCarty: But they basically made an engineered line of petunia that when you paint it with this small molecule called dexamethasone, it turns those regions pink as the flower opens. And so you get this. This flower has actually been painted.

Desnoyer: We have a few more of them as well. You can see that you can make different patterns based on how you treat it. And so I want this flower to really become a medium for art, where the flower really becomes the canvas. I’m calling it Paintunia because it’s based on Petunia hybrida, which is probably the second most studied flower in the world just after Arabidopsis. But there’s a lot more known about the colors in Petunia versus Arabidopsis. It’s the most studied in that realm. And what’s really nice about it is you see you have these five fused petals that are papery thin. They’re like three cell layers thick. And so it really seems like a perfect canvas.

McCarty: You want to take this line essentially that somebody had already developed and you want to add more colors so that there’s different small molecule inducers that would turn on a cyan, a yellow. How hard do you think that’s actually going to be?

Desnoyer: This might actually be the first time that you have a flower producing all of these different classes of pigments. For example, flowers that make betalains don’t typically make anthocyanins. And so we don’t necessarily know how those blends of colors are going to look.

McCarty: But what color is a betalain? What color is an anthocyanin?

Desnoyer: Right. Because this will be a living canvas for painting, it needs three major colors, cyan, magenta, and yellow. And with this, you can make blends to reach the entire color spectrum. Right now I have magenta, which is anthocyanins, but I’m going to eventually make this to be the cyan, or at least try and get close to cyan. For cyan, I’m going to now cross this to ornamental varieties that are a bit more blue. A mauve. Actually, I think we have one here. Yeah. In this background, this is also anthocyanins, but it’s a bit more blue. Ideally, it would be a perfect cyan. I’m going to try and introgress some of the anthocyanin biosynthesis genes in this background into that flower. When you add the dexamethasone, it turns a bit more blue. And then for the magenta, I’m looking for betacyanins, which are betalains.

And then for the yellow, I’m going for either betaxanthines or carotenoids.

McCarty: Do we have a yellow one?

Desnoyer: Oh yeah, down here. This is a more newer line. I think this one popped up about five years ago. And this one was exciting because they didn’t have a true yellow petunia before this. This one’s called Bees Knees and it’s a completely different class of pigments. It’s called carotenoids. This is probably accumulation of beta carotene within the chromoplasts. But what’s cool is that once you now have this true yellow, you can cross this to all the anthocyanin lines and then see how does this pigment blend with the others.

McCarty: Is that what you’re doing? To get your second color for the pentunia?

Desnoyer: Totally. For example, if you were to cross this yellow with this dark purple line, you get a flower called Black Mamba. And this is... I wouldn’t say true black. It’s almost a philosophical question about what true black is. But because you have all these anthocyanins that are absorbing a certain spectrum of light and the yellow doesn’t overlap with that spectrum, it’s just absorbing a lot of the light. And so you get kind of a black flower. The reason I’m going for CMYK and not RGB is these are two classes of colors. What RGB is, is additive. So it’s like a screen. A screen starts as black and then you add red color or you add blue color and you see how those blend. Whereas in real life when you’re not illuminating something, but it’s rather the absorption of the light, it’s subtractive. And so that’s why printing in physical paper is CMYK.

McCarty: When you actually think about designing a flower, what are the dimensions that you can control? What are the sorts of parameters that you think about? And how much can you actually deviate from natural designs?

Desnoyer: I break it down into two major principles of flower design that helps me organize all the thoughts. And this is a simplification. It’s just based on the visual of the flower. There’s other dimensions like smell and taste, maybe even hearing that we can get into. But this is just based on the visual. I think of sculpting the flower and painting the flower.

In terms of sculpting, you can think about rearranging the architecture of the flower. That’s where and what organs are placed and how many. And then once those organs are placed, how are they shaped? Their size and shape, that’s more organogenesis. And then on the painting side, I think of changing its color or pigmentation and then where those pigments are placed. The patterning with those four major principles, you can think about designing any flower you’d want.

McCarty: For example, this flower here, this is just a wild type flower, right? This is a plant that people bred probably to look like this. People have basically been creating flowers that have certain forms for thousands of years, right?

Desnoyer: Yeah.

McCarty: In terms of construction or the first part of this, the sculpting: is there really any additional benefit you get from genetic engineering? Or is it basically all breeding?

Desnoyer: Right. Just for some context, there’s about 300,000 flowering plant species. And then when you think about all the varieties that we’ve made, it’s tens of millions, even the varieties of the wild type and the ones that we’ve bred. What’s really beautiful about breeding is anyone can do it. It’s extremely accessible. All you need to do is just cross two flowers, collect the seeds and then start selecting the traits that you like. This goes back like 10,000 years. For example, modern corn comes from tiosinte, also a flower. And this has actually been fundamental to the success of humanity. It’s probably the biggest driver of human success is flower breeding. And so there’s breeding for purely aesthetic purposes, but also obviously breeding for food. It’s very powerful. Almost every flower you find in the grocery store has been bred, and you can make so many different forms and funky colors, everything under the sun.

Bioengineering, in contrast, it’s a bit more modern and new, and you have, I would say, less control. I would compare it to kind of like levers that you have that you can introduce to create new traits. And what I’m really interested in is creating traits that you wouldn’t be able to make with breeding. And so with breeding, you’re stuck within the plant family that you’re able to cross these flowers with. And so if you try and cross this rose with another species of rose, maybe they’re incompatible. All you have is the natural genetic variation, which still there’s a lot. Every single one of these petunias comes from an original cross of Petunia axilaris from like 200 years ago. But what really makes me excited is with bioengineering, you break out of that natural variation and that plant species and you can access the entire tree of life.

And so then you can think of all the molecular inventions that evolution has created. And so some of these flowers are taking molecular inventions or genes from bacteria, fungi, humans, you name it.

McCarty: The thing I’ve been most impressed by, hanging out with you the last two weeks, is that you have this somewhat encyclopedic knowledge of papers and mutations and phenotypes. You can recall things such as, “Oh, in 2002, this lab made this cross. They found this mutant and the flower had these traits.”

Desnoyer: Yeah.

McCarty: So basically, a big part of your design process is just knowing which mutations people have already found and what that led to in the flower. And then crossing those in interesting ways. There’s actually not a lot of genetic engineering involved, right? There is in the designs and we’ll talk about other projects later where you’re adding pigments and these things. But the sculpting process seems to be finding the right mutants and then breeding them.

Desnoyer: Right, exactly. It’s not like a typical design process where you would start with a customer and think about what’s the ideal product for them and then go back and this year limited to the code. It’s like, what can the code make? And do we know the code that makes that? And so for example, in Arabidopsis, it’s the most studied flower in the world. It’s got 34,000 genes and we know the function of most of the important ones. And so that is like a whole sandbox for genetic engineering and all the tools that are made for Arabidopsis. That’s probably the most powerful medium for genetically engineering a flower in terms of sculpting.

McCarty: There is a database of Arabidopsis mutations. Basically, there’s so many people who study that. But you started working in Petunia and there are much fewer known mutants, right? Why do you not spend the first part of your project just making mutants? How difficult would it be to just make a ton of Petunia mutants and discover your own phenotypes?

Desnoyer: Well, I would love to do that. But at the moment, it’s just me and a colleague, Juan Pablo. And I still have a postdoc project, actually, that consumes most of my time. But the ideal scenario would be that I have a garden, I have every possible mutant, I can do mutagenesis and pick up ones that aren’t even found yet. And so most of the diversity that I’m actually working with, they’re commercial cultivars or they’re from the Amsterdam Petunia germplasm collection, which is maintained by Francesca Quattrocchio and Ronald Koes, who have totally helped me get into this.

McCarty: If you wanted to make a huge library of mutants, how would you actually do that? How do people mutagenize flowers and then screen for phenotypes? I would imagine it’s just a lot of waiting. Are they using chemicals? Are they using x-rays? Does it matter what you use?

Desnoyer: There’s a lot of ways you can mutate plants. A famous one, especially in Arabidopsis, is called EMS, which is a chemical that induces—causes the polymerase to make more mistakes and you get all of these mutations. You soak the seeds in this chemical mutagen. Another way is fast neutrons. I don’t know too much about this, but I guess you blast it with neutrons. And this is—breaks the DNA and then repairs, possibly. Another way is actually just putting them in full sun. UV radiation induces more mutations. Well, that all falls under breeding. One of the coolest ones is transposon-mediated mutagenesis, and this is a really cool flower here. This is actually why petunia was a major model organism for so long, is because they found this mutant. This flower, it’s now called W138. It originates from a commercial line called Roter Vogel, which means red bird in German.

And some scientists brought this into the lab and they made it inbred or they tried to homozygose it so that all the progeny are more or less the same and you can have reproducible results. But something cool popped up, which is that this transposon had jumped into one of those transcription factors that regulates anthocyanins, AN1. And it popped in just at the border of an intron in this gene. And so the transposon, it’s about 230 base pairs and it’s extremely unstable. But essentially the scientists saw this. They saw all these streaking patterns. They could tell that a transposon is moving. And then they just propagate this plant knowing that this transposon is going to excise and jump into other places of the genome. And so most of the mutants that we have for Petunia derive from this line and you can figure out what the causal mutation is quickly because they know the sequence of the transposon, so you do flanking PCR.

And they maintain all of those mutants actually in Lyon, in France. I’ve heard that they lost actually all of them due to a flood, which is really sad. And so when I was at the Petunia Community Conference, they were talking about how do we reestablish all of these mutations? What I think would be a really beautiful project actually is because these are wild type, but they don’t have any transgene, these you can grow in your garden. And so I would love to have a community breeding project where you give this out to like 10,000 gardeners. They can cross it to their own petunias or just have these grow and document these transposition events and the weird things that pop out. And this could come along with a guidebook of all the mutants that we know in petunia. They could figure out what the causal mutation is.

McCarty: And so these patterns of red, right? These are basically cells where the flower is naturally white. The flower is naturally white, but when you see the red, that means that the transposon has jumped.

Desnoyer: Right. This is in an AN1 mutant background. And so the flower is typically white because it has no anthocyanins. And the mutation is this DTPH1 transposon that sits within the gene. And so it can’t be fully translated because it has an early stop codon, but when it jumps out early during the petal primordium development, then it restores its function and all the function of all its daughter cells. And so you get these sectors where you can essentially trace the lineage of that cell. And depending on when that transposon jumps, you’ll have a different size of the sector. These really large sectors that happened really early. Whereas if it jumped really late in development, you can have it even in a single cell.

McCarty: There’s other petunias here too, right? Let’s talk about more petunias.

Desnoyer: Sure.

McCarty: That’s the transposon petunia. What are these ones?

Desnoyer: Let’s see.

McCarty: This is called Night Sky, right?

Desnoyer: Yeah.

McCarty: What’s going on with Night Sky?

Desnoyer: You can tell this one’s Night Sky because it has all of these white spots, but the spots can actually be much different than this. This is an example of post-transcriptional gene silencing. I think this one’s even cooler. It’s like an inversion.

You almost have this Rorschach looking pattern and you can see it’s quite diffuse. If you see this diffuse white and not a dark purple background and you especially see this gradient of color going white to red, you can be almost certain that that’s post-transcriptional gene silencing. Because these are actually—the silencing is mediated by small RNAs and these diffuse through the plasmodesmata. And so you get these gradients along the border. Whereas the transposon that I showed before, this is a genetic event. And so there’s none of that diffusion. This is actually where post-transcriptional gene silencing was discovered.

McCarty: Tell me that story.

Desnoyer: This comes from another petunia background called V26. It’s a really purple line. It’s this one right here. And I think it was back in the 70s, maybe 80s, some scientists wanted to make this flower even darker purple. And so they overexpressed one of the early genes in anthocyanin biosynthesis called chalconsynthase. And when they transformed this plant and overexpressed it, it actually didn’t get darker purple, but it reverted back to white in many cases. And so I think like 40% of the transformants, it went back to white or had these weird blotchy patterns or circles or even star patterns along the Corolla fusion sites. And there was a WTF moment when that happened. Apparently the first author of this paper was bringing it around showing people, coming up with crazy explanations about how this could be because back then they didn’t know about gene silencing.

McCarty: And there’s some other petunias here too, right? This is the original transposon line?

Desnoyer: Roter Vogel, I believe. This is a nice coral red. This is just another flavor of anthocyanins called peligronin.

McCarty: Why do some anthocyanins look red and some look purple?

Desnoyer: It’s just they have different enzymes that shift the pathway. Anthocyanins are a pretty complicated biosynthesis pathway where the compound is being hydroxylated or glycosylated. And so depending on the presence or absence or expression levels with these different enzymes, you shift it towards different colors. There’s actually an even another layer of the coloration, which has to do with the acidity of the vacuole. These anthocyanins, they move into the vacuole, which is quite acidic environment. And then depending on the pH of that, they can change from blue to pink. Like hydrangeas that you may have in your garden if the soil is really acidic, it’s more blue. If it’s more basic, it’s more pink.

McCarty: And what else? Let’s see. You brought a lot of petunias, right? What’s going on with this petunia?

Desnoyer: This one. You have Queen of Hearts, King of Hearts. This one, we didn’t know actually what it’s called. My colleague called it Silent Prince. As you can imagine, this is also post-transcriptional gene silencing.

McCarty: This looks much more controlled.

Desnoyer: It looks much more controlled, but the breeders still don’t even know why. It’s in this star pattern. And so with gene silencing in petunia, it’s the reason for almost all the patterns. But they don’t know why it arises in certain places. They’re learning more and more about that though.

McCarty: Just to be clear, this flower would normally be purple. It’s just the gene encoding the purple gets silenced at a certain point in development. And so that the cells that become these white part of the petals look white. And you were telling me too, while I’ve been here, that most flowers are just naturally white, right? White is the default state of flowers.

Desnoyer: I would say white is, you could say, the default state of all the organs and that you just have pigments that are expressed over it. And so sometimes you get an effect called virus-induced gene silencing in leaves. And scientists often target the gene called PDS to know if the virus-induced gene silencing is working. And in that case, you get all these white sectors in the leaf and that’s because the chloroplasts are being degraded where the chlorophyll isn’t accumulating and so it goes back to white. You can get a fully white leaf.

McCarty: And so last petunia here.

Desnoyer: Oh yeah.

McCarty: I’m interested in this one because it has this really beautiful—the color flows through the veins. It looks like veins. What’s going on here?

Desnoyer: This one, I don’t actually know. But I would assume that you have more anthocyanins accumulating in the vasculature. Probably the phloem companion cells. There is a promoter called SUC2 that’s a really famous promoter for phloem companion cells that you can drive the expression of different genes. And so perhaps it’s more stressed within the vasculature and so you have more anthocyanins. There’s actually a gene expressed there. But it’s an interesting exercise to think, if I want to now bioengineer this flower, how do I access the different domains of the petal? How can I just get it along the margins? How can I get it just within the veins or the center or these spots?

McCarty: It’s basically when you think about your design, you’re looking to nature. You see things that pop up naturally. And then in your head, you’re like, “Oh, I bet I could figure out a way to use that!” Like getting color in the vasculature or adding new layers of petals like this, right? Double stacking petals.

Desnoyer: Right.

McCarty: Or using gene silencing to get these really beautiful patterns, which is just amazing, right? It’s like you look to nature, you see the natural variation. Then you’re like, “Oh, I can cross these things to get the features I want.”

Desnoyer: Or you think, how does the plant make this? How can I reverse engineer that with bioengineering?

McCarty: Which also makes you look more closely, probably.

Desnoyer: Totally. It completely actually changes how you view nature. And I think that’s the most beautiful part of the project. It’s like how your job influences your perspective on life. If you’re a carpenter, maybe you’re just always looking around at the wood and taking a closer look at that. I probably didn’t even think about what type of wood this table is made of. But it influences how you see the world. And so as a flower designer, you really start to look at all plants as amazing inventions of evolution. And it really sparks your imagination as to how they make these things.

McCarty: This is technically a petunia, even though it doesn’t look like it, where the leaves are forming...

Desnoyer: There’s basically no petals.

McCarty: So it’s leaves. One thing I’ve noticed is why don’t we see flowers that have green? Surely there are green pigments or there’s ways to put chlorophyll back into the petals. Why do we not see green flowers?

Desnoyer: If you don’t find something in nature that means either it hasn’t evolved because it’s not fit or because it’s something really difficult to evolve. This is the case of the former. It’s just not very fit. As I mentioned before, petals have the sole purpose of attracting insects and having them pollinate them. And so a green petal may not stand out to a pollinator. In fact, it may attract more herbivores that want to eat that petal. Obviously there’s many ways you can make a green petal. All you need to do is have it revert back to sepal identity or not lose its chloroplast. There’s tons of mutations that you can do to make that. But that doesn’t typically evolve.

McCarty: But if you wanted to, could you breed this in the lab to enlarge these so that it looks like a flower that is green?

Desnoyer: Yeah. There’s a flower that we don’t have here, but it’s called the Picasso petunia. It’s actually my favorite and it’s like a lime green, with pink pigmentation in the center. It’s less dramatic than this, but I would imagine that those petals have a bit more leaf identity. There’s also quite a few pathogens that actually infect plants in order to revert the flower back to a vegetative state. And these bacteria are usually transmitted by insects. And so it’s kind of co-evolution of the flower, the insect and the bacteria where the bacteria is trying to revert the flower back into food so that it recruits the insect. And then when the insect eats the flower, it obtains the bacteria and then moves to another flower. There’s also some ecology in flower design.

McCarty: The other color that I basically never see is true blues. There’s a lot of purples. We see blues in animals, right? There’s blue lobsters. Why do we not see blue flowers?

Desnoyer: I was looking for the most blue one we have here, but it’s a bit more mauve. And actually when you call something blue, you have to be very careful because it can be a bit scandalous. There was the first blue rose that came out. It’s called Applause. It came from a company, Floragine. It took 10 years to make. And when they produced it, they said, this is the first blue rose. But anyone with eyes can see it’s purple. And so the blue, it’s like, as you said, it’s really rare to have true blue pigmentation. You’ll find a lot of blue in nature, but a lot of it is structural color. It’s some kind of co-evolution between underlying pigment and then bending that light using structural variation on the cuticle or the waxy surface of the petal. And so that tells you something: that tells you that blue actually is fit, right, because it evolves.

And so it’s really that second category of it’s just difficult to evolve. And that’s probably because there’s something about blue compounds that are hard to biosynthesize or we don’t have the code for or the chemical is unstable. The most classic true blue in flowers is a flower called, I think it’s butterfly pea. It’s known as Clitoria, so you can imagine what it actually looks like. And there’s a guy here at the John Innes Center that actually did his whole PhD on Clitoria and defining what are the anthocyanins that lead to blue in this flower. And so he actually shared his PhD with me. And it’s amazing because he has the whole lore of blue pigments in his introduction. It’s also a beautiful thing that you can do storytelling about all the different colors. We could sit here for the next two hours and talk about every single color and all the organisms that have evolved different ways of making those colors and where they’re from.

And blue is an interesting one.

McCarty: And there’s also this flower that you were telling me about that turns translucent when it rains and it’s native to Japan. What happens with that flower? Why does it turn transparent?

Desnoyer: Yeah. That flower, it’s called the skeleton flower. And actually most petals, if you soak them in water or infiltrate them so that they become completely full of water, they will become transparent. If I take a white petunia and inject in water into the air spaces or the mesophyll of the petal, you’ll be able to see through it. That’s because it’s not an actual white pigment that the flower has. It’s more that it’s just diffuse light. And so that’s caused by the difference in the refractive index of the cell wall of the cell in that air space. And so the air’s refractive index is somewhere around one, and the plant cell wall is like 1.5. And so when it goes through the plant cell wall, it bends the light and then this hits another plant cell wall, bends. And so you get light scattering everywhere.

Of course, this looks white. But when you fill it with water, the refractive index of water is one. It’s very close to air. And so it just goes right through and is transparent. And so I imagine with the skeleton flower, it has large stomata or has some way of becoming soaked really easily.

McCarty: Have you thought about designing flowers that turn transparent?

Desnoyer: I haven’t actually, but that would be a cool one. You would need to probably get rid of the waxy cuticle on top of the flower because flowers don’t want to become soaked. When you get soaked, it’s really easy to get a pathogen infection because it’s full of bacteria. Most plant surfaces have a waxy coating to prevent that.

McCarty: Do you feel like we’ve mapped all the levers? So you also have this thing with smell, right? You have an ongoing project. But smell—are there, beyond smell, are there other levers as well? How big can we really go with flower design?

Desnoyer: I think you’re kind of limited in what code we have and what code has been discovered. And with bioengineering, I see it as complementary to breeding; you can create traits that you wouldn’t normally get within that plant family. And then once you create that, you can start to introgress it into different lines with breeding. Flower design encompasses both breeding and bioengineering. The exciting part of bioengineering is you can create traits that would have never evolved within that plant family. For example, recently people made a glow-in-the-dark petunia; the firefly petunia from Light Bio that’s being sold in the US. It’s a bit dim, but still it glows in the dark and it’s literally emitting photons and recycling the substrate for the luciferase, which is quite incredible. And so things like that get me really excited about new to nature traits and also this adventure aspect of biology in which you look out over the entire tree of life, at what are all the molecular inventions we have, and then thinking creatively about how can we implement those to create new patterns, new colors, and new structures.

McCarty: And what is the smell project that you’re doing?

Desnoyer: The smell project, we first wanted to go for something that we are very sensitive to, which is the smell of rotting flesh. And so there’s a few carnivorous plants. They’re called malodorous plants because actually they want to attract insects and want to mimic rotting flesh. Like Rafflesia, which is the biggest flower in the world. It’s in Malaysia, this really large red flower. But it produces a chemical, DMDS, through a two-step pathway that was recently discovered. And so actually it was my professor now, Sophien [Kamoun], who knew the person that discovered these enzymes and was like, “You should make smelly petunia.” But this is also potentially cool because maybe you could attract new types of pollinators like flies. So we synthesize the genes and we always test them first in bacteria. We always use microbiology to do quick tests, and so we boosted the expression of these two genes in bacteria and the first thing I did was make a liter of this bacteria and I let it grow overnight.

It’s a bit of a funny story because the next day I came in so excited to see if it really smells horrible and I was actually a bit worried because when I walked into the building, the whole building smelled like sewage and people were even talking me about how they couldn’t work there. And I immediately went white in the face. I was like, oh no, this is all because of me. But there was also a part of me that was like, by God, we’ve done it. And I’m thinking that this is going to be the most legendary story of the Sainsbury lab. And so I get closer, I’m telling people, I’m like, I think it’s me. And I get closer to the shaker where we’re growing the bacteria and I open it up and it smells just like E. coli.

And so it just so happened that the same day that I grew this massive flask culture of this corpse flower gene bacteria one of the sewage pipes had broken. With that said, when I smelled the bacteria it still smelled a bit like sewage, and we did a blind smell test throughout [The Sainsbury Laboratory] where I brought people controls that don’t have the gene. And they did double blind smell tests, and seven out of seven people could distinguish them. Now we’re going to see if putting these genes in petunia, if it triggers that smell. You’ll need the precursor of the metabolite and the gene product will have to be in the right place of the cell. Those are always the two considerations. And if those work, then we hope to actually test it with different pollinators, seeing if we can do pollinator switch.

I would love to have petunia, which is normally pollinated by hawk moths, hummingbirds and bees to now be pollinated by flies.

McCarty: Why did you choose a bad smell instead of a good smell?

Desnoyer: It’s just the first thing actually that came up. Sophien just happened to know a person that was working on it. But later, after telling this story, I realized it’s actually probably the best thing to go for first because it elicits a really strong reaction. And some people really don’t like it and some people love it. And probably the best art is art that divides the audience or produces a really strong emotional response. And what produces a more emotional response than the smell of rotting flesh in a flower? But it’s also a beautiful art project because it questions what is beauty. To us, we’re like, oh, this is a disgusting flower and maybe also to the hummingbird. But now that it’s attractive to the fly—well, beauty is subjective. The fly may find that flower more beautiful. And so it brings up this interesting question.

McCarty: And also, of course, flowers did not evolve for our aesthetic sense. They evolved to look the way they look to attract pollinators, right? That’s the whole function of the flower.

Desnoyer: Almost every flower that you can find in nature which is insect pollinated is a physical manifestation of what that insect finds beautiful. And so over generations and generations, they keep selecting what they find the most beautiful. And it’s a physical manifestation of that insect’s aesthetic sense.

McCarty: Let’s talk about Arabidopsis. Your first flower design project was micro-roses, which were basically Arabidopsis that you bred and then you added a transgene to get color. And so let’s just start with the wild type. This is normal Arabidopsis, right? It looks so unremarkable. And that’s because you were saying plant breeders have just inbred it for generations. Why were they inbreeding it in the first place?

Desnoyer: Scientists continued to self it so that it became an almost completely homozygous line. That means all the seeds from that plant will have more or less the same traits and that makes experiments more reproducible.

McCarty: I see.

Desnoyer: Oftentimes if a scientist wants to adopt a new model organism for plants, the first step is to make it either double haploid where it’s completely homozygous or you just keep selfing it for like 10 generations.

McCarty: Basically they just breed it with itself until all the progeny are the same.

Desnoyer: Right. But still they are slightly different. If you were to find wild type in different labs around the world and sequence them, I’m sure that they would each be slightly different. And this actually may lead to some discrepancies in results. There’s also different wild type backgrounds. There’s Landsberg, there’s Columbia, there’s Erecta, and then there’s tons of other ones. But the most common one which you see here is Columbia, like Columbia, Missouri.

McCarty: If you were to walk around outside though, you would see Arabidopsis all over the place. But it wouldn’t really look like this, right?

Desnoyer: No, it would look very, very similar. I believe even in the wild, Arabidopsis thaliana is still mostly selfing. Which is a sad story because once a plant evolves a selfing strategy, it’s very efficient at the beginning because you can produce a lot of seeds. But it’s more or less an evolutionary dead end because you can’t create any new variation.

McCarty: And so when you started to think about designing Arabidopsis, what was the first thing that you did?

Desnoyer: Actually the first time I designed Arabidopsis was with fluorescent proteins. My PhD was all on plant reproduction and I was doing live imaging of double fertilization within the flower. But through this, I developed all these different fluorescent lines of the flower. And so I started to cross all of these and actually think of it as genetic painting with fluorescent proteins, trying to highlight different parts of the plant. And then a new tool came online that actually a postdoc brought into the lab. I still remember his journal club on this paper. And he brought in a tool called RUBY And RUBY is three genes from beetroot—CYP76AD1, DODA, and a glucosyltransferase gene—separated by P2A linkers that are self-cleavable so that you get one mRNA that can produce three proteins. It’s elegant. And they use this as a selectable marker or reporter for gene expression.

And so this postdoc, he wanted to put it under a pollen promoter because we all studied the pollen. He wanted to use it as a marker for tracking the pollen tube growth. But he noticed on the single cell level, the pollen tubes are still translucent, so it didn’t really work. But there’s a lot of serendipity that occurs when you transform a plant with a construct like RUBY. It can have multiple copies or it can go into different parts of the genome. And so he picked up all these transforming lines that had pink pigmentation in the full flower or just the petals or just the pollen. But for his experiment, it didn’t work.

McCarty: Right.

Desnoyer: He was ready to throw them away. But me, having played with genetic painting of fluorescent proteins for so long, I was like, wow, this is actually visible. At the same time, right next to these flowers, I had these agamous mutants. And in short, instead of producing petals, sepals, stamens and carpels, they produce sepals, petals, petals, sepals repeating. It becomes an indeterminate flower. And I was teaching a course. That’s why I was growing these flowers, and the students always love this one the most because it’s actually the genetic underpinning of a lot of commercial flowers that you see in the grocery store that have this double flower phenotype. They’re like little roses. I crossed a carrier of agamous—these don’t produce any reproductive organs—with this RUBY line. This is actually a funny story because it was a carrier for agamous, but this is a really old EMS allele.

And in fact, we have no selectable marker for this mutation.

McCarty: What does that mean, EMS?

Desnoyer: This is a chemical mutagen that can induce mutations. Back in the early days, people were mutating Arabidopsis to do forward genetic screens, where they would treat them with EMS. They put the seeds in the chemical and then look for interesting mutants that pop up. And you know if your EMS worked, if some of the leaves have white sectors, but also about one in 10,000 plants will yield an agamous mutant. They use this as a proxy to see if the mutagenesis worked. But to propagate these is quite hard because these ones don’t make seed. The only way to propagate these would be to vegetatively cut them, which you can’t do with Arabidopsis, or propagate them through carriers of the mutation. So you find wild type looking plants that would be heterozygous for the agamous allele. But because this is an EMS allele and we didn’t have any selectable marker for it and we couldn’t even genotype it, I do it through classical genetics where I plate all the seeds of a wild type looking plant and see: does the mutation pop up in the next generation?

And so all I did was cross a plant which I thought was a carrier for agamous. I had only two thirds chance that this was a carrier and crossed these two plants. And so in an alternate universe, that plant was the one third that was wild type and didn’t get anything. And I would have just trashed them. That’s what I mean by there’s a bit of serendipity in it. And so I crossed these two flowers. I grow the first generation, which is called the F1, filial one. And again, I’m not genotyping them. Also I don’t know if the progeny carry the mutation and they’re just a little less pink than this because now they’re heterozygous for that. I selfed that plant and then in the F2 hoped to find some pink little micro-roses popping up. In the F2 population, then deriving from this cross, I found a few plants that were homozygous for agamous, as in they had this double flower phenotype, and homozygous for the RUBY, which looked a bit like this.

I thought this could get a lot more interest within the course and that this actually might be a better way to reach the public than genetic painting with fluorescent proteins. And so I brought this through the entire institute showing everyone I could find and was met with mixed reactions. So the younger students, like the undergrads and the masters, they were like, “Whoa, that’s so cool. You transform this into this and now it’s big and pink.” But when I brought it to the postdocs and the professors, they were like, “Yeah, you made a cross. Now get back to work. What are you doing?” But I was quite convinced that people will find this cool especially when you compare the wild type and the new version to show people what you can do with biotech.

I had already been making scientific art that I put online for about four years up until that point. And I posted this post and it was the first time that people actually reached out to me offering money. And I had always hoped that this would happen one day. And so I was really excited after I made this micro-rose, I showed it to my professor and he was like, oh, you could introduce this mutation and this mutation to make it larger or to have more petals. And so I took that as inspiration to upgrade these roses over the next year. And so rather than looking at it as a potential startup, I was exploring it more from an artistic perspective of: is flower design a real thing that we can do? Or is it just a simple cross like the postdocs were saying? And this was one of the most exciting times in science because it really for me, because it started to change how I saw the science.

I was writing my PhD, but also exploring all of the literature of Arabidopsis research, which has over 10,000 papers on it. But now I was starting to see the utility of this research actually. When you try and make something out of a flower, you start to see what are the actual impactful papers that we can start to create stuff with. I started skimming these papers and looking at them more as catalogs of mutations: what could make these flowers a little more aesthetic? And there’s very few genes that actually make the flower a little more aesthetic. I went for about five and I just ordered some primers for CRISPR and started to transform these pink roses with the CRISPR. And I also went for a few low hanging fruits of what I thought could be achieved with bioengineering, but not breeding, which is making the flowers interactive; such as using inducible gene cassettes that you can add a small molecule and you can direct the production of pigmentation.

Over the course of the year, I explored many different combinations of mutations and many different transgenes to control the color. And all the while I had this vision in my head of a flower that’s—instead of being sepals, petals, petals, sepals, which makes it a little bit mangly—it’s much larger. It has sepals and just petals, petals, petals all the way down, is a bit more symmetrical and large, and has a bleeding red center, which I thought I could achieve with the inducible system. And so I remember clearly going to bed and having this burning vision of: I want to create this. And over the course of the year I explored a lot of funky things, realized what works, what doesn’t. And around December 2024, one line popped up that was homozygous for erecta, agamous, and superman and heterozygous for clavata3.

And it had the RUBY transgene as well in the background. It had about five different edits you could say. And this—I was extremely excited because it was much more beautiful than, in my eyes, than the original. And this time—and when I brought it through the Institute, I was met with much different reactions. In fact, I brought it to some professors that had worked on Arabidopsis their entire career. Like 40 years. And they were asking me what species it was. And so I took that as affirmation that I’m onto something.

McCarty: What were some of the interesting micro-roses that you got? And if you were still working on that, are there other colors or what else were you thinking in terms of expanding that?

Desnoyer: Right. Some of the other genes that I explored were more to do with organogenesis. There are two genes, DA1 and EOD1, that regulate the size of the petal. As I mentioned before, petals are dispensable and their shape and size don’t really affect their physiological function. And so you can introduce mutations like frl1 that make frilled edges or da1 and eod1 that make them larger. You can introduce mutations like clavata3 that actually affect the floral meristem size and balance. That’s where you also get some kind of freaky things that look like fasciated flowers or even tumors.

McCarty: We have a fasciated flower, right?

Desnoyer: We do, yeah.

McCarty: Show that.

Desnoyer: This is almost certainly a fasciated stem. And so the vegetative meristem likely became too large at some point in this mutant background and all the stems started to fuse together.

McCarty: But yeah, it’s like two stems. It’s like a very thick, flat stem. I’ve never seen a plant like this.

Desnoyer: But now that you’ve seen this, you’re going to start finding it in nature, especially if you look at dandelion. Dandelion clonally reproduce and so they don’t have much genetic variation. And so they’re very prone to making mutations. And sometimes they get mutants in clavata3 and start to fasciate like that.

McCarty: And that also happens in the meristem, right? Instead of one center, it has a split center.

Desnoyer: That’s right.

McCarty: Basically to sculpt the flower initially, you were doing crosses with mutants and your professor suggested some other things you could cross in. But then you started to do CRISPR. And I’m curious about that because before I came here to visit you, I had no understanding of the plant world. It was almost like a mystical thing when people would talk about tissue culture or whatever, or agrobacterium. And so maybe you could walk me through that. When you—we have some plates here, and could you explain what some of these are and how you actually get genes into the Arabidopsis?

Desnoyer: There are many ways you can get a gene into a plant. It depends on which species you’re working with. Arabidopsis is such a famous model because you can really easily transform it. And so you put your genes into a bacterium called Agrobacterium that occurs in the wild. This naturally infects plants and produces what’s called crown gall disease. Chemical signals from the wounded plant, such as acetosyringone, activate the Agrobacterium. The bacteria produce this secretion system, which is like little needles that poke into the plant cell, and they’re able to insert what’s called T-DNA. In the wild, this T-DNA carries tumor-inducing genes. And so the plant essentially becomes a home for the bacteria and starts to feed them metabolites. Some really clever scientists thought, “Oh, we should just take away those tumor-inducing genes and use this beautiful invention for inserting genes into plants.”

And then there was a guy who took that Agrobacterium and just dipped the flowers of Arabidopsis in the bacterial solution, collected the seeds, and found that, wow, this is enough to get that gene into the egg cell of the flower. The next generation then carries the stably integrated gene.

McCarty: You can just dip them into bacteria?

Desnoyer: Exactly. And that only works so well in Arabidopsis because of the development of the flower. There’s a stage very early on when the carpels, which eventually bear the seeds, are fusing and make a furrow. During that stage, the Agrobacterium can get in, and later the carpels enclose it. And that’s what infects the cells that become the egg cell. So it’s really based on the anatomy of Arabidopsis. If you dipped this petunia in Agrobacterium, it wouldn’t work.

McCarty: And so, speaking of petunia—to transform petunia, which I know we’re jumping back, you have to use tissue culture, right? How does that work?

Desnoyer: Essentially, for petunia, you take a cutting of one of its leaves and put it onto a medium with different concentrations of phytohormones, namely auxin and cytokinin. With the right balance of these phytohormones, the cells go from a differentiated state back to a dedifferentiated state, like stem cells. And they start to produce what’s called callus. Callus is totipotent. It’s undifferentiated cells that can produce an entirely new plant.

So what you do is, early on, when you cut the leaves, you co-culture them with the Agrobacterium. The Agrobacteriumstabs into the leaf cells and inserts its T-DNA. Along with that T-DNA, you add a selectable marker. This could be something like the RUBY transgene that colors it or an antibiotic-resistance gene. After the co-cultivation, you kill all the Agrobacterium with antibiotics and put the leaf tissue on a selection medium that again induces the callus. This time, only the callus that carries the transgene is going to survive and produce a whole new plant.

And so then shoots start to grow from this leaf and produce the next plant. Eventually, you move it from a shooting medium into a rooting medium, which has little to none of the phytohormones, so that the shoot can start to produce roots. The auxin and cytokinins actually have different modes of regulating genes in the shoot and the root.

McCarty: And so the process—if you have to do tissue culture, like for petunia—takes much longer, right?

Desnoyer: This one might even have a date on it. Actually, this is when we transferred it. But I can tell you this one is probably two to three months after cutting the plants. It’s quite remarkable. This whole plant came from a single cell from a leaf.

McCarty: There’s a physical limitation in how long it takes for a plant to grow. But when you dip Arabidopsis, you can immediately take it and regrow a plant, right? How much faster is it to transform and grow the next generation of Arabidopsis versus petunia?

Desnoyer: Right. Arabidopsis takes about a month and a half to grow. You can get flowers within four weeks. You dip those, wait about two to three weeks for it to produce the seeds, get those, and you have your transformed plant. You can go generation to generation in about a month and a half. And that’s how, within a year, I could squeeze in about five generations. You can never do that with petunia, but petunia is also a much larger, hardier plant. So it just takes longer to grow.

McCarty: And so with the micro-rose project, right? You went from this to petunia, presumably because you have a bigger flower to start playing with, right? But it also takes much longer, and researchers have mapped fewer mutations compared to Arabidopsis, right? What are the trade-offs when you choose a flower? What are you thinking about?

Desnoyer: Actually, when I was first making these micro-roses, I was kind of convinced that I was going to work on this for the rest of my life. I was calling it the Thousand Flower Collection. I was like, “I’m going to make a thousand different versions of this flower.” But I realized that even with Arabidopsis, there’s still a disconnect with the public. Most people don’t know this flower, and it’s just a bit too small for people to really get interested in. Mostly, they’re just not familiar with it.

This first project was also mostly impactful because of the sculpting of the flower. And that was possible because of all those mutations that we knew in Arabidopsis. So for my next project, I wanted to move from predominantly genetic sculpting to genetic painting, which is when I had this idea of the Paintunia.

And petunia is probably the second most studied flower, just after Arabidopsis. You know a lot more about the color. There’s a lot more standing variation that you can use. And it’s a beautiful canvas. So lately, I’ve completely switched to petunia. But after I make the petunia, I would like to switch to another system as well.

That’s what’s beautiful about flower design: it’s really endless. Once you get bored of one thing, you can move to a different medium. With each flower, you start to learn its growth behavior, how it reacts to certain treatments—even tissue culture you have to master. I like to think of mastering this organism. You have that in your tool belt, and through doing that, you develop a kind of deep connection with the plant. And you keep moving on and mastering different plants.

It’s like if you moved from watercolor to oil painting to clay.

McCarty: Also, at one point while I’ve been here, you were talking about how there are also viruses that you can infect plants with, right? There are so many dimensions to engineering plants. If you think about it, it’s breeding, CRISPR, transgenes, pathogens, and all these things—you’re kind of tuning the development, I guess, of the plant. What were some of the viruses that you were thinking about? Although I know you’re not doing it anymore because people don’t like to hear about engineering viruses.

Desnoyer: Exactly. Actually, if you want to look at engineering a plant, one of the best places to go looking is at plant pathogens because they naturally hijack the plant machinery. Phytoplasmas, for example, revert the flower back into a vegetative meristem. They have all these genes that can regulate the plant at the genetic level or at the protein level, and pathogens make all sorts of patterns.

In fact, if you go out for a walk, just take a look at a leaf, and almost certainly you’ll find one that’s infected. One of the most famous infected flowers is ‘Semper Augustus,’ a tulip that had this fiery, flame-like pattern with red streaks in it. At the time, people had never seen this type of tulip. And so they called it tulip mania because this tulip became absurdly expensive.

However, it was a bit of a scandal because later they found out that a virus was causing this effect. And so it’s not heritable, and a lot of the flowers start to wilt over time. It was a big waste of your money to get this flower. Now, with breeding, you can get tulips that look like ‘Semper Augustus,’ but it’s post-transcriptional gene silencing now.

McCarty: It’s not stable?

Desnoyer: It is. It is stable. You can get stably unstable gene silencing in tulips. I used to think that viruses would be a really cool way of making all these different types of patterns. But as I started to learn about them, I realized that because they’re not alive and because they’re just particles, they don’t really have a way of actively moving through the plant. So they just move through the plant based on bulk flow or diffusion. They get into the vasculature, spread out through the veins, and then later through the plasmodesmata. A lot of the patterns that you see from viruses are actually more or less the same. It’s just diffusing through the vasculature. With that said, because they’re introducing new genes, you have what’s called virus-induced gene silencing, where you can get some interesting patterns.

And you can imagine how, in theory, you could engineer a virus to deliver a pigment as it travels through the plant to create cool patterns. I had thought about that because we actually have licenses to work with plant viruses. But that was an important consideration because whenever you’re doing biological design, you have to think not just, “Can we do this?” but, “Should we do this?” And I don’t think that is worth it, and it doesn’t send the right message either.

McCarty: We’ve been talking while I’ve been here, and you said that one of your dreams would be to build a greenhouse full of mutant flowers and maybe even mutant insects that pollinate different types of flowers than they naturally would. I guess my question is: Why is that your dream? Why build something like that? Isn’t the diversity and beauty of nature sufficient for our appreciation? Why do we need a mutant greenhouse?

Desnoyer: First off, there are already thousands of botanical gardens around the world where you can showcase the diversity and natural beauty of the living world. Those are my favorite things to go see. Anytime I go to a new city, the only thing I care about is going to their botanical gardens. It’s like a museum—a natural history museum of all the beautiful things evolution has created and what we’ve made with breeding. And there’s something really kind of magical…

But with this, it’s just admiration. You look at the flower and appreciate it for what it is. But I think there’s an added layer of beauty when you start to interact with the flower or even interface with its genome. You start to understand it at a deeper level that almost feels a bit magical. In fact, it’s the closest thing to fantasy in real life that I can find.

And so that’s the real motivation: to make the Studio Ghibli of biology. Something that instills that feeling of magic and fantasy when people see it. I think we can achieve that by looking at nature more through a scientific lens. There’s a school of thought that by learning more about nature, you actually diminish its beauty. But I think it’s the exact opposite.

There’s a famous monologue by this guy Richard Feynman. I’m sure you’ve heard of this guy—a very famous physicist. He has a whole monologue about flowers. It’s called “Ode to a Flower.”

Feynman [quoted passage]: I have a friend who’s an artist, and he’s sometimes taken a view which I don’t agree with very well. He’ll hold up a flower and say, “Look how beautiful it is.” And I’ll agree. And he says, “You see, I, as an artist, can see how beautiful this is. But you, as a scientist, take this all apart, and it becomes a dull thing.” And I think that he’s kind of nutty.

First of all, the beauty that he sees is available to other people and to me too. I believe, although I may not be quite as refined aesthetically as he is, that I can appreciate the beauty of a flower. At the same time, I see much more about the flower than he sees. I could imagine the cells in there, the complicated actions inside, which also have a beauty. I mean, it’s not just beauty at this dimension of one centimeter; there’s also beauty at smaller dimensions, the inner structure, also the processes. The fact that the colors in the flower evolved in order to attract insects to pollinate it is interesting.

It means that insects can see the color. It adds a question: Does this aesthetic sense also exist in the lower forms? Why is it aesthetic? All kinds of interesting questions, which the science knowledge only adds to the excitement, the mystery, and the awe of a flower. It only adds. I don’t understand how it subtracts.

Desnoyer: And so I would like to create a botanical garden that you walk into, and you don’t just appreciate the flowers, but you actually understand them a bit deeper through a scientific lens. And importantly, I want these flowers to only exist there. It’s like a magical moment where you realize you’ll only find this flower here.

I don’t want to use a typical breeder’s rubric of, “This one has really nice vase life,” or, “This one’s really full and big and colorful.” I want them to be weird and wonderful. When you walk through this garden, you get that feeling of magic, and with each flower, there’s a little description of what the mutant is. Almost certainly, each time you walk through the garden, you’ll feel compelled to take a closer look at what’s going on.

And that’s really the takeaway: once you leave this garden, you want to take a closer look at nature.

McCarty: It’s like when we were looking at the plants right before—the petunias—and every petunia has a different thing, has a different story. Some are gene silencing, and some are transposons.

Desnoyer: Do you watch Studio Ghibli, though?

McCarty: I love those films.

Desnoyer: And there’s a certain feeling you get, right, when you watch those films?

McCarty: You feel the humanity.

Desnoyer: You feel Hayao Miyazaki’s vision. If you learn a little bit about him too, you see that bleeding through into his story. And there’s that feeling that you get after watching, like, Howl’s Moving Castle or Princess Mononoke—I want to create that feeling in biology. How is that possible? And I think that is possible.

I think I told you one of my favorite podcasts is Lex Fridman with Neri Oxman. She said there are realists who are addicted to dreaming, and there are dreamers who are addicted to realism. And the people you want to hire are dreamers who are addicted to realism. They live for the dreams, they live for the fantasy, but they’re super grounded in reality. They just want to make that real life.

As a kid, I was a dreamer addicted to dreaming. I was having all these fantasies, and I had a moment, like when you lose your innocence, where I was like, “Damn, the world is a really boring place.”

As you grow older, you just get into all this monotony, and there is no Final Fantasy. There is no Kingdom Hearts. There are no grand adventures that you’re going to go on. And actually, it’s kind of stupid. But as a kid, because I had all those feelings and I could see the adults who clearly didn’t, I literally made a promise to myself: “This feeling is going to go away, so I need to remember it, hold on to that.”

I made a promise to myself when I was like seven years old. I was fighting with swords under a willow tree, and it was raining outside. I still remember that. And that came back to me once I started the flower design project. Now I’m like, “Oh, you can make a fantastical world that’s grounded in reality.” And that is biology.

That is science. Biology can be endless adventures. That is going out into nature, finding that thing, and bringing it back into the lab. The problem right now is there’s friction between the lab and nature. Nature drives you into the lab as you grow up loving all the biodiversity, and they all have their own special powers and stuff. Then you come into the lab and you just forget that because people don’t do that. You work on this single protein, but can you connect it back to what originally got you into it?

I feel like we’re kind of losing sight of that, and there’s a whole area of biology that’s completely missing at the moment, where the whole purpose is to just bring you closer to nature. Most of the tech we’re creating is driving us away from nature. But can I make tech that actually makes me feel like I understand nature better and feel closer to it?

That’s the future I hope we have as humanity: that we don’t have this cyberpunk future that’s just all these skyscrapers and robots, but rather a really high-tech future where we’re more grounded in nature. And the world looks a lot more like a garden with all these unique kinds of spaces.

McCarty: Some academics have kind of pooh-poohed what you’re doing, I would say. They’ve said, “What’s the point of this?” Is the argument essentially that by designing flowers in this way, you’re deepening our scientific understanding, or are you just deepening people’s curiosity about what’s going on? What is the pitch you make to people who maybe attack this or don’t understand it?

Desnoyer: I think there are many reasons, and a lot of their reasoning is justified. I don’t think everyone should switch to flower design. The world would fall apart, and we wouldn’t have as much food. But I do think there’s an interesting aspect going on, particularly in funding schemes: they’re really targeted at deliverables that are closer to things we just need and that are essential for our survival, like food and medicine.

But it’s actually the things that are least essential for our survival that make us human, like singing or dancing or laughing or gardening. Every summer, people go out and plant flowers in their garden, and some of the bulbs only pop up for like a week. You could ask, “What’s the point of that? Shouldn’t you be growing food?” But that’s not what makes us human. So I think beauty for itself is a justifiable purpose.

It enriches our lives. It adds to the quality of our lives. And so I feel there could be a bit more emphasis on that. I do see kind of a cultural shift, especially when you look at science a few hundred years ago, or even architecture a few hundred years ago: they put more emphasis on ornaments and creating beauty. I would like to see that a bit more in science.

But there are other angles that you can look at it from. Flowers are extremely familiar. So if you want to educate someone about biotechnology, there’s no better approach than a flower. They already know them. They already love them. They don’t have to eat them. These flowers aren’t going into the environment, and I’m showing exactly how I make them. It’s a very neutral approach.

You could also look at it through the artistic lens, which I think is justifiable in its own right: this is a new medium of art where the canvas is living cells.

That’s intriguing. And then, from a scientific lens, how much can we learn from designing and playing around with flower design? If we’re only focused on crop yield, most of the mutations and tools we’re going to use aren’t going to radically change that. Or most of them are going to be deleterious. Whereas flower design is like a creative sandbox. You’re implementing tools in new, creative ways that you wouldn’t normally go straight for if you were trying to increase the yield of a crop.

McCarty: Do you feel like you’ve made any discoveries from designing flowers?

Desnoyer: I wouldn’t say any groundbreaking ones, but I would say I’ve made a lot of observations that I didn’t know of before. I then go look them up and find that someone discovered this in a similar way. That happens all the time when you’re working with things that are super visible.

If you’re just working with fluorescent proteins, you’re only going to find a weird phenomenon if you dissect out that part of the plant, put it under the microscope, and look through a ton of samples. But if you’re working with something visible that’s growing in the glasshouse, it’s going to catch your eye, and you’re going to ask what’s going on there. That’s actually how people made a lot of discoveries. That’s how Barbara McClintock discovered transposons, with the different colors in the aleurone layer of the corn grain.

The white eyes that Thomas Hunt Morgan found in Drosophila—things like that.

McCarty: There are millions of gardeners. Gardening is an extremely popular hobby. And I do sometimes wonder why biotech doesn’t do more advocacy through gardening, right? Instead of having a community lab where people engineer E. coli to make insulin or something, why don’t we have a community lab where gardeners come and breed flowers? It’s biotech, but much more approachable. Is that something you’ve thought about?

Desnoyer: Absolutely. I think that’s an amazing approach because a lot of the pushback scientists get about genetic modification is because the general public either doesn’t understand it or just has no skin in the game. There are many years of education that you have to go through to even get into the lab. So it’s a big privilege, and you have this disconnect.

Whereas if the average person could already start to dabble in it a little bit, they would see that there’s nothing to be afraid of and that there’s nothing inherently wrong with modifying a plant. Just like anything else, it’s a tool, and you should use it responsibly. I think breeding is a nice kind of intermediate that anyone can get started with. And there are interesting things like the transposon line that has this active transposon that can mutate the flowers.

You don’t need to use CRISPR to find these mutations. You could already grow this in your garden, cross it with other flowers, and start to get cool mutants that you can introgress.

McCarty: Nick, what’s the next step?

Desnoyer: I want to complete this mission of creating the Paintunia. I think I’ve got about two more years before I finish the final product, which, again, is this petunia flower where you can apply copper, dexamethasone, or estradiol literally with a paintbrush onto the bud and create your own bespoke flower.

I would like to get this Paintunia into over 100 classrooms and have this be standard material for a university practical where people can visually see gene expression. They can modify a plant on a really short timescale, and they can blend art and science, so that even if that student doesn’t pursue biology, they’ll forever remember the Paintunia and how they triggered gene expression. That would be enough. However, to get there, I still need to fund the project because, as we’ve discussed, this is kind of hard when your goal is art and education and beauty.

I have a fellowship that I’m on right now, working on a completely separate research project, but that terminates next year, in August 2027. I’m looking to raise about $250,000 for that year to fund me, a research assistant, and an artist to tell the story of the making of the Paintunia.

I want to make this into a really accessible book and sell it very cheaply. Also, make a digital copy that has motion graphics that you can scroll through and see the entire story. And keep posting about it online so that people get live updates, and eventually they can get it into their own hands.

I’m seeking to do that fundraising, but also thinking: What comes after that? My grand vision is to make a flower design lab slash design studio that you walk into, and it’s a fusion of the modern lab and an art studio.

There are tablets for digital painting, cameras, and microscopes, but there’s also dirt for gardening. And just outside is a glasshouse where you can maintain all these plants. Now that I’ve been privileged enough to see the beauty of a flower and design it myself, I would like to have a place where you can invite others in so that they can do it themselves.

McCarty: It’s a great vision.

Desnoyer: Thanks.

McCarty: And you’re right that it’s so much more tangible than E. coli. When we teach biotech in schools, it’s like, “Oh, engineer the E. coli to have GFP.”

Desnoyer: Yeah.

McCarty: My experience of being here—everything is more visceral. When I walk around with you and see the different flowers, even the ones that people have just bred—flowers that they haven’t even engineered—you tell the story of, “Oh, here’s how they made this, and here’s why it looks like that, and it’s because of this molecular…” So you’re moving from the macroscopic all the way down to the molecular level.

Desnoyer: Totally.

McCarty: In a way that brings the biology together in a way that, for me, it almost never comes together.

Desnoyer: Right?

McCarty: And it’s not just the flower, but you talk about the ecosystem and insects that carry viruses. It’s just so beautiful to be a part of this. And thank you for hosting me.

Desnoyer: Thanks for being here. I’m going to miss you when you leave. And I’m sure we’ll work on more stuff together.

McCarty: And Nick, thanks also for doing the podcast.

Desnoyer: Thanks for having me.

Our goal at Asimov Press is to share the biggest stories in biotechnology. We do this by publishing original articles, as well as releasing technical and historical podcasts and printing beautiful, physical books.

This work is made possible by you, our readers, who shape our coverage more than you might expect. When we meet you at events, we often hear great ideas for stories but don’t know suitable writers to take them on. In a quest to find them, we are releasing our third list of Stories We’d Like to Publish. If any of these ideas speak to you, please send an email with the word “Pitch” in the subject line to editors@asimov.com.1

In your email, tell us why you’re the right person to tell this story, and also include examples of prior, non-academic writing (outlines are also welcome, so we can see how you might structure the article). If we accept the pitch, we’ll support you throughout the writing process, from outlining to editing, then copyediting and publishing. When the article is finished, we’ll pay you, host the article online and, perhaps, print your work in a forthcoming, physical book.

We are no longer seeking to publish “pure histories,” or articles that solely look to the past. History is important and might be integral to the article, but we are after arguments and ideas. We generally publish two types of pieces:

1. Reported features are stories that explain a technology’s promise while looking hard at its setbacks and current hurdles. These features should consider the future with an eye to the past, making both the science and the stakes clear. They should investigate the drama of discovery and the people moving a field forward, yet not shy away from prior decisions that might have led it astray. We typically pay $1-$1.50 per word for reported features and seek to work with experienced writers with a strong narrative skillset. We’re open to funding travel for in-person interviews.

2. Essays should make an argument, advance a big idea, or put forth a new way of looking at an idea. Essays, at their core, explain what should exist in the world and why. What problem are you trying to solve? What would it take to solve it? We want ideas that readers and funders can ingest, contemplate, and act on. We typically pay a flat rate of $1,200 for essays.

Here is a (high-level look) at some of the ideas we’d like to publish.

Wolbachia mosquitoes in Singapore

What happens when a city releases mosquitoes to prevent disease? In Singapore, researchers reared male Aedes aegypti mosquitoes carrying Wolbachia, a type of bacteria that prevents eggs from hatching when these males mate with wild females, and released them around the city. In a randomized trial across neighborhoods home to more than 724,000 people, the mosquitoes reduced the risk of symptomatic dengue by about 71–72%.

How did Singapore make this work in a densely populated city? How do you rear, sort, and distribute enough males, and how much does it cost to keep releasing them? How did residents react? A recent Works in Progress article, “Mosquitoes are a choice,” explored the idea of mosquito control and its regulatory barriers; now we’d like writers to build on that article via on-the-ground reporting to figure out what it would take to scale this technology elsewhere.

Freeze-drying any protein

Can we freeze-dry more proteins and other molecules so that drugs and diagnostics could be shipped without refrigeration? Some proteins, when frozen and dried, unfold or clump together, and finding a formulation that prevents this often involves a lot of trial and error. For a recent study, researchers used AI tools to create solid-state mRNA vaccines that retained their full bioactivity for two months at 37°C. The method worked by optimizing the ingredients around the vaccines and then drying them under vacuum.

This is a cool demonstration, but could we refactor more of our supply chain around dried molecules? It might allow for shipping more molecules where they’re needed; pharmacies in remote areas could keep powders ready to go, unrefrigerated. We’d love an idea-driven article examining how to make more proteins and other medicines stable without refrigeration.

DNA screening is a difficult problem

If someone orders a DNA sequence encoding a known pathogen or toxin, a DNA synthesis company can flag it and refuse to make the order. DNA screening algorithms are becoming quite good, but can they keep up with advances in biotechnology?

A recent Nature Biotechnology paper describes a method to assemble more than 1,000 gene fragments from a large oligo pool in a single reaction, with near-zero misalignment between genes. For a separate study, researchers at Arc Institute used Evo, a generative AI model, to design and synthesize nearly 300 bacteriophage genomes; 16 were viable, and several outperformed the natural ΦX174 template, despite having sequences that differed greatly from natural bacteriophages. How do DNA screening algorithms detect pathogenic sequences that do not share high homology with known pathogens? How might we engineer these algorithms to counter manufactured biothreats (not to mention benchtop DNA printers)?

Islet cell transplants for diabetes

It’s possible to treat type 1 diabetes by taking insulin-producing beta cells from a dead person’s pancreas and injecting them through a patient’s skin into their liver. The beta cells stay trapped inside the liver and pump out insulin, relieving some recipients from insulin injections for several years. Still, these patients need ongoing glucose monitoring and long-term immunosuppressive drugs.

There has been some recent progress in combining donor islet transplants with treatments intended to establish immune tolerance, but how close is this to widespread use? Is there a more permanent way to “cure” type 1 diabetes?

Gene delivery to the brain

A gene therapy can only work if it reaches the right cells. The brain makes this especially difficult, in part because its blood vessels restrict what can pass into surrounding tissue, and reaching one region does not mean reaching the various cell types spread throughout the brain. Researchers are trying many options to get genes into the brain; everything from direct injections (through the skull) to engineered viruses that cross the blood-brain barrier. One team redesigned an AAV capsid to bind the human transferrin receptor and showed that this improved brain delivery in mice engineered to express that receptor.

We’d like a reported guide to what works for brain deliveries, for which cells, and in which species. How do delivery through the bloodstream, spinal fluid, and direct brain injections compare? Which neuron types remain hardest to reach, and why do promising results in mice so often fail to translate? One way to focus this article might be to explore how people have tried to deliver gene therapies to the brain for, say, Huntington’s.

Gryphon Scientific

Gryphon Scientific was a small consultancy that grew into one of the most important biosecurity organizations. Founded in 2005, it worked at the intersection of biology and national security for years before those issues became mainstream. And when the U.S. government paused certain gain-of-function experiments involving influenza, SARS, and MERS, the guidelines were informed by a risk and benefit analysis performed by Gryphon. The company also did some of the early work on DNA synthesis screening, laboratory containment, and risks at the intersection of AI and biology. (And wrote one of our favorite papers of all time, “The accelerating pace of biotech democratization.”)

Deloitte acquired Gryphon in 2024, so now is a good time to revisit their work. How did Gryphon Scientific choose their projects, recruit people, and build credibility in the biosecurity space? We’d love a reported company profile of their work and early years; such a piece might offer useful context for other entrepreneurs working at the intersection of biosecurity and philanthropy.

On the unreasonable pain of engineering trees

North Carolina State University is famous for its engineered trees. It’s home to multiple tree engineers who use gene-editing tools, such as CRISPR, to modify how quickly eucalyptus, poplar, and other varieties grow, resist disease, or make wood. But this task is unreasonably painful. Even within one species, for example, trees can respond so differently to tissue culture and transformation that researchers have to frequently rebuild their protocol, even in the same species. You cannot just make an edit, grow a tree, and see what happens (as the process takes many, many years).

Given that a single experiment in tree biotechnology can take decades, how is it possible to design edits when iterations are so slow? We’d love a piece profiling “tree engineers” that also investigates what might be done to solve these bottlenecks. Trees are arguably the most important class of organisms on Earth, and yet this topic is largely unknown.

Release the sentinel plants

Plants can be engineered to act as a warning system for farmers. Researchers can create a gene circuit that senses a pathogen, toxin, drought or nutrient stress, coax it into plants, and then design an output module that causes a visible change in leaf color, the emission of a gas, or anything else that can be read out using a drone or tractor. Sean Cutler’s group at UC Riverside has engineered plants that turn red in the presence of a banned pesticide, for example, and a company called InnerPlant is developing soybean, corn, and cotton that fluoresce under stress. We’d like to publish a feature about how these sentinel plants work, and how they might be deployed at scale.

Once-weekly insulin

The FDA recently approved Onswik, a once-weekly insulin injection for people with type 2 diabetes. This molecule was engineered in some intriguing ways; for one, it’s attached to part of an antibody, which slows its clearance from the bloodstream. And two, the insulin protein was engineered to bind its receptor less strongly, helping it to act gradually over time. We’d like to publish an article about these once-weekly insulins, how they work at a molecular level, and what difficulties arose before the FDA approval.

Protein logic drugs

Researchers have been making “smart” insulin proteins that dynamically change their binding activities based on how much glucose is in the bloodstream. One of these experimental therapies, called NNC2215, is an insulin protein attached to a glucose-binding ring and a small chemical group that looks a lot like glucose. So when glucose is low in the blood, these chemical attachments hold the insulin in a shape that makes it harder for the protein to bind to its receptor. But when glucose levels rise, the glucose molecules in the bloodstream compete for the binding site and favor the more active shape. We’re curious about whether similar sorts of dynamic therapeutics are being explored for other conditions. How do these “logical proteins” work mechanistically, especially given advances in protein design tools?

Plants that clone themselves

Some plants naturally make seeds without sex; their offspring are genetic clones of the parents. Researchers are trying to take this trait (naturally found in dandelions, for example) and engineer it into rice and other important crops. If they succeed, the major seed companies—which currently cross plants every year, harvest their seeds, and sell them to farmers—could be circumvented. Farmers would be able to plant their crops and reuse their seeds year after year, because the F1 hybrid “vigor” would remain unchanged.

Researchers have already produced hybrid rice that generates more than 95% clonal seed across several generations, but transferring the system reliably to other crops has been difficult, with problems related to seed yield and the stability of clonal reproduction. We’d like to publish a feature explaining synthetic apomixis, why it has taken so long, and what it might enable at scale.

Jellyfish sensors for ocean mapping

John Dabiri’s group at Caltech is building electronic devices that can be attached to living jellyfish and used to measure pH, pollution, and other things in the ocean. His plan is to use these animals to map the ocean by taking measurements through waters they naturally travel, while transmitting data along the way. This would provide an alternative to expensive submersibles. Could we similarly attach chemical sensors to birds to measure air pollution? Would that even be ethical? We’d like to publish a profile of Dabiri that considers this emerging approach to animal sensing.

Real-time RNA sequencing

Transcriptomics is a method to measure the abundance of messenger RNA in a cell. The gist is that you grow cells in a culture (or individual cells in a microfluidic chip), slice them open, extract their RNA molecules, and convert the molecules into DNA copies. Then, you use machines to “count” the abundance of each sequence. The approach has a downside, though, in that the readout reflects only the quantity of RNA molecules at the particular moment when the cells were cut open. Cells are constantly dialing RNA abundances up or down (as gene expression is not static), but this dynamism isn’t captured.

We’d like to commission a piece on “real-time RNA-seq,” or the plethora of methods that enable RNAs to be studied across time. Live-seq, for example, is a method to extract RNA molecules from single cells without breaking them open. Felix Horns, a researcher at Arc Institute, has also developed a method to coax cells to export their own RNAs, enabling the molecules to be studied over time. Why aren’t these methods widely used? Are they biased or finicky? What will it take to make them mundane?

Noninvasive biology

In a related question, we think it’s very strange that so many methods used to study biology measure a particular molecule at a particular moment in time. These methods are incredibly one-dimensional, and we could do so much more in biology if we had ways to study molecules across space and time without killing the cell.

Historically, we could fuse fluorescent proteins to a target in a cell and then watch those cells under a microscope to see where and how that protein moves. But there are all kinds of other interesting noninvasive methods coming online. For example, hyperspectral imaging could theoretically be used to quantify molecular abundances without destroying the cell. Raman spectroscopy similarly allows for studying cells by measuring the vibrational frequencies of molecules. And Dan Needleman at Harvard is developing imaging methods to study metabolism in living cells. We’d like articles on “future” technologies that help normalize and encourage people to work on them.

Getting clinical data from inside the body

Let’s say a drug researcher has invented a new therapy for cancer. The drug is given to patients in a clinical trial, and tumors are measured for shrinkage. While symptoms are tracked, it can be difficult to measure exactly how much drug reached the tumor, which cells it entered, or whether the tumor shrank because of direct or secondary effects. Biopsies, scans, and blood tests help answer some of these questions, but not very well.

There has to be a better way to study how drugs behave inside the body with higher spatial resolution. Consider CAR T-cell therapies, where T-cells are taken from a patient’s body, engineered to recognize the patient’s cancer and then re-infused. The T-cells circulate, find cancer cells, and destroy them. Researchers have engineered CAR T-cells to express gas vesicles (little protein shells that exclude water, and thus show up on ultrasound scans) and tracked them in mice for up to five days. Could similar methods enable us to follow these cells inside patients? There are many other things that could be measured in vivo, too, if only we had better methods.

Blood draws are outdated

Hims & Hers acquired YourBio Health, which makes devices that use tiny needles and suction to collect blood from capillaries in the upper arm. Its TAP Micro Select device can draw up to 500 microliters, or half a milliliter, of blood with the push of a button. In one study, participants rated their pain when using several upper-arm devices, including TAP, and ranked this device below a 1 on a 10-point pain scale. So how is it possible to make a blood draw hurt so little?

A piece on this idea should explain how these capillary blood draws were invented and why we don’t yet find them everywhere. Why, when we go to the doctor, do we still give vials of blood? If it’s because certain assays require a large amount of blood, how can we change those assays to test for more things using less blood? This might sound like Theranos, but we’d love a piece examining which assays these devices are compatible with, which they aren’t, and how their use might be extended.

A “universal” nanopore

The holy grail of biosensing is arguably to create a universal solid-state nanopore. We’ve had several conversations with the founder of Demonpore, a company trying to build this technology, but it’s been hard to get people in the field to speak to us about the bottlenecks. Today, nanopore sensors are best known for DNA sequencing, where individual DNA molecules pass through a tiny protein pore and alter the ionic current flowing through it. In principle, the same basic idea could be extended much further; a solid-state nanopore, for example, is simply a tiny hole (not a protein) drilled into a membrane, and different molecules passing through that hole should perturb the electrical current in different ways. If we collected enough data on the electrical signatures produced by amino acids, proteins, RNA molecules, nucleotides, and other compounds, could we train a model to distinguish among them and turn a nanopore into a general purpose sensor for biology?

Egg manufacturing

Vaccines for seasonal influenza have been manufactured in chicken eggs since the 1940s. But eggs could make much more than vaccines. Neion Bio is engineering chickens to make therapeutic proteins—including monoclonal antibodies—in their eggs. Eggs are abundant, and each contains roughly six grams of protein. When Ozempic soared in popularity, there were worries over manufacturing constraints. Could eggs help address manufacturing constraints for some drugs?

What are the bottlenecks? Are GMP-grade manufacturing requirements holding it back? Are eggs too variable from one batch to the next to ensure drug consistency? (For example: Not all influenza strains replicate equally well in eggs, and growing them in eggs can select for changes that affect vaccine performance.)

Bottlenecks on microfluidic chips

Microfluidic chips allow for the manipulation of tiny volumes of liquid and make it relatively easy to study single cells, but making the chips can be finicky. Many methods require photolithography, UV light, specialized equipment, and a lot of technical expertise. The dream would be to have a 3D printer that could just print the chip you want. Researchers already print channels only tens of micrometers across, but achieving the resolution needed remains difficult.

What are the issues holding back microfluidic manufacturing? How do we scale this so that more labs can use microfluidic 3D printers instead? Related: Where Is My Mother Machine?

Can we eradicate cervical cancer?

We have vaccines that can prevent most cervical cancers, and yet the disease still kills hundreds of thousands of women. WHO estimated 660,000 new cases and 350,000 deaths in 2022. In one Swedish study, though, HPV vaccinations given before age 17 were associated with an 88% lower risk of cervical cancer.

Is it possible to eliminate cervical cancer? WHO’s goal is fewer than four new cases per 100,000 women each year, but we’d like to publish a feature about how close we are to full eradication. Where are rates already falling, and what combination of vaccination, screening, and treatment would let other countries follow? (h/t Saloni Dattani)

Hepatitis is weird

Hepatitis B and C both infect the liver and can cause cirrhosis and cancer. But our tools against them have a strange asymmetry. We have a highly effective vaccine against hepatitis B, yet treatment merely suppresses a chronic infection instead of curing it. For hepatitis C, antiviral drugs can cure more than 95% of infections, but we still have no effective vaccine. WHO estimates that 240 million people were living with chronic hepatitis B in 2024.

We’d like a mechanistic, reported piece about the missing hepatitis B cure and hepatitis C vaccine. We’re open to treating these as two separate stories, or as one big feature.

Biotechnology in the Global South

Biotechnology innovation, historically, has mostly come in the northern hemisphere, particularly in the United States, Europe and (more recently), China. The world would benefit from biotechnology flourishing in the Global South, as well. There are deficits in venture capital and research funding in many countries of the Global South, as well as infrastructure issues around cold storage and the electric grid. A lack of biologic materials hinders progress. There are talent issues, as well, with young researchers leaving to train in established and well-funded labs elsewhere, and sometimes staying.

What other issues might be holding back biotech progress in the Global South? What countries are focusing on building a bio-economy, and are their goals achievable? And what do promising young scientists in the global south think about all of this? We’d love a reported feature on this topic.

How $1B remade Boston

The state of Massachusetts made a 10-year, $1 billion investment to make itself into a biotech and pharma hub. The initiative overhauled Cambridge, Massachusetts, and the surrounding universities. It also bore fruit with drug approvals. With a lack of clarity around future federal funding for basic science, perhaps it is time to again use state tax dollars to drive innovation and growth in biotech. California is moving forward with its own plan, but what lessons can be learned from Boston’s earlier initiative, and can they be applied to other states? We’d love a feature investigating this past and applying it to the future.

Editing mice against Lyme disease

Mice Against Ticks is a collaboration between MIT and Tufts University, with the aim of making white-footed mice immune to Lyme disease. Mice are a reservoir for the disease, and the initiative is seen as a possible tool for also fighting hantavirus, arenavirus and leptospirosis. Many ecological issues arise around engineering the genome of an animal in the wild, of course. The initiative has said it prioritizes local input (the project began in and around Nantucket and Martha’s Vineyard in Massachusetts) and responsibility, but how has that worked in reality? What results have been seen so far, and where has data been published? And how would a relatively small scale field trial be scaled to curtail Lyme disease across the country? Meanwhile, Lyme (and alpha-gal syndrome) is on the rise, with the CDC estimating that nearly half a million Americans are treated for the infection each year. Having seen how unpleasant it can be up close, we’d love to see a reported feature on the technology behind Mice Against Ticks, what it might be able to solve, and the problems inherent in genetically modifying a population of wild animals.

Pig genome engineering for organs

How much of a pig’s genome has to change before its organs can work inside a person? For one study, researchers made 69 edits to the pigs, including changes that removed immune-provoking sugars, added human proteins, and inactivated copies of porcine endogenous retroviruses. A kidney from this design was transplanted into a living patient in 2024, though the patient died 52 days later. Other groups are pursuing different combinations of modifications, and it’s not clear which approach is most likely to succeed and why.

We’d like an article that explains the reasoning behind these edits and what the transplant results tell us about them. How far away are we (actually) from pig organs being transplanted into humans and working inside of them for decades? Why do these studies keep failing?

Paper-based sensors

Similarly, one can combine cell extracts with a synthetic gene circuit, freeze-dry the mixtures onto paper, and then reactivate those molecular circuits with water, saliva, or other liquids. These cell-free circuits can be programmed to sense a particular RNA sequence, protein, small molecule, or toxin. By freezing this molecular program in time on a piece of paper, it basically becomes a portable biosensor. Researchers used these paper-based biosensors to detect Ebola and Zika in the laboratory, and even showed they worked on real patient samples for chikungunya.

But progress seems to have stalled. Were these sensors not sensitive enough, or did they get outcompeted by RT-LAMP and other assays? We’d love an article that tells the story behind these paper-based sensors and where they stand today.

Portable vaccines

Many years ago, researchers showed that you can take molecular machinery from engineered bacteria, freeze-dry it, and then “reactivate” those molecules to create a vaccine. These cell-free systems can read DNA sequences dropped into a vial, basically, and then read it to make proteins. Researchers showed it’s even possible to attach bacterial sugars to these proteins, in these vials, to make conjugate vaccines. The reaction takes about an hour, and the freeze-dried mixtures can be stored at room temperature. In one experiment, a vaccine made from these cell-free mixtures protected mice from a normally lethal bacterial infection.

These “portable vaccines” were touted as a way to make custom vaccine manufacturing more feasible in low-income countries. But they have not yet surfaced. Did they fail for economic, geopolitical, or technical reasons?

Theories are feats of scientific imagination that illuminate and unify our understanding of reality. Far from being mere generalizations of empirical data done after the fact, theories orient us toward the unseen, toward what lies “beyond our grasp.” As systems biologist Jeremy Gunawardena puts it,

If scientific research is stumbling around in a dark cellar looking for a black cat, then biology is doing so without knowing there is a cat there until one accidentally falls over it. Theory can sometimes conjure up the cat before the accident.

In biology, the cats in question are particularly elusive. Sometimes you don’t even get to fall over them, because all you can see is a smile lingering after the rest of the cat melted into the air. You may hypothesize the existence of ion channels or cell receptors but it might take years or decades before someone else identifies them experimentally. In the meantime, you have to hold on to the cat’s smile because it is so suggestive of the whole cat and explains so much.

But imagining that an unseen entity exists is only one kind of theoretical leap in biology. Some cats instead belong to a whole clowder, and you must figure out not only the cats themselves but also how they interact with one another. In other words, you must imagine how an entire biological process works: which components are involved, how they interact, and what those interactions are used for. In some cases, theoretical biologists have made such leaps in the absence of any direct experimental evidence. These are amongst the most imaginative achievements of the scientific mind.

Two of them, the chemiosmotic theory and the clonal selection theory of adaptive immunity, are particularly noteworthy. They proposed mechanisms for two fundamental processes of life — how cells generate energy to power themselves and how the body mounts a defense against pathogens — before any pre-existing analogs were known.

What were the reasoning paths that led to these theories? Can they be traced at all?

It is rarely acknowledged to what extent biologists’ broader philosophical commitments guide their theoretical thinking. This is especially evident in the two theories above. In both, the scientists’ philosophical background allowed them to break away from existing theoretical frameworks and offer their own, at first seemingly outlandish and bizarre, but later ultimately experimentally validated, theories. By examining the ideas they borrowed and assumptions they questioned, we can begin to understand how such feats of imagination become feasible and learn to make others.

Chemiosmotic Theory

The British biochemist Peter Mitchell formulated his chemiosmotic theory of oxidative phosphorylation in 1961 without “a shred of experimental evidence.” But the theory didn’t come out of nowhere either. It was ultimately a result of a lifelong development of Mitchell’s idiosyncratic philosophical framework related to general foundations of life.

When he was a graduate student at Cambridge in the 1940s, Mitchell proposed a system of entities called statids, fluctids and fluctoids, echoing the ideas of the pre-Socratic philosopher Heraclitus of Ephesus. As their names suggest, statids are stable structural elements while fluctids are constantly in flux, exchanging matter with the environment, like a flame. Fluctoids are a combination of a flowing element with a static element, like a Bunsen burner.1 Though this framework may seem too abstract and far-fetched to be useful in biology, Mitchell integrated it productively into his life’s work in various biological contexts, culminating in the formulation of chemiosmotic theory.

The central problem that chemiosmotic theory addresses is the source of the energy that fuels ATP synthesis in mitochondria during cellular respiration — a process known as oxidative phosphorylation (so named as it requires oxygen to proceed and results in the phosphorylation of ADP into ATP).

The prevailing chemical theory of oxidative phosphorylation at the time postulated the existence of a “high-energy” chemical intermediate, similar to what happens in glycolysis.2 But no such intermediate was ever isolated, despite persistent attempts. On the other hand, it wasn’t possible to reproduce oxidative phosphorylation in a membrane-free in vitro system, even though the chemical theory didn’t require membranes.

The leap of imagination that Mitchell made in 1961 was to connect the metabolic reaction of ATP synthesis with the active transport of ions across membranes. In an earlier paper from 1959,3 Mitchell presciently wrote:

In view of certain similarities between mitochondrial membranes and the plasma- membranes of bacteria, I would venture to suggest that the function of the membranes of mitochondria and of the endoplasmic reticulum may be, like that of the plasma membrane, to act as chemi-osmotic links between the media that they separate.

This connection was in line with the innovative concept of “vectorial metabolism” that Mitchell developed — metabolic processes that have a direction in space, as opposed to “scalar metabolism” where the direction doesn’t matter, as was generally assumed to be the case in biochemistry at the time. A scalar enzymatic reaction could become vectorial if the reacting substances arrived on one side of the membrane and the products left on the other.

Because he wasn’t constrained by the then prevailing chemical theory, Mitchell was able to resolve the two major issues that the chemical theory was unfit to explain — the lack of a chemical intermediate in the process in question and its dependence on membranes. In Mitchell’s terms, the chemiosmotic theory of oxidative phosphorylation describes a fluctoid system with a structural statid element, the semipermeable membrane, and a flowing fluctid element, the protons and electrons moving across the membrane, thus providing energy for ATP synthesis.

Mitchell’s conception of vectorial metabolism was also influenced by his readings of early 20th century biologists and philosophers of science, including D’Arcy Thompson, Joseph Needham and Joseph Woodger. Woodger’s ideas were especially crucial: first, the notion of life as persisting through change in response to the environment, and second, the conception of the cell as a “spatio-temporal structure in which the various kinetic processes are an integral part of that structure.” From this Mitchell inferred that the interaction with the environment (through the transport across cellular membranes) and the processes happening inside the cell (such as metabolism) had to be inextricably linked.4

Mitchell was also aware of earlier studies that showed dependence on respiration of the mineral ion uptake from the soil by plant roots, and the same dependence was demonstrated for acid secretion in the stomach. These studies hinted at a tantalizing connection between the process that feeds into oxidative phosphorylation (respiration) and the ion transport across membranes. Taken together, Woodger’s theoretical considerations (which in essence aligned with Mitchell’s original fluctoid system) and the indirect experimental evidence from plant roots and animal stomachs provided the basis for Mitchell’s leap into formulating chemiosmotic theory.

But the theory wasn’t met with enthusiasm among his contemporaries, and Mitchell, who also suffered from gastric ulcers, left academia in 1963 to set up his own organization, the Glynn Research Institute, with his family’s financial support.5 There he continued to revise his theory, working on its experimental verification for the next few decades. As experimental support for the chemiosmotic theory accumulated over time, the resistance against it in academic circles subsided, and Mitchell was awarded the Nobel Prize in Chemistry in 1978.

Clonal selection theory of immunity

The self vs. non-self recognition that is at the heart of the immune response represents a unique philosophical problem in biology, and understanding it has required theoretical ingenuity far ahead of what was known from contemporaneous experimental work. The clonal selection theory of immunity describes a fundamental process by which an organism develops its ability to recognize and mount an adaptive immune response against any foreign agents it may come in contact with.6 This process was deciphered over decades, spread out among several scientists — the German Paul Ehrlich, the Dane Niels Kaj Jerne, the American David Talmage and the Australian Macfarlane Burnet. All except Talmage were awarded, in different years, the Nobel Prize in Physiology or Medicine for other discoveries, but none for the clonal selection theory itself.

The origins of the theory date back to “the father of immunology”, Paul Ehrlich,7 who in the late 19th century anticipated many of the fundamental principles of adaptive immunity, some inaccuracies notwithstanding. Without any direct experimental evidence,8 he suggested that all cells in the body carry Seitenketten (“side-chains” in German) that normally bind nutrients, and that antigens from foreign substances like toxins can be recognized by some of those side-chains (receptors). As a result of this interaction, the cells are functionally activated and produce an excess of antigen-specific side-chains, which are then released into the bloodstream as antibodies that neutralize the toxin. He also considered that antibody formation could be a dedicated function of the cells of “haemopoietic tissue.” What he didn’t realize is that each antibody-producing cell bears receptors of a single-antigen specificity, rather than a variety of them at once.

Most presciently, Ehrlich proposed that there must be a preexistingrepertoire of antibody specificities for a variety of antigens that an organism may encounter throughout its lifespan, and that the antigens act to select from among this pool of specific antibodies. This appears to be a major leap of imagination on the part of the German scientist. However, Ehrlich was himself somewhat conflicted about this point:

It would not be reasonable to suppose that there were present in the organism many hundreds of atomic groups destined to unite with toxins, when the latter appeared, but in function really playing no part in the processes of normal life, and only arbitrarily brought into relationship with them by the will of the investigator. It would indeed be highly superfluous, for example, for all our native animals to possess in their tissues atomic groups deliberately adapted to unite with abrin, ricin, and crotin, substances coming from the far distant tropics.

By this time it was already experimentally shown that antibodies against those substances can in fact be raised in animals. Combined with the evidence from the 1920s9 that specific antibodies can be induced even in response to synthetic chemicals, this led to the abandonment of Ehrlich’s preformation theory and an increasing adoption of alternative, template (also known as instructionist) theories. Such theories postulated that the antigens themselves act as a template, “instructing” the formation of a complementary antibody structure.10 Linus Pauling, for example, suggested that newly synthesized native globulin molecules can fold around antigen molecules, with antigens thus serving as templates that transform the otherwise neutral blood proteins into specific antibodies.

But in the 1950s, Niels Jerne dealt a decisive blow to the template theories. To be sure, these theories had multiple known problems, including their inability to explain the prolonged production of antigen-specific antibodies after the antigen had cleared from the organism (how could such antibodies persist in the absence of a template?). Neither could they account for the booster phenomenon — the dramatic increase in the concentration of antigen-specific antibodies after a repeated injection of an antigen.

Finally, the template theories were also incompatible with another empirical observation — that the avidity of antibodies (the “tightness” of their binding to the antigen, measured by how quickly they neutralize it) increases significantly in the course of immunization, with later antibodies becoming much more avid than the early titres. Jerne’s natural selection theory solved all of these incongruities by postulating a mechanism remarkably similar to Ehrlich’s preformation theory, and yet independent from it, and with a more developed philosophical grounding.

Before he formulated his theory in 1955, Jerne spent more than a decade researching the avidity of antibodies and developing new sensitive quantitative assays along the way. In particular, he used plaque formation by bacteriophages as an indirect readout of antibody binding. In a crucial series of experiments, he immunized a horse with high titres of bacteriophage T4 and took serum samples over a time series.

Jerne found T4-specific antibodies already in the early first samples, but more surprisingly, he was able to detect anti-T4 activity even in the controls — in the serum of non-immunized animals. This could either be because they had already been exposed to the T4 bacteriophage earlier in their lives (which would be in line with the template theories) or, thrillingly, because it was instead a “natural antibody” spontaneously produced by all healthy animals even if they had never encountered T4 before. Jerne went with the latter interpretation, even though, strictly speaking, his experiments did not demonstrate that.11 It was a theoretical supposition that relied more on Jerne’s philosophical considerations than on direct empirical evidence.

Why was the existence of natural antibodies so important? Scores of immunologists before Jerne had been aware of the phenomenon,12 yet none went in the theoretical direction he pursued. But Jerne had a rigorous background in biostatistics and a keen interest in Neo-Darwinism through the works of R. A. Fisher and Max Delbrück. This unique intellectual background prepared him to make the inference that the presence of natural antibodies was an indication that a selection process was at play. From a preexisting pool of continuously produced natural antibodies, antigens select ones that can inherently bind them (rather than “instruct” their formation de novo), which leads to the amplification of this specific fraction of antibodies.

In other words, “the function of the antigen is to exert a population pressure on the distribution of a heterogeneous globulin population,” as he wrote in a letter to a colleague.13 In a way, Jerne pioneered the use of Darwinian ideas to explain physiological phenomena:

Darwinian ‘selection of the fittest’ had hitherto been applied only with regard to the diversity of plant and animal species. I think I was the first to propose that the Darwinian selection principle was also possible and indeed applied, within the diversity of cells within a single polycellular organism, i.e. ‘physiologically.’14

In addition to Neo-Darwinism, there was another philosophical influence that Jerne highlighted in his famous retrospective essay, The Natural Selection Theory of Antibody Formation: Ten Years Later (1966). It seems that his idea of the preexisting antibody repertoire drew in part from the interpretation of the Socratic notion of “learning as recollection” by the Danish existentialist philosopher Søren Kierkegaard:

Can the truth (the capability to synthesize an antibody) be learned? If so, it must be assumed not to pre-exist; to be learned, it must be acquired. We are thus confronted with the difficulty to which Socrates calls attention in Meno (Socrates, 375 B.C.), namely that it makes as little sense to search for what one does not know as to search for what one knows; what one knows one cannot search for, since one knows it already, and what one does not know one cannot search for, since one does not even know what to search for. Socrates resolved this difficulty by postulating that learning is nothing but recollection. The truth (the capability to synthesize an antibody)cannot be brought in, but was already inherent. [parentheses added by Jerne]

Accordingly, Jerne’s theory predicted that there must be a “random mechanism” that generates the initial diversity of antibodies in the absence of foreign antigens. It additionally postulated a “purging mechanism” that ensures the elimination of antibodies which recognize the body’s own antigens. However, his “selective mechanism” for the propagation of the best-fitting antibodies wasn’t flawless: natural selection acted at the molecular level, on the antibodies themselves, but the theory couldn’t convincingly account for how the antibodies expanded and multiplied in numbers in the course of an immune response.

Jerne did suggest that the antibody-antigen complex is somehow engulfed by a phagocytic cell and transported to other cells that can produce more antibodies of the same kind. But this threatened to challenge the central dogma of molecular biology: How can proteins cause synthesis of more of their kind? That would require a reverse flow of information from proteins to DNA. Thus, several scientists, including Linus Pauling and James Watson, rejected Jerne’s theory outright, though his more Darwinian-minded colleagues, like Günther Stent and Joshua Lederberg, welcomed it more enthusiastically.

Jerne’s theory didn’t perish but was instead modified into a cellularselection theory. In a 1957 review David Talmage credited Jerne with originating the modern idea of selection in immunology and added a tentative but crucial detail (a rare case where the advancement of science happens in a review article):

… it is tempting to consider that one of the multiplying units in the antibody response is the cell itself. According to this hypothesis, only those cells are selected for multiplication whose synthesized product has affinity for the antigen injected. This would have the disadvantage of requiring a different species of cell for each species of protein produced, but would not increase the total amount of configurational information required on the hereditary process. (italics added)

This shift from the selection at the antibody level to the selection at the cell level proved decisive. The Australian immunologist Macfarlane Burnet was also among those who responded favorably to Jerne’s theory, especially given its Darwinian mechanism (in contrast to the Lamarckist implications of the template theories), and found it“attractive, though obviously wrong.” Burnet’s solution was to replace antibody molecules with cell clones and their membrane receptors, similar to what was proposed by Talmage earlier that year.15

In a paper published in 1957, which he judiciously titled A modification of Jerne’s theory of antibody production using the concept of clonal selection, Burnet proposed that large numbers of lymphoid cell clones were programmed to produce antibodies of a large variety of distinct specificities. When an antigen enters the body, the cell clones that happen to produce antibodies against that antigen are thus “selected” for and expand into much larger clones, which start releasing more antibodies against the invading antigen.

This is now known as Burnet’s clonal selection theory. To his credit, Burnet continued to acknowledge Jerne’s role in its origins: “As I hope I have always been careful to say, its ‘onlie begetter’ was Niels Jerne.”16 Interestingly, for Burnet the main virtue of the theory was in explaining the “self vs. non-self” distinction, which became a major focus of his later investigations.

Yet as Burnet initially had his doubts about the clonal selection theory, it is telling that he published it in the obscure Australian Journal of Science, explaining the choice in his autobiography: “If, perish the thought, there was something very wrong about the clonal selection concept, the publication in an out-of-the-way Australian journal would mean that very few people in America or England would see it.”

Happily, the first confirming experimental evidence for the clonal selection theory arrived in 1958 when Gustav Nossal and Joshua Lederberg showed that each lymphoid cell produces only one kind of antibody by binding specificity. In 1976 Susumu Tonegawa’s discovery of V(D)J recombination further clarified the molecular foundations of the generation of antibody diversity.

In his autobiography, Burnet ultimately stated, “I regard the development of the clonal selection theory of immunity as my most important scientific achievement.”

When a biologist considers a theory his main contribution to science, that says something! Indeed, the clonal selection theory of immunity laid the foundations for understanding immunological memory and immunological tolerance (it was for its discovery that Burnet shared a Nobel Prize in Physiology and Medicine with Peter Medawar in 1960), and eventually made the development of monoclonal antibodies possible. Its organizing force in immunology was akin to what Mendeleev’s periodic table did in chemistry or what, indeed, Darwinian evolution did in biology as a whole.

To discern order in a dizzying complexity of nature is theoretical thinking at its finest. Theories provide explanations for our observations and unify existing disjoint bodies of knowledge. But they also leap forward, ahead of what we can assert based solely on our observations.

The most overarching theories, like universal laws of physics, light up the whole dark cellar for us to see the various cats that might be hiding there, to come back to Gunawardena’s image. A more circumscribed mechanistic theory in biology helps us predict the location and shape of a particular cat (or a bundle thereof) so we can prepare for our encounter with it, by designing experiments to identify it (or realize that it is not there at all), rather than stumbling upon it serendipitously in the dark.

Ulkar Aghayeva is a science writer and columnist at Asimov Press. She also writes at the science history blog Measure for Measure and music blog The Bass Line.

Header image by Ella Watkins-Dulaney.

Cite: Aghayeva, U. “Great Leaps in Biological Theory.” Asimov Press(2026). DOI: 10.62211/56hq-78uy

“There is a mask of theory over the whole face of nature, if it be theory to infer more than we see.”

— William Whewell, 1840

“I have an old belief that a good observer really means a good theorist.”

— Charles Darwin, 1860

Introduction

At the beginning of his famous lectures on physics, Richard Feynman describes experiment as the foundation of empirical science and shows where theory enters into scientific practice:

The test of all knowledge is experiment. Experiment is the sole judge of scientific ‘truth.’ But what is the source of knowledge? Where do the laws that are to be tested come from? Experiment, itself, helps to produce these laws, in the sense that it gives us hints. But also needed is imagination to create from these hints the great generalizations—to guess at the wonderful, simple, but very strange patterns beneath them all, and then to experiment to check again whether we have made the right guess. This imagining process is so difficult that there is a division of labor in physics: there are theoretical physicists who imagine, deduce, and guess at new laws, but do not experiment; and then there are experimental physicists who experiment, imagine, deduce, and guess.

In biology, by contrast, theory and experiments seem to be more entwined and are usually practiced by the same scientists.

And yet, in a provocatively titled 2014 article, Biology is more theoretical than physics, systems biologist Jeremy Gunawardena challenged this view about the relationship between biological theories and experiments. He made a case for a number of biological entities that were postulated – imagined – to exist theoretically long before they were discovered experimentally, in a “division of labor” similar to that described by Feynman in physics. These theorized entities include such fundamental concepts as genes, receptors, enzymes, ion channels, and tumor suppressors.

But even more impressive are the theories that predicted entire mechanisms of biological processes and deciphered fundamental principles of life itself, way ahead of their experimental demonstration. Among them are the genetic code, the chemiosmotic theory and the clonal selection theory of adaptive immunity. Some of these theories emerged untethered to experimental evidence and yet proved correct, while others were developed in a tighter feedback loop with experimental work.

Still, theoretical biology doesn’t seem to have attained the same standing and institutional support that is granted to theoretical physics, or, for that matter, other fields that study complex systems, such as economics. I’ve long wondered why this is.

A look at the history of ideas in theoretical biology offers an answer. Theoretical biology encompasses a surprising variety of subdisciplines, ranging from biosemiotics to modern systems biology. This variety is a testament to the soaring imagination of biologists in their attempts to make sense of the living world, and at the same time to their persistent desire to model biological theories on the more established and successful theories in physics.

Through this examination, it might be possible to also address the more fundamental question: can biology be as compressible and comprehensible as physics? Or are biological systems unique in their complexity in that they don’t lend themselves to compact theoretical description in the manner of physical systems? Though the ultimate answers to these questions still remain out of reach, it turns out that there are, in fact, surprising commonalities between the models that describe the behavior of both physical and biological systems.

The Varieties of Theoretical Biology

Theoretical explanations of biological phenomena date back to the early days of natural philosophy. But the notion of theoretical biology as a distinct field of inquiry is much more recent, originating in the early 20th century. As the zoologist William A. Locy wrote in his 1908 book, Biology and Its Makers:

In the rise of biology the facts have accumulated constantly, through observation and experiment, but the general truths have emerged slowly and periodically, whenever there has been granted to some mind an insight into the meanings of the facts. The detached facts are sometimes tedious, the interpretations always interesting.

Some of the earliest theories in biology persisted for centuries until they were refuted or subsumed into more modern theories. For one, the theory of spontaneous generation posited the emergence of living things like fleas, maggots and frogs out of non-living matter like mud or decaying meat. It was first fully articulated by Aristotle in his History of Animals, based on the works of earlier Greek natural philosophers. The long debates over spontaneous generation were finally resolved by the simple but ingenious experiments of Louis Pasteur in the early 1860s that decisively ruled out this theory.

Sometimes, overturned theories live on—albeit in a modified form—within modern biology. The two contending views on embryonic development, known as epigenesis (dating back to Aristotle’s Generation of Animals) and preformationism (espoused by the 17-18th century anatomists and microscopists), differed on whether the embryo develops gradually from unformed material or has fully pre-formed parts, coming from the egg or the sperm, that grow into an adult organism. We now know that development is, indeed, partly specified in advance and partly emergent through cellular interactions, so in a sense both competing theories have been incorporated into modern embryology, though as such they remain only a matter of historical interest.

Early theories in biology were often formulated as verbal descriptions of the inner workings of a living organism. But gradually, biology started reorganizing itself around mathematical methods, following the Kantian principle that “in every discipline of natural sciences there is present only so much real science as there is mathematics.”1

By 1680, the Italian mathematician Giovanni Alphonso Borelli introduced the earliest work of biomechanics, De Motu Animalium (“On Animal Motion”).2 Published just seven years before Newton’s Principia Mathematica, it relies on geometry and Galilean static mechanics to explain limb movements in humans and animals, including running, jumping, swimming, and flying. Borelli also applied his mechanical approach to basic physiological processes such as breathing, heartbeat, blood circulation, and liver function. His book can be regarded as the earliest formal work in theoretical biology.

A direct descendant of Borelli’s line of investigation is D’Arcy Wentworth Thompson’s On Growth and Form, published in 1917.3 In it, the Scottish biologist and mathematician argues that biological form is a consequence of physical processes, such as crystallization, adsorption and diffusion, and of mechanical forces that constrain the evolution of organismal morphology. Needless to say, evolutionary theory itself (1859) was a major feat of theoretical biology and one of the most generative theoretical contributions in all of science.

To study physical constraints on biological systems, Thompson wrote, one has to make extensive use of mathematical and physical methods, the way Borelli did for the mechanics of motion:

Cell and tissue, shell and bone, leaf and flower, are so many portions of matter, and it is in obedience to the laws of physics that their particles have been moved, moulded and conformed. They are no exception to the rule that Θεὸς ἀεὶ γεωμετρεῖ (‘God always does geometry’). Their problems of form are in the first instance math­e­mat­i­cal problems, and their problems of growth are essentially physical problems; and the morphologist is, ipso facto, a student of physical science.

Thompson applied this vision to entire organisms, devising a geometric method where a Cartesian coordinate grid is projected onto an organ (or a whole organism) and then undergoes a geometric transformation, such as compression or dilation, to produce the body shape of another related species. In this way, a porcupine fish can be transformed into a sunfish:

Admittedly, On Growth and Form was still largely descriptive, without providing a mechanistic understanding of the generative processes behind the development and morphological variety of organisms. But the book was influential and inspired more rigorous research in embryology, including discovery of morphogens and tissue patterning.

In the early 20th century, theoretical biology emerged as a distinct and a more self-aware field of research. German and Eastern European scientists in particular approached it with high ambitions, inspired by the success of theoretical physics in compressing and generalizing physical phenomena into natural laws.

They used the term “theoretical biology” to describe a range of newly emerging disciplines: philosophy of biology, general theories of living matter and physico-chemical foundations of life, attempts to mathematicize biology (especially in ecology and population genetics), and later systems theory and cybernetics. Some of these theories were developed by biologists themselves while others owe their origin to an influx of philosophers, physicists and mathematicians into biology.

Among German biologists, theoretical biology was initially associated with philosophy of nature, epistemology, and metaphysics rather than with formulation of scientific hypotheses and theories. The botanist Johannes Reinke, who coined the term “theoretical biology” in 1901 to differentiate it from “empirical biology,” was himself a proponent of the revived philosophy of neo-vitalism.4 Jakob Johann von Uexküll’s Theoretical Biology, published in 1920, was to a large extent also a philosophical work.

Best known for introducing the notion of Umwelt,5 Uexküll was one of the pioneers of the field of biosemiotics6 which studies communication within and between living organisms in terms of signs and codes conveying distinct interpretable meanings—a synthesis of biology, philosophy and linguistics.

Biosemiotics emphasizes that an organism doesn’t respond indiscriminately to the physical world, since only certain signals from its environment have biological significance for it. For each species, there is a range of wavelengths of light, frequencies of sound, odors and tastes that are relevant and meaningful to it and that it can interpret and respond to. This was a major theoretical innovation in how it is possible to build a scientific understanding of life from a very general, meaning-making, perspective.

In parallel to the emergence of general philosophical and physical theories of living matter, several new disciplines within biology started to heavily borrow mathematical methods to describe and solve biological problems. These are mathematical biophysics, biometrics, population ecology and population genetics.

In 1938, a Russian émigré, Nicholas Rashevsky published his major work, Mathematical Biophysics: Physicomathematical Foundations of Biology, at the University of Chicago.7 His goal was to lay out “a mathematical biology which would stand in the same relation to experimental biology as mathematical physics stands to experimental physics.” Rashevsky advocated for

… a certain autonomy of the mathematical natural sciences, which must develop in agreement with the results of experimental research but should not be made mere handmaidens of the experimentalist. A theoretical problem may have an interest of its own, and should not be tabooed only because at present it does not appear applicable to a definite experiment. The history of physics shows how frequently such ‘purely theoretical’ developments led, a few decades later, to the most astonishing practical results.

This way, he could distinguish his new field from previous attempts to build non-quantitative, narrative theoretical biology.

Rashevsky came up with one of the first models of the neuron that was able to recapitulate the all-or-none character of neuronal firing known from experiments. He also noted that networks of these neurons could produce complex behavior and even serve as a model of the brain. However, a lot of his other research endeavors were ultimately highly abstract and formal and had little contact with experiment, which earned him mixed reactions or outright rejection from contemporary experimental physiologists.8

But perhaps Rashevsky’s longest-lasting contribution to theoretical biology was in fact administrative: he founded the first journal of mathematical biology and the first PhD granting program in mathematical biology at the University of Chicago. Established in 1939, The Bulletin of Mathematical Biophysics9 is still an active publication.

One of the most remarkable papers published in this journal in 1943 was first to develop the logical theory of neuronal activity. Its authors, Warren McCulloch and Walter Pitts, showed that networks of simple neurons with all-or-none behavior could be treated as logical circuits that could carry out complex computations. The paper laid the foundations of computational neuroscience, another branch of theoretical biology. Later, John von Neumann used the McCulloch-Pitts model as the basis for the logical design of digital computers.

Rashevsky’s mathematical biophysics was mostly concerned with microscopic biological processes within living organisms, such as cell division and nerve conduction. In contrast, the Polish-American physical chemist Alfred J. Lotka focused on macroscopic population-level phenomena and developed his own version of a biological “theory of everything.” His Elements of Physical Biologyfrom 1924 was “almost cosmological in scope,” seeking to derive general laws of chemical dynamics of structured systems and the interactions between living organisms in terms of energy and matter. Lotka saw evolution as an unfolding, irreversible transformation of the universe, and life itself as a large-scale chemical process.

Today, Lotka is mostly remembered for a set of differential equations describing the population dynamics of two interacting species, a predator and its prey.10These equations were also independently derived by the Italian mathematician and physicist Vito Volterra. The Lotka-Volterra equations have since been used as a staple mathematical tool in population ecology. Though based on idealizations and simplifying assumptions,11 these equations can nonetheless reflect observed fluctuations in the population dynamics of real-world animal species and even beyond biology, such as in economic scenarios.

Both Lotka and Volterra were motivated to understand population dynamics in agriculture and fisheries. Despite the differences of their approach,12 they pursued the same overarching goal, which was

to show that theoretical, mathematical approaches had a place in biology … that theory could guide experiment and research, and that it was not worth waiting until all the facts were in before engaging in speculation with the help of mathematical models.

But perhaps the biggest success story in applying mathematical methods in theoretical biology was in population genetics, established as a discipline by J. B. S. Haldane, Ronald A. Fisher and Sewall Wright. Together, the three men developed the modern synthesis of Darwinian evolution and Mendelian genetics, reconciling natural selection with the laws of heredity.13

J. B. S. Haldane, aside from population genetics, was also the first to realize the theoretical possibility of and experimentally demonstrate genetic linkage in mammals. He proposed the now famous “primordial soup” hypothesis of the origins of life.14 He even authored the modern formulation of the Michaelis-Menten equations of enzyme kinetics.

Haldane was also first to note a link between malaria and the prevalence of genetic blood disorders in tropical regions. In 1948, he hypothesized that certain genetic diseases, when inherited in the heterozygous state, could be protective against malarial infections by deforming red blood cells and making them inhospitable for the malaria parasite. The hypothesis was empirically validated in 1954.

Haldane’s prescient theoretical thinking across many fields of biology goes on to show, once again, that mechanistic theories have preceded many of the major biological discoveries, rather than arriving later as explanations of already known empirical facts.

This is true of all sciences, but it especially struck me while researching the history of theoretical biology that each of its new branches invented its own language to make sense of biological phenomena: biosemiotics speaks in terms of biological signs and meanings, while the abstractions of population genetics are at the level of allele frequencies and fitness landscapes. Cybernetics, too, introduced an idiosyncratic vocabulary to describe organism-level regulation in the most general terms, which also apply to systems beyond biology.

The roots of cybernetic ideas can be traced to Walter Cannon who coined the term “homeostasis” in 1926,15 referring to the ability of an organism to maintain its healthy physiological state in the face of environmental perturbations. In the language of cybernetics, homeostatic regulation can be expressed as a problem of feedback—taking into account the information about the difference between the current and the target state of the organism and using it to move closer to the desired state.

In 1943, the father of cybernetics, Norbert Wiener,16 coauthored a landmark paper with the physiologist Arturo Rosenblueth and engineer Julian Bigelow, titled Behavior, Purpose and Teleology. They argued that both machines and organisms could be studied from a behaviorist perspective, using the same functional concepts of feedback, control, and communication. This means analyzing the inputs (changes in the environment that affect the system) and outputs (changes exerted by the system in the outside world) while disregarding internal processes within the system. Purposeful, goal-directed behavior of organisms can thus be explained in terms of deviations from a target state that are fed back into the organism to reduce those deviations, creating a circular causal relationship.

The field of cybernetics was conceived as inherently interdisciplinary, drawing from physics and engineering as much as from biology, and later sociology and economics. In a sense, cybernetic principles are substrate-independent, since the same abstract feedback relations could characterize a thermostat, an animal, or a human society.

Cybernetics gave birth to new theories and whole fields, including the famous “good regulator theorem,”17 perceptual control theory, modern theories of self-organization, and complexity science18—all direct or indirect contributions to theoretical biology.

Another intellectual descendant of cybernetics is systems biology, though it also traces its origins to several other theoretical forebears, including Ludwig von Bertalanffy’s General System Theory. GST argued that biological phenomena are to be understood in terms of organization and relations among parts rather than through the properties of isolated components. This principle later grew into the “central dogma” of systems biology: that “it is system dynamics that gives rise to the functioning and function of cells.”

In 1968, Serbian systems engineering scientist Mihajlo Mesarović published an essay, Systems Theory and Biology—View of a Theoretician,19 discussing the application of engineering techniques to biological problems and a search for “general biological laws which govern the behavior and evolution of living matter in a way analogous to the relation of the physical laws and non-living matter.” This is reminiscent of the earlier work of Nicholas Rashevsky, but this time around Mesarović’s proposal launched a new research program that became the modern discipline of systems biology.

A parallel stream of ideas that fed into systems biology, also starting in the late 1960s, was in the mathematical modeling of biochemical reactions and of gene regulatory networks.20

Finally, the genomic revolution of the 1990s and early 2000s transformed systems biology into an empirical science, by making systemic measurements of biomolecules possible. The new “omics” fields like genomics, transcriptomics and proteomics, followed by single-cell methods, generated enormous amounts of structural and functional molecular data at previously unattainable scales.

Taken together, these fields are sometimes referred to as “pragmatic systems biology”, in contrast to “systems-theoretic biology,” which is concerned with general principles of biological organization. Another way to articulate this distinction is “data without models” versus “models without data,” pointing at the disconnect between the data deluge and the theoretical frameworks that could make sense of it.

In contrast to the concise mechanistic explanatory theories of earlier days, there’s now a rich tradition of computational modeling that aims to predict the behavior of biological systems based on large datasets of experimentally measured microscopic parameters. But it has remained challenging for these modeling methods to discern simpler patterns, if they exist, within the data, that would amount to a theoretical understanding of biological systems.

This state of affairs prompted me to ask several further questions: what does, indeed, theoretical understanding in biology mean, given our access to increasingly comprehensive quantitative characterization of biological systems? Is it a faithful reconstruction of all their component interactions, or an account of the causal mechanisms producing their collective behavior, or a more general set of organizational principles from which such mechanisms could be derived? Or is simply being able to accurately predict these behaviors enough? How different is theory building in contemporary biology compared to physics? These questions will reappear later, but it is worth pondering what the more recent introduction of artificial intelligence models adds to theoretical biology.

The expanding use of AI models in biology sharpens the separation between modeling prediction and theoretical explanation even further. AlphaFold2 demonstrated the ability of foundational models to predict protein structures with high accuracy without relying on a mechanistic theory of protein folding. Similar approaches are now being developed for genomes, gene expression patterns, and single-cell multi-omics. Increasingly complex biological behaviors seem to yield to prediction from learned statistical representations rather than from theories constructed explicitly in terms of mechanisms or general principles. Still, sometimes sophisticated foundation models fail to outperform far simpler statistical models. The models are surely learning something about the biological systems they are trained on, but how much that constitutes theoretical understanding and how it can be made intelligible to human scientists remains an open question.

Epistemic Cultures

The history of theoretical biology has abundantly shown that theory building in biology has been at least as fertile and consequential in the process of discovery as in physics. Yet, the institutional standing of theoretical biology has arguably lagged behind that of theoretical physics. This was noted already decades ago by the British developmental biologist Conrad Hal Waddington, who wrote in a 1968 paper titled Towards a Theoretical Biology:

Theoretical physics is a well recognized discipline, and there are departments and professorships devoted to the subject in many universities. Moreover, it is widely accepted that theories of the nature of the physical universe have profound consequences for problems of general philosophy. In contrast to this situation, theoretical biology can hardly be said to exist as an academic discipline. There is even little agreement as to what topics it should deal with, or in what manner it should proceed.

I find it remarkable that this was said at a time when theoretical biology had already developed multiple productive branches of inquiry, from population genetics to computational neuroscience. But its institutional presence was still scattered within primarily experimental labs, and rarely manifested itself as independent standalone institutions, such as Nicholas Rashevsky’s short-lived mathematical biophysics program at the University of Chicago.21

Waddington himself was a member of the Theoretical Biology Club at Cambridge University (also known as “Biotheoretical Gathering”) that flourished in the 1930s.22 The Club, founded by embryologist Joseph Needham and philosopher of science Joseph Woodger, was sympathetic to an organicist philosophy that stood in contrast to the reductionism prevalent in natural sciences at the time. They followed a “third way” between vitalism and reductionism, focusing on morphogenesis and hierarchical levels of biological organization. Arguably, epigenetics as a field also traces its origins to these gatherings. However, it is telling that the Club disbanded when the Rockefeller Foundation refused to fund their research (the same grantmaking organization that can be credited with starting the field of molecular biology and that was generally invested in the “technology transfer” from physical sciences into biology).23

Purely theoretical work in biology has generally been harder to secure funding for than experimental research. But theoretical biology has historically spanned a broad territory, from what we now refer to as a philosophy of biology all the way to mathematical modeling of biological systems.

Even though theoretical biology as such does not carry the lofty connotations of theoretical physics, its successful instantiations have individuated into distinct fields that go hand in hand with experimental work, like population genetics, population ecology, evolutionary game theory, computational neuroscience, systems biology, and bioinformatics, often housed at dedicated university departments. So the scattered presence of theoretical biology in research organizations is a consequence of its development in tight connection with experimental work, often by the same scientists. And despite being fractured in this way, theoretical biology is in fact a flourishing multitude of fields of research.

Theoretical physics, on the other hand, has retained its distinct identity, relative independence, and institutional prominence for many decades.

Now, a useful distinction can also be made between theoretical biology, considered as a whole, and theoretical physics, in that they make up different epistemic cultures.24 These cultures are defined by the entire process of discovery and knowledge generation in each field of inquiry, including the nature of that knowledge and its representation.25

Arguably, biology faces a greater difficulty in compressing its empirical data into theories compared to physics. There’s only one kind of electron, and what applies to one, applies to all, making them fundamentally interchangeable. But biological cells and organisms are products of individual developmental and evolutionary histories and are interchangeable only up to a limit.

Sweeping generalizations in biology are harder to justify, and almost every rule has a known exception. Of course, unique biological entities can in principle be handled statistically, and cells can usually be categorized by cell type, and organisms by their species. But biological variation can itself be irreducible and causally meaningful, rather than behave merely as noise around an ideal theorized parameter value. This represents a major theoretical challenge in biology.

Thus there is a sense in which theories in physics and in biology might point to different kinds of epistemic constructs. In physics, to explain a phenomenon theoretically normally means to subsume it under a natural law—a robust and general statement about the (inanimate) world. In contrast, theoretical explanations in biology usually take the shape of a descriptive mechanism and do not appeal to universal laws. Some doubt that natural laws are possible in biology at all, as biological phenomena are historically contingent, being the products of evolution (the evolutionary contingency thesis).26 Though to this one might retort, aren’t laws of physics and chemistry themselves cosmologically contingent, the way biological laws and mechanisms are evolutionarily so?

Philosophers of science have long pondered whether biological phenomena can be fully reduced to physical ones. Waddington wrote that no conceptualization of a living system is complete “unless it includes at least four importantly different time scales, those of metabolism, development, heredity and evolution.” Mendelian genetics comes perhaps closest to a law in biology, being a mathematically mature discipline. However, fundamental exceptions to Mendelian laws are well-documented. Another theory that imposes a kind of general law-like logic onto the living world is, of course, the evolutionary theory and the principle of natural selection.

But most research in biology is a quest for specific context-dependent mechanisms rather than all-encompassing, unifying laws. Such mechanisms describe a biological process or a structure with defined functions and how its components operate and relate to each other, such as the Krebs cycle or physiological homeostatic feedback loops, or gene regulation circuits. Many biological mechanisms are expressed as verbal descriptions or diagrams, rarely accompanied by equations or other kinds of mathematical formalism, unlike most physical theories.

There is also, in biology, the problem of parameters. From a physicist’s point of view, when trying to model a biological system, the number of parameters that needs to be accounted for can be daunting, whether they are genetic or biochemical and subject to change on evolutionary or physiological time scales. Such a state of affairs is rather “unsettling for a theoretical physicist,” as William Bialek of Princeton University notes:

Our most complete theories of the natural world certainly have parameters, but there is a sense that if we are focused too much on these parameters then we are doing something wrong. If parameters proliferate, we take this as a sign that we are missing some additional level of unification that could relate these many parameters to one another; if our qualitative explanation of phenomena hinges on precise quantitative adjustment of parameters, then we search for the hidden dynamics that could make this apparent fine tuning happen more naturally.27

Indeed, in modern physics, some of the most beautiful fundamental theories are nearly parameter-free. They are effective not because nature is itself simple but because, as it turns out, most of its complexity is causally irrelevant and can be done away with for the purposes of both theoretical understanding and prediction.

This is illustrated well by the renormalization group, a theory that relates changes happening in physical systems at different length or energy scales. If we zoom out to larger and larger scales, the microscopic details of the smaller scales largely cease to matter, and it is possible to make universal, coarse-grained predictions about a system’s macrostates, free from its microscopic parameters. Only a few of them, such as critical temperature or proton mass, contribute to macroscopic or observable behaviors, such as phase transitions, while the remaining parameters can be safely ignored.

The renormalization group thus provides a powerful theoretical compression by showing that enormous amounts of microscopic detail are irrelevant at larger scales. This also means that completely different microscopic systems can in principle share the same large-scale behavior. For example, systems made of very different materials can display identical scaling behavior near a critical point provided they belong to the same “universality class,” whether it is a ferromagnet near its Curie temperature or a liquid near its boiling temperature.

In a way, the renormalization group resembles the argument made by American physicist Philip Anderson in his famous 1972 article More is Different: that higher levels of organization have their own emergent concepts and laws that don’t automatically follow from those at lower levels. Biology, with its nested levels of organization from molecules to cells to organisms and populations, is a prime domain in which Anderson’s general argument holds. But unlike in physics, it is not clear whether an equivalent of the renormalization group applies in the living world.

Is it possible to reconcile the elegance and parameter parsimony of theoretical physics with the explosion of parameters in realistic models of biological systems? Bialek considers several possibilities:

First, it could be that biological systems are really so complex that they must be described by models with an irreducibly large number of parameters, each contributing to their behavior. This is perhaps the most pessimistic view if one hopes to come up with effective theories that advance our understanding of biological phenomena, and not just our ability to predict their behavior with black-box models.28

Alternatively, it could be that the behavior of a biological system is largely robust to variations in parameter values, rendering accounting for most parameters inconsequential and their exact measurement unnecessary.

A third possibility is that evolutionary forces have driven biological systems toward a few precise, adaptive parameter values, such as protein binding specificity or beak shape. Such evolutionary optimization can even reach limits permitted by physical laws, as seen in the sensitivity of the visual system to single photons, and in bacterial navigation of chemical gradients that can be sensitive to single molecules detected at the cell membrane.

In this case, if it is possible to identify the selection principle that brought about such evolutionary fine tuning of parameters, we could build parameter-free theories. In other worlds, knowing the selection principle would help us predict the behavior of the system without relying on specific parameters. Or the optimization itself could be used as a principle from which the underlying mechanisms could be deduced,29 not unlike the principle of least action or minimization of free energy in physics.

And, finally, it might be that something like a renormalization group applies to biological systems after all, as much as it does to the inanimate world. This would allow us to understand macroscopic behaviors of an organism largely independent from its parameter-heavy microscopic details.

As it happens, a broad range of phenomena — including biochemical reaction networks, insect flight, the eukaryotic cell cycle, and diffusion and magnetism — are well described by so-called sloppy models, whose properties have interesting implications for Bialek’s suppositions.

In such models, only a few parameter combinations are important to the observable behavior of the model, called “stiff” parameters. The rest, “sloppy” parameters, can be varied over orders of magnitude and still have a negligible impact on the system behavior, and can therefore be discarded for modeling purposes.30 This way, inside a complex or highly parameterized model, it is possible to isolate a simpler and more intelligible effective theory that explains the observed phenomena just as well.

The ubiquity of sloppiness means that even if we are not able to measure most microscopic parameters of a system, it is still possible to make good predictions about its behavior, as long as the model predictions depend on the same stiff parameters as the real-world data.31 The challenge is to identify the most relevant stiff parameters.32 In other words, how can microscopic components of a system be compressed into a few effective degrees of freedom?

This is exactly what the renormalization group in physics does, taking advantage of the scale invariance and symmetries within physical systems. But in the case of biological systems, such as biochemical or gene regulatory networks, this kind of simplification is challenging due to their inhomogeneity and lack of symmetries.33 This brings us back to evolutionary contingency and the irreducible variability of biological systems at all levels of organization. Sloppiness still applies to them but it does not seem to be enacted through an equivalent of a renormalization group that makes physical models so elegant.

And so, it appears that high-dimensional models often possess low-dimensional predictive structure. What is becoming clear through recent research is that the underlying reasons for such compressibility may be system-dependent. Low-dimensionality can arise because of functional robustness—the insensitivity of the system behavior to parameter variation over a broad range of parameter values. It can also emerge through the independence of specifically higher-level behavior from lower-level parameters, as in coarse-graining of the renormalization group. Low-dimensionality can even be a result of evolutionary optimization.

William Bialek has most recently been pursuing the optimization program as a path toward building elegant theories for biological systems, and it turns out that his program and the sloppiness research have converged. Bialek’s group discovered that even the landscape of optimization itself is characterized by sloppiness:

If [evolutionary] optima are small, sharply defined regions in a rugged terrain, then plausible dynamics for the exploration of high dimensional parameter spaces are unlikely to find the optimum; similarly, even small variations in parameters would drive the system far from optimality … Instead, the dependence of functional performance on the underlying parameters is very gentle or ‘soft,’ so that at least some combinations of parameters can vary substantially with very little effect.

Evolution may constrain biological systems because some functional performance measure has been selected (such as kinetic rate of enzymatic reactions or variability of a birdsong). But sloppiness means that this constraint need not determine all microscopic parameters uniquely. And so, a few combinations of parameters that matter for the biological function can be constrained while leaving a large number of other combinations free to vary. Because of this, evolutionary optimization and variability can coexist.

Relatedly, sloppiness has been associated with the remarkable robustness of biological systems to environmental perturbations. Thanks to the vastness of the sloppy parameter space, scores of parameters might not affect the model’s behavior, and this is thought to parallel the function and behavior of the biological system itself. For example, the circadian rhythm in cyanobacteria maintains a 24-hour period over a wide range of temperatures through the phosphorylation dynamics of three interacting proteins, despite the known effect of temperature on the kinetic rates of biochemical reactions. In cases like this, evolution may have to tune only a few stiff parameters to achieve the desired behavior at any temperature within the organism’s reaction norm.

Some accounts generalize even further and connect sloppiness with the comprehensibility of the world itself. Why indeed is science such a successful human enterprise? Whence comes “the unreasonable effectiveness of mathematics in natural sciences”? Is there a selection effect at work, in that we choose to study only those subjects where our brains can in fact discern patterns—low-dimensional representations—and describe them through effective theories? Or are there fundamental external reasons, such as laws of cosmology and evolution, that channel observed behaviors into patterns comprehensible to us?

We do not have ultimate answers to these questions. Yet sloppiness may provide a new way to frame and explore them rigorously. The potential of sloppiness for a unification of physics and biology is beyond being a mere curiosity and deserves attention of both epistemic cultures.

Ulkar Aghayeva is a science writer and columnist at Asimov Press. She also writes at the science history blog Measure for Measure and music blog The Bass Line.

Cite: Aghayeva, U. “On the Role of Theory in Biology.” Asimov Press (2026). DOI: 10.62211/27ry-34qp

Header image by Ella Watkins-Dulaney.

During my interview with CEO Erika DeBenedictis, I asked about the myriad bottlenecks to growing life on Mars. The planet has tons of water, for example, but only as ice. (So much water, in fact, that if you melted all of it, it would fill an ocean 100 feet deep across the entire planet!)

Similarly, we talked about the dirt. Martian regolith has salts, heavy metals, and perchlorates that are toxic to many organisms. Some microbes on Earth are able to break down these toxins, though, and get energy from doing it. Pioneer is engineering these organisms to make them better at eating toxins while adding nutrients back into the soil, which could possibly support plant life in the future.

My main takeaway from this interview is that biotechnology can be used to do so many things; not only for medicine, but also about bioremediation and protecting animal species and, yes, seeding life on other planets. I hope you enjoy the interview.

Watch on YouTube or listen on Apple Podcasts or Spotify.

Timestamps

0:00 A first microbe engineered for Mars
5:03 Simulating Martian chemistry
8:27 The five "pioneer species"
17:20 Water, warming, and nitrogen
26:39 Radiation, growth rates, and poor measurements
35:49 Starship
41:05 Why create a second biosphere?

Further Readings

• How to make a microbe for Mars, one step at a time. Pioneer Labs, 2024.

• A biologist’s guide to Mars dirt. Pioneer Labs, 2024.

• Our first target microbe for Mars, and how we chose it. Pioneer Labs, 2025.

• An introduction to Mars terraforming. Pioneer Labs, 2025.

• What’s the best way to make a Mars microbe?. Pioneer Labs, 2025.

• BioBloom: Bringing scanning mutagenesis to whole genomes. Pioneer Labs, 2026.

• No, Mars doesn’t need a magnetic field. Pioneer Labs, 2026.

• Bioplastic production from simulated 100% in situ Mars resources. bioRxiv preprint, 2026.

• Defined Mars Media (DMM), a chemically defined simulant of the soluble macro- and micro-nutrients in Mars regolith for use in biological research. bioRxiv preprint, 2026.

• BioBloom, a method for barcoded saturation mutagenesis of an entire bacterial genome. bioRxiv preprint, 2026.

• An introduction to Mars terraforming, 2025 workshop summary. arXiv, 2025.

• The case for Mars terraforming research. Nature Astronomy, 2025.

• Making Mars habitable. Nature, 1991.

• Technological requirements for terraforming Mars. AIAA, 1993.

• Inventory of CO₂ available for terraforming Mars. Nature Astronomy, 2018.

• How to get to Earth from Mars: Solving the hard part first. Casey Handmer, 2016.

• Atmospheres and terraforming. Casey Handmer, 2018.

Transcript

Erika: That is the ultimate win condition. If we succeed maximally at making the five organisms for Mars, we’ll have a complete set of organisms that you could just crash land onto Mars and they’ll just grow themselves and make the planet nice.

Niko: So today’s guest is Erika DeBenedictis. She’s co-founder and CEO of Pioneer Labs, a nonprofit research group that is engineering microbes to terraform Mars. And what we’re going to talk about today is why we need a second biosphere beyond Earth. So Erika, welcome.

Erika: Thank you.

Niko: And we’re here, of course, filming in the lab at Pioneer because you have an announcement.

Erika: Yeah, we’ve engineered the first microbe for Mars. It’s useful. You can take it to Mars and use it to help build a city. It’s safe. It can’t be invasive. And most excitingly, it’s a first step toward terraforming. It actually grows using the nutrients in Mars dirt. You don’t have to cart stuff from Earth to grow it. And there’s a lot of Mars dirt on Mars. And that’s what you need to actually make a biosphere at scale.

Niko: And you have this here, right?

Erika: I do. Yeah. So this is one of the microbes we engineered. So this is a C. necator that we then evolved to be better at using the nutrients in Martian regolith. And we’re using it to produce PHB bioplastics.

So bioplastics are a class of materials that we often interact with in everyday life. They make up everything from 3D printing filament to compostable cups. Many different materials with a lot of good properties fall into this class. And we can manufacture things in that class of material with this organism.

Niko: So you have some of the sample here, too, of what these microbes are actually making, right? So you’re growing them in a simulated Martian liquid, which has all the toxins and all the kind of nutrients we would expect to find on Mars. And then they make this. And I’m curious, why did you want them to make this? Why is this useful?

Erika: Yeah. So this is an example of raw bioplastic right after it comes out of the organism, but before you form it into its final form. I think the first application we’re excited about is building astronaut houses with it. So you can make plastic liner that would contain the pressure that you need to breathe inside, and you can also use it to make panels to hold up your house.

And bioplastic is cool because microbes on Earth often make bioplastics for the same reason that we put on weight if we eat too many donuts. It’s a waste product and it’s a result of having too much carbon that they have to get rid of.

So unlike many other bioproducts that it’s an enormous amount of energy for the organism to make it and if it can figure out how to evolve away from making it or something, unlike those bioproducts, making bioplastic is the thing they do by default when they’re stressed, which is kind of how they’ll be on Mars.

Niko: And so you chose something that they could make without evolving away or breaking the function.

Erika: Exactly. And indeed, if we take these organisms to Mars and we continue to grow them, they’ll evolve to get better, not worse. And that’s important because you don’t want to take something to Mars and rely on the bioprocess and then have it break.

Niko: And this is maybe speculative, but let’s say you send microbes to Mars and they’re making bioplastics on the surface. How do you think about shaping this now into an actual dome that you can now populate with other things?

Erika: Yeah. And this is why I say this is right what comes out of the organism. This is just the fat on their hips, right? It is straight up extracted, and you get this molecule that you would then need to process with equipment that you would have to bring from Earth to form it into useful actual materials.

And the equipment required to do this processing is something we’re trying to minimize. So we’ve tried to make a first organism that is useful and that would require some processing equipment, but hopefully not too much. And we’ve actually completely enumerated how many boxes of equipment would you need to bring to Mars to do all of the processing and 3D printing and stuff like that to use this material. And we’ve actually gone through the full exercise of full TEA (techno-economic analysis) of how much weight do I need to bring to Mars to actually use this stuff?

Niko: So your vision for terraforming is microbes plus robots working together where the microbes are doing the biochemistry needed to create the material. One thing I’m interested in is why C. necator. It is kind of a strange microbe.

Erika: C. necator. Yeah.

Niko: I’ve seen papers with that being used. But again, you’re actually trying to get this thing to work in a very weird medium.

Erika: That’s right.

Niko: And so, how did you actually do the experiment of creating this? What is in the medium? Were you just putting a bunch of microbes in it and then doing evolution to figure out what could even survive? And then you have to engineer that to make bioplastic.

Erika: What’s kind of funny about this announcement, we’re really proud that we have made the first microbe for Mars. But the making of the microbe was the easy part. The hard part was figuring out how to choose which microbe and how to correctly simulate Mars so that we could make it better. And once we got that figured out, doing the evolution for many months, that was easy, right?

Niko: So how do you simulate Mars?

Erika: Yeah, so we’ve never brought Mars dirt back to Earth. Instead, what we have is we have beautiful rover data from multiple rovers that have really nicely engineered hardware for scooping up a little Mars dirt on Mars, adding water and then taking chemical measurements just on site. And so we have a good idea, actually, and really nice measurements of what’s in the dirt on Mars in multiple locations.

And from that, what we did was we created a defined media that accurately simulates what happens when you add Mars dirt and water and what the chemistry is of that mixture.

And this is a little bit different than how scientists have used what they call regolith simulants or Mars dirt simulants before. So people have had Mars dirt simulants for a long time, but they’re mostly designed to help you test your rover tires. So people find rocks on Earth that are roughly the same bulk mineral properties. They grind them up to the right dust size and then they’ll send it to you in a little sample and you can test your rover tires. But they’re...

Niko: You have some of that dirt here, right?

Erika: I do. So these are some samples of regolith simulant from Space Resources Technology. This is one that’s designed to mimic the dirt at the Curiosity rover landing site. And this is not under armed guard because this is not from Mars. This is just Earth rocks ground up. And this does a good job at allowing you to start doing tests mostly of mechanical stuff.

But this is not a sufficiently high fidelity simulation of the chemistry and what would matter for biology, especially both the nutrients that are in Mars dirt and the toxins. And so we created a defined media recipe that accurately simulates those things.

And that was what we used to take a bunch of terrestrial microbes, including C. necator. Yeah, it’s not E. coli. It’s not as famous as E. coli, but it’s often used in terrestrial biomanufacturing in part because, as organisms go, it’s kind of a generalist. It’s reasonably good at accepting unusual feedstocks and making use of them, which is exactly the task here.

Niko: So just to clarify, you have a microbe that uses Martian nutrients to make a material because in your mind, the first step to terraforming Mars is not to terraform the whole planet. It’s to make a mini greenhouse that we can fill with oxygen and then put microbes in there.

Walk me through the actual steps. You mentioned this is the first microbe for Mars. I think me naively, six months ago, I would have thought, oh, it’s just going to be one microbe that we send to Mars and it’s going to terraform the planet and grow and do everything. But your roadmap at Pioneer is actually a series of microbes that do different things.

Erika: Absolutely. So think about Earth. We don’t just have one microbe, right? Nature is really good at this task of taking a harsh environment that sucks and is sterile — think post forest fire or volcano, something like that — and transforming it into a complex ecosystem that has lots of different organisms in it, including not just one microbe, but thousands of microbes, even just in a gram of dirt on Earth.

And so Pioneer Labs, the name comes from this idea of creating the pioneer species that allow that ecological succession to take place where you begin with this sterile environment that’s mostly rock or, in the case of Mars, rock dust. That’s what the dirt is on Mars. It’s just sterile rock dust. And then you have to convert that material step by step into something that can support a wider variety of life, one pioneer species at a time.

So a pioneer species is something that comes in and can grow in a harsh environment and makes it easier for the organisms that come after it by adding nutrients to the soil or breaking up big pieces of rock into little pieces of gravel that then plants can grow in. And so our goal as an organization is to figure out what is the smallest number of these pioneer species we actually need so that we can go to Mars and step by step make it more habitable.

Niko: OK, so this is the first microbe for Mars. I’m curious, what is the full process that you envision? So you’re going to send this thing to Mars with robots and other people will build the robots probably, right? You’re just kind of providing the biological option. What is next? What is microbe two and three and four and five? And when do we have plants growing on Mars?

Erika: Yeah, so the goal is to get to a scalable way to make arable land where you could grow plants that make oxygen and food.

Working backward, plants need a warm environment that has enough liquid water to grow. And so in the short term, you need to make really, really simple enclosures. Could be as simple as basically a plastic bag, something that’s transparent so it lets light through, but it holds in just a little bit of that pressure so that your water doesn’t all evaporate immediately.

And we think we could make such really simple greenhouses with the microbe that makes bioplastic, which bioplastic has really good tensile strength so it can hold in the pressure, combined with a second organism that can make a transparent material. And there’s a bunch of options for that and that’s what we’re working on now, is what is the correct choice for making big sheets of transparent material on Mars that, again, requires very little processing. The more robots, the more expensive it’ll be. And ideally can be done super scalably and entirely, again, using Mars dirt, water and air. You don’t want to ship stuff from Earth.

So the plants, they need an environment to grow in and then they need good dirt. So the dirt on Mars has a toxin in it called perchlorate. And fortunately, there’s a little bit of perchlorate on Earth. It happens in really small quantities in high altitude deserts. It is produced when UV radiation hits sand. And this is why Mars has a ton of it and we have only a little bit of it.

And there are some microbes on Earth that can actually eat it. They degrade it. They get a little bit of energy out of it. And so we are now also working on microbes that you can add to Mars mud and they’ll just eat the toxin and add nutrients to the soil, add organic matter and nitrogen and so on.

And so altogether, tensile strength materials, transparent materials, get rid of the toxin and make the dirt nutritious. Those are the first four. And then the fifth is a photosynthetic thing that can grow in that environment and create oxygen and food.

Niko: Why not just ship a tarp, a clear... You keep saying it’s like a plastic bag. That’s not that heavy. Why don’t we just send the plastic bag, put in the perchlorate degrading microbe, put in a microbe that makes oxygen. Boom, terraform dome, right? Why do we need all five microbes?

Erika: We may not. So I think the fastest way to dome off a local area probably is to manufacture something on Earth and ship it there. I think the question of scale is always the thing that the biology is really good at.

So yes, totally. You might be able to manufacture a structure and send it to Mars and live in it. If you want to manufacture a bigger structure, then suddenly it starts becoming an attractive option to use biology to convert local materials into that larger structure.

Niko: You could also have a rover that’s setting up domes.

Erika: Absolutely. And as a concrete example, so the first microbe we engineered, we’ve done the full TEA to figure out: is this actually useful? Is it better to send that microbe to Mars and use that microbe to manufacture houses out of local materials? And is that better than just sending the materials from Earth and not bothering with any of that?

And the answer is yes, it is better. But the crossover point takes a couple years. So you have to send a bunch of infrastructure to Mars and it takes several years of that infrastructure cranking out product for it to justify its mass.

But that is fundamentally the thing, which is: where is the opportunity to use the fact that biology can convert awful toxic dirt on site into houses? When does that become useful? And it is useful in this scenario, exactly this scenario where Mars is far away, shipping stuff is expensive. And if we’re setting up shop, we will massively benefit from being able to manufacture things on site without being worried that our shipment of bioplastic from Earth is going to blow up or get delayed.

Niko: Do you think it would ever be possible to find the right combination of microbes? No equipment, no robots, just microbes and just put them on a light sail and send them to Mars and have life on Mars? Would they survive?

Erika: That is the ultimate win condition. If we succeed maximally at making the five organisms for Mars, we’ll have a complete set of organisms that you could just crash land onto Mars and they’ll just grow themselves and make the planet nice.

Niko: But they need robotics. They need bioreactors.

Erika: I think this is the question. I mentioned we’re trying to strip out as many of these human processes as possible, that each abiotic process is itself a bottleneck that requires equipment that you would send to Mars, equipment that could break, equipment that if you want to do that process over the whole planet, you have to send many copies of that equipment. And so this is the outstanding question, is how many of those robots can we get rid of?

I think more realistically, the path that seems eminently doable from an engineering perspective, but more expensive and requiring some robotics, does involve maintaining some of these processes that have these abiotic choke points, which in some ways is also a feature, right? Because I think it is its own different and scary prospect to have a fully biological system that doesn’t need human intervention and that can self propagate indefinitely.

And so maybe we like it that we need a little bit of robotics. We’re still important. We still have some control and we’ll see how the engineering pans out.

Niko: And you’re mentioning things like liquid water. You’re mentioning UV radiation, intense radiation, toxins in the dirt. And so I want to talk about some of the bottlenecks of growing life on Mars, because I think that, me as an engineer, that always helps me understand, what are we actually engineering for?

So I guess the first question here is how do we get water on Mars? And how do you ensure that your payload lands near where that water is? Right. If we’re not shipping water because it’s so heavy, how are you solving that problem?

Erika: Yeah. So I think it’s a super common misconception that Mars doesn’t have water. So 3.7 billion years ago, Mars and Earth were little twinsies. So both freshly formed planets both had liquid water. Following that, their paths diverged. So Mars was a little bit smaller. As a result, it cooled faster. Its magnetic field stopped, its core stopped convecting. And so its magnetic field turned off.

Something else must have also happened because we can’t figure out where Mars’s atmosphere went. We can’t fully account for it. But one way or another, Mars lost a lot of its atmosphere. Today it has sub 1% of Earth’s atmospheric pressure, basically all CO2. And it’s really cold because what atmospheres do is they’re little blankets for planets.

And what happened to all of its water is that its water is now ice. So Mars actually has a ton of water. If you melted all of Mars’ water and distributed it evenly over the planet, you would get an ocean 100 feet deep.

And much of that water is permafrost in the northern hemisphere, so if you land kind of anywhere that’s mid northern latitude on Mars, it’s a really thin layer of dirt on top of basically a glacier that is permafrost. And so in that sense, Mars looks like a desert. It looks like the Atacama Desert, but actually it’s kind of more like Alaska. And that means the water is actually really easy to access.

You can use a type of well that people use in Greenland where you start with some water, you heat it up and you just pump it into the ground and use the heat that you can dump into the ground with the water as a way to melt more and then you suck it back up. And so you can actually get water out of Mars super easily.

Niko: So there are also people working on extremely potent greenhouse gases, right? And the idea is that they would take some of this and ship it to Mars and heat the planet by basically capturing more energy.

If you read the documents that Pioneer Labs has put out, which I’ll put in the description, that would take 30 years, people think, to heat the planet by 30 degrees Celsius. So do you just not want to wait 30 years? Doesn’t that seem like an easier thing, to ship greenhouse gases to Mars and heat the planet, wait a few decades and then start shipping microbes?

Erika: One could. Mars is too cold. There’s a bunch of different ways, proposals for how to heat it up. One proposal is this idea that perhaps we could manufacture something that you could sprinkle in the atmosphere that might be little bits of glitter that basically reflect heat radiation back toward the surface.

And yeah, the numbers look shockingly good for making another planet. So in 30 years for the cost of the U.S. interstate highway system, you could heat Mars up to liquid water compatible levels, which is amazing because then we could have another planet.

The downside being you have to convince a government to spend a couple billion dollars a year doing this for 30 years. And if you stop, it goes back to normal. And also you do have to continue to do it indefinitely or else again, the glitter settles out and then it cools back down.

Another option is to use solar sails. Reflect Orbital is a company that is deploying solar sails in orbit that would reflect light onto the ground. And you could use that also to... If Mars is too cold, just add more sunlight energy and thus heat it up. But this approach has a similar order of magnitude math, where you would have to spend a lot of money and just practically speaking, scaling up that manufacturing would take quite a long time to approach the magnitude of sunlight needed to be reflected onto Mars to heat the whole planet.

And I think both of these are really cool and exciting possibilities that some combination of all of these methods could be used in the long term to heat up Mars, keep it warm and control its climate. But practically speaking, the thing that is so much easier to implement is something where you get an immediate benefit in a local area.

So with the engineered aerosols, you’d have to heat the whole planet for 30 years with a lot of money before any benefit is acquired. In contrast, if you greenhouse off a small area, you can go live in that area. That can be where you have the first human mission to Mars and you can just expand the area.

And so again, all of these techniques could be used. They’re complementary. But I think we keep returning to this fact that regardless of what additional techniques you use to heat the planet, it’s a really good idea to focus your efforts and money on a small local area in which you can get immediate benefit.

Niko: I’m just imagining a little bubble with a drill pumping boiling water down. It’s very strange to think about. And there would be microbes in there.

Erika: So another thing about greenhouses, if you literally just plop down a greenhouse above permafrost, water vapor will be released into the greenhouse. So you don’t even need the drill if you have a really simple greenhouse. All you need to do is heat up a glacier to get water out of it.

Niko: And nitrogen. So Earth is mostly nitrogen, lots of nitrogen in the atmosphere. And organisms need that to build proteins and other things. But Mars, the nitrogen, it’s 3% or something or less. So how do you solve that problem? Is there enough nitrogen on Mars to terraform the planet?

Erika: So actually, when we first started this... Mars was not the only option for a planetary body that maybe you could make a planetary scale biomass on. So we went through all the other planets that you might think of and we computed if you converted all the carbon and hydrogen and oxygen and nitrogen and phosphorus and sulfur that’s on that planet into biomass, how much do you get? And is it a big number? And for Mars, it’s a really big number.

So yeah, there’s a ton of nitrogen on Mars. There is relatively little nitrogen relative to Earth in the atmosphere. So Earth is mostly nitrogen gas in the atmosphere. On Mars, the nitrogen gas comprises a very small fraction of what is already a very thin atmosphere. And where the nitrogen is instead: it’s already in the dirt.

So by Earth standards, Mars dirt is pre-fertilized. It has fixed nitrogen in it already as nitrate. And so in the short term, that’s what you use. You can put organisms in the dirt that do not need to fix nitrogen out of the gas and can use the nutrients that are already there.

Now, the problem is that on Earth we have denitrifying bacteria that take nitrogen and basically breathe it and they release nitrogen back out into the atmosphere. And so on Earth we have a whole nitrogen cycle where some bacteria, they take the nitrogen gas, they make it solid basically. Other bacteria, they take the solid nitrogen, they make it gas again and around and around we go.

And so the question is on Mars, does that loop still close? On one hand, denitrification, so making the gas again would be slower because there’d be more oxygen in the atmosphere because the proposed atmosphere and that sort of proposed composition of the greenhouses is mostly oxygen. Or it’s a much higher percent oxygen. And so that step is slower. Similarly, fixing the nitrogen once again would happen. It would just happen way slower.

And so together do those things close? And that is an outstanding scientific question that remains one of the things that could actually be a deal breaker for terraforming. If somehow those two edges of the cycle don’t balance, that could be a big problem.

Niko: Wow. And so another bottleneck, of course, is... Let’s assume you’ve shipped this stuff there. The second big bottleneck seems to be UV radiation.

So it seems like you’re just trying to evolve microbes kind of naturally where you’re not putting in engineered systems or complex logic. But is there any concern that these microbes would just break because the radiation is so intense or they’re mutating so quickly that they stop producing bioplastics? How can you be sure that the biological function persists?

Erika: In the case of the first microbe, the idea is that you grow it in a controlled, contained environment. You grow it in a stirred heated bioreactor, which doesn’t have UV. So it’s protected from those particular extremes. And what it’s challenged to do is just deal with the really weird chemistry of what the raw materials are that it has access to on Mars. And so in that sense, it’s a good first step. It sidesteps all of these issues of: how do we deal with the temperature and pressure and UV?

But you’re totally right. In the broader sense of if we’re proposing to make a whole biosphere, it’s a big planet. And that means that all of our attempts to prevent or control what evolution might happen will probably not work. We have to expect we’re making a natural system that will evolve. We’re making real stuff for real outdoors. Not for controlled environments where you can sterilize everything and start over.

On the point of UV, Mars has a lot of surface radiation in part because again it has really thin atmosphere. It has so much surface radiation that it would be a problem today for prolonged exposure for a human and that’s because we have great big bodies and so we have many targets to hit. And if a piece of radiation comes in and hits my one cell over here, it can become cancerous. And it’s a whole problem for the whole bit of me.

In contrast, prokaryotes actually are quite radiation tolerant. So they have such small bodies. Essentially, there’s very little target to hit and they divide very quickly. So if you have an E. coli that’s dividing every 20 minutes, it’s kind of no problem if once an hour one of the E. coli gets killed. And so the prokaryotes are kind of not the issue here.

Some of the eukaryotes, like algae, it starts to be something you have to think about. There are some plants that are quite radiation tolerant, others not so much. When you think about the biosphere, it’s, I think, a little bit more of a matter of carefully choosing which organisms to use and making sure that they’re growing fast enough to outrun the radiation rather than spending huge amounts of energy repairing damage.

Niko: Do you have any indication of how fast microbes would be growing under such toxic nutrients, very low quality kind of nutrients, right? My hunch would be that their division rate would be maybe once per day or once every other day.

Erika: To be determined. And I think that’s what we’re trying to get to. So over the past couple years, we’ve tried to push toward again being able to actually simulate Mars. We started with the easy case of can we just simulate just the chemistry of the dirt? We got to get that right.

And now we’re moving into simulating not just the chemical environment, but also the physical environment and the physical reality of what would your experience be being a microbe growing in a pond in a simple greenhouse?

And so we’ve started to do things like we have a bunch of thermal modeling for what is the day night temperature swing and the seasonal temperature swing in a greenhouse on Mars. And so we’re starting to actually get to the level of detail where we can actually put numbers on things like how fast would this organism grow in a greenhouse?

And we also are starting to do things like... We have many little bits and pieces of equipment all around for many different ways to simulate the physical conditions of Mars, the low pressure, weird temperature swings and so on. At one point we were using a food dehydrator that you usually use to make jerky but you can hook it up to low pressure and stuff.

This is just one of our new boxes. You can depressurize this. You can put the whole cube into a freezer and you can shine lights on it in order to mimic the atmospheric pressure and composition, the temperature profile, and the illumination profile in a target environment on Mars.

Niko: And so you’re basing all of this on measurements, right? And again, as an outsider, as a biologist primarily who hasn’t studied Mars, my sense would be that we would have very good measurements on things like the atmosphere of Mars, these physical things, how much sunlight hits Mars, but that the actual chemical measurements of the dirt from these rovers might not actually be a great representation of what is on Mars, right?

It’s a rover in a place on Mars. And based on these very scattered measurements, you are designing this media that the whole mission kind of relies on, right? Will these microbes be able to live in this media and then make bioplastics? How can you be so confident that we have good enough measurements to simulate Mars on Earth?

Erika: This is also the thing that makes me nervous. And that’s why we started with the chemistry, because everyone knows how to make a vacuum chamber. And you’re right, we know what the pressure is. And so as you say, the place where the scary unknown unknowns lurk is in the dirt.

A couple causes for hope. So cause for hope number one, unlike on Earth where dirt on Earth, the composition of it is super dominated by what just died in it. Unlike that, Mars lacks that entire level of dirt variation that is generated by the entire history of things living and dying in a spot.

Niko: But things used to live on Mars and probably they lived on... Well, I guess we don’t know that, right?

Erika: We don’t know that.

Niko: So I’ll cut this out.

Erika: But we have measured this. So people have looked for organics in the dirt on Mars. So we know dinosaurs used to be here because there’s hydrocarbons. So people looked for hydrocarbons on Mars and guess what they found? Teeny, teeny, tiny trace amounts.

And we’re at the point with search for life on Mars where the teeniest, tiniest trace amount was super exciting to everyone, but it’s so low that it’s not going to matter for growth. So we have an upper bound on how much we should care about organics in the dirt. And it’s negligible from a perspective of growing new stuff.

And Mars has global dust storms. And so the top layer of regolith is super homogenous all around the planet. And we can confirm that in part because we have measured it at multiple spots. And that top layer, the fines, is quite homogenous. So that is some cause for hope.

That said, there’s no way to know whether or not there’s an unknown, unknown chemical in Mars dirt that maybe is at super low amounts, but is so toxic, nothing could ever grow. And that we don’t even know to look for it because we’ve never seen it or something. One certainly can imagine some sort of really unusual unknown unknown that would be a deal breaker. And the only way to resolve that existential risk is to go to Mars and try to grow stuff.

Niko: The problem is if you send something to Mars and it doesn’t grow, you have no idea why. It’s because it’s dead. And the problem with biology is we can only study things that are alive. And if it’s dead, it could have been a million different things.

Erika: It’s true. Although with good experimental design, much could be done. So you can send a microbe to Mars. You can bring some Earth LB. You can take a sample of Mars dirt and then you can do a dilution series between the two and see where it craps out. And at least you know at what concentration it dies, right?

Niko: You could try to get it back to Earth.

Erika: That would be ideal. And man, wouldn’t it be nice if we had sample return?

Niko: But you keep mentioning Starship and launching stuff to Mars and the cost.

I guess my question is even if you made all these microbes, which seems possible, it’s totally feasible to make a microbe that recycles perchlorates and you’ve already made progress on microbe number two doing the clear materials and you already have microbe number one. So how do you get on Starship? Is this a private donor is going to come and buy a Starship to launch to Mars? Or how does this...

Erika: I’m so glad you think we’re doing the thing. Yes, we’re doing the thing. So let’s just pause for a moment and just celebrate that you think that I’ve convinced you it’s possible because we’ve actually gotten started, which... Let’s just take a moment.

Niko: I’m not a skeptical audience.

Erika: That is true. Yeah. Look, you’re right. One of the cool things about this is that we have reduced the formerly incredible sci-fi concept of terraforming into an actionable engineering plan that consists of make five things with biology that seems feasible. And that itself is crazy and very cool.

Okay, separately, how do you actually get to Mars? In the past 10 years, our access to space and just Earth’s launch capacity has increased by at least a thousand fold in part because of Starship in particular, which is our new heavy launch vehicle that is kind of reusable and will eventually be actually fully reusable.

It’s now possible to contract a private company to take your payload box and put it on the moon and provide it with power and comms. That’s called a CLPS lander. So they’ve made a modular standard for if you want to send your box to the moon, here are the attachment points that it needs to have and here’s its dimensions. And it’s becoming commoditized to send stuff to the moon.

And the equivalent of CLPS for Mars doesn’t quite exist yet, but it’s coming. And it’s in the proposed congressional budgets to push toward that. Again, Mars is the next step beyond the moon and there remain things, engineering things that need doing. There needs to be some politics to make it possible to commoditize sending stuff to Mars. We need to be able to refuel Starship in orbit, which hasn’t been done yet, but will happen in the next few years.

And overall just instead of being this tremendously bespoke process to get to space where you have to find a government that wants to either make a brand new launch vehicle for your mission or manifest your payload on something they’re already doing for totally other science reasons. It’s becoming actually honest to God a thing that you can go to the marketplace and buy. Ten years ago, this would have been a crazy world to imagine.

Niko: How much does it cost?

Erika: There is not public prices for it yet because it hasn’t happened yet. But the movement is real toward expanding the public sort of reasonably public pricing or sort of understood pricing for the moon to Mars. I would expect that 2031 is going to be the first launch window when a commercial entity actually lands on Mars. And we’re hoping to have a payload on that spacecraft.

Niko: So Pioneer Labs is a nonprofit research organization. You’re many years out from this 2031 mission. You’re not spinning out companies, right? You’re developing a lot of tools internally that could be very useful to companies, but you’re not spinning out for profits that I’ve seen. So how do you actually sustain this? How do you keep going for the next five plus years?

Erika: So Pioneer Labs is a nonprofit, independent nonprofit. It’s in the style of a focused research organization. So the idea of a focused research organization is that there are some problems that are too big for academia or require too much multidisciplinary expertise or scale of resources to solve, but they’re also not sufficiently immediately profitable to be funded by venture capital.

And so what you need is a dedicated basically nonprofit startup that solves the bottleneck and then moves on to the next phase. And so that’s what Pioneer is.

And in general, our goal is to make the microbes and then figure out how to get them to fly, whether that is through a spin-out of our own or by licensing the technology out to others or facilitating it through collaboration. Our job is solving that bottleneck and then giving that technology the actual genuine best chance of being implemented by, in some cases, giving it away, or in other cases, creating for-profit incentives that can drive deployment.

Niko: I want to understand really explicitly your motivation for creating a new biosphere in the universe. Is it simply... So Chris McKay, a pioneer in terraforming Mars research, published many papers on this. He has just said that the goal of terraforming is to enhance the richness and diversity of life in the universe.

One thing I’ve always been curious about is if we just look out in the universe at all the planets and all the stars, the probability that there is life elsewhere, my view has always been that it would be kind of selfish and egotistical to assume that we’re the only life forms in the universe.

And so why do we need to go and create another biosphere? Is it that mapping this probabilistically is not enough? That taking it on faith that there’s life in the universe is not enough? That we need to be sure that there’s another planet? What is your motivation really?

Erika: This is such a crazy thing to say because when I think about it, it seems crazy to me that there would be life somewhere else. There’s just no evidence. There’s no evidence one way or another. You sort of arrive at this, oh, surely there’s life somewhere from a, well, it would be weird because if there isn’t, it’s egotistical.

Niko: They just don’t want to talk to us.

Erika: They just don’t want to talk to us. I think the thing that is fascinating about Mars and why so much of space science, mental energy is directed toward Mars rather than, say, the moon is because it has this weird twin situation going on with Earth.

Mars truly used to be nice. It looks exactly or used to look exactly like a place that really should have been able to support life 3.7 billion years ago. And so it is such an irresistible target for both searching for life and also trying to figure out whether it could be possible to create life that grows there.

And I would say our options are becoming kind of limited for what could we do to put some priors on do we think there’s life somewhere else in the universe? We can go to the ice moons in our solar system and check. They are great candidates. They’re far away. It’s going to take 20 years to just schlep out there to check.

What else are we going to do? Go to the nearest next habitable exoplanet and check. That’s going to take longer than my lifetime, right? So our options are kind of limited for how to gather data about the real universe we live in that might inform whether or not we’re alone.

I am impatient and I don’t know if we’re going to get our act together to go to ice moons, but a thing I could do is see if I could make life that grows on Mars. Because if so, and if it’s not just possible to get anything to grow in a Mars simulation chamber, but actually make a planet scale thing, can you actually make a biosphere that closes all of its loops on another planet that has similar but kind of different base chemistry and base raw materials?

That to me is something that changes my priors on is there life elsewhere in the universe? If it’s possible to actually have a functional biosphere on Mars, my probabilities go up.

Niko: If we actually wanted to increase the diversity and richness of life in the universe, wouldn’t it be better to not terraform Mars with Earth microbes, but just to kind of treat it as a directed evolution experiment where it maximizes the diversity of what things grow there? Is there any concern that just sending five microbes from Earth and seeding the planet with that will just limit its diversity long term, right? Because you’re kind of just duplicating stuff that already happens on Earth.

Erika: Yes, but I’m not sure how to do better, right? So I think we are constrained by life as we know it. So we know of carbon based water based life. That’s not necessarily the only thing there could be. But engineering something different is beyond current science, right? And so I can’t prove it’s impossible, it probably is possible, but we also can’t do it.

What we know from evolution experiments in the lab on Earth is if you grow something for long enough, ecology will evolve. So the long term evolution experiment literally evolved ecology. They were just passaging E. coli for decades and it essentially speciated all on its own. And I would expect that to happen on Mars. Even if we only send five things, they’ll figure out how to speciate on their own very likely.

Niko: The selection pressures are unique.

Erika: The selection pressures are so unique, so unusual and so unlike any combination, any environment we have on Earth. None of them are quite like Mars. Some of them have one overlapping piece, but none of them are quite like it.

Niko: If NASA announced tomorrow that they discovered life elsewhere in the universe, would you still be motivated to work on this?

Erika: Absolutely, yeah. I grew up as a teenager watching Star Trek. And I liked Star Trek and I think it set my philosophy for thinking about the universe in the sense that it’s this really positive vision for what humanity could grow up to be as this species that overcomes our technological adolescence and learns how to use our technology for good both socially and technically and retains the spirit of exploration and curiosity and I like that.

I think in the Star Trek universe they have this concise and satisfying way to do the ethics about life where it’s: if the civilization is pre-warp then you don’t talk to them. If the civilization is post-warp then you get to talk to them. So in this imaginary world where life is plentiful in the universe the ethics of how do you go out and interact with other life forms is sort of obvious.

And I think in a universe that doesn’t have as much life in it, that’s much more dicey because suddenly even microbial life or even the hint of possible life becomes something that we don’t know how to handle because we’re searching so hard for someone else to talk to. And so I really hope there is a life somewhere else.

Niko: No, but wouldn’t that update your priors on the motivation to work on Mars was you can see it. And if we can close the cycles, it increases your priors that there’s life elsewhere in the universe, right? But if there is life elsewhere in the universe, then that’s no longer a motive, right? I think then it becomes about having a backup biosphere and...

Erika: No, it’s not.

Niko: Well, then what would...

Erika: I hate this term, “backup planet.” Oh, my God. People say this. No one should say this. It’s like saying you want a second kid as a backup. That’s not why you want a second kid. You want a second kid because you had a first kid and you want at least two, right? Biospheres are magic and we should have at least two, right? That’s why. Nature good, more nature better. Life interesting, more diverse life better, right?

I would love to understand whether or not there’s more life in the universe, but that’s the scientific motivation. There’s more motivations, too, of just my experience on this planet is one of wanting to go outdoors and smell the flowers and taking so much delight in how curious and weird all these crazy plants and animals and microbes are on this planet.

Complex systems are cool in part because you can’t study them by isolating just one component. They only have emergent properties when you put them together and nature is that way where you can’t just have one tree in a pot. You have to have a forest and forests have cool and exotic and fascinating properties because they’re complex and composed of many parts.

And so I think it’s that complexity, the ability of nature to surprise us and delight us and evolve and have emergent properties. That’s the thing that is cool and intriguing. And I would like for future humans to have more of that rather than less of that.

I hate this progression, this constant progression toward a life that is totally predetermined and entirely based on totally robust and predictable, abiotic processes where we get our calories from some chemical process that makes vinegar for us to drink. I just think it sounds horrible. I like the nature part of life, and I think we should have more rather than less of it.

Niko: Well, Erika, thank you so much for coming on the podcast.

Erika: Thank you so much. This was fun.

I flew to Minnesota and interviewed Adamala for about three hours. First, I wanted to understand the bottlenecks with SpudCell. What engineering changes will they make to get it to reproduce continuously?

Second, Adamala is one of the leading opponents of creating mirror life, a program funded in part by Coefficient Giving through the Mirror Biology Dialogues Fund. I pushed her on the idea that SpudCell, or the ability to create life from the bottom up, could uniquely enable the creation of mirror life. I wanted to understand how we avoid that.

The third part of our interview was about understanding the point of making synthetic life. SpudCell, as you’ll hear, is nothing like cells on Earth today. I think it stretches the imagination to claim that making these cells teaches us much about biology. It’s more of an exercise in hacking biology to get the outcomes we want! I wanted to understand why that matters, and Adamala had some really good responses about how SpudCell could expand life’s chemistry, enable us to make new types of medicines, and help biologize the economy.

Watch on YouTube; listen on Apple Podcasts or Spotify.

Timestamps

(00:00:00) – Opening
(00:00:52) – SpudCell is not alive
(00:05:16) – How SpudCell works
(00:13:14) – Recycling as a major bottleneck
(00:19:24) – Genome inheritance and cell division
(00:34:19) – Bottom-up vs. top-down cells
(00:40:38) – What synthetic cells could be used for
(00:46:29) – Evolution and complexity of synthetic cells
(00:54:51) – Why Kate changed her mind about mirror life
(01:01:27) – Why life evolved a single handedness
(01:05:12) – SpudCell as an enabler for mirrored life
(01:14:38) – Biotic
(01:21:58) – China and 2050 goals

Transcript

00:00 — Opening

Kate Adamala (00:00) Life as a phenomenon is probably very easy, given how quickly it started on Earth. Many people were saying that even making growing, replicating cells from chemicals is impossible, that you need some mystery spark of life. And we just did it. We put together molecules, and they’re growing and replicating. There’s nothing magical to me in how molecules make themselves into life if you give them enough time and the right conditions.

Niko McCarty (00:25) I flew all the way to the University of Minnesota to interview Kate Adamala, whose group, about a month ago, published the first synthetic cell that eats, grows, and divides, built purely using molecules that the scientists put there. In this interview, I want to know not only how they built Spud Cell, as it’s called, but also what it’s good for. Kate, welcome.

Kate Adamala (00:51) Thanks for having me.

00:52 — SpudCell is not alive

Niko McCarty (00:52) My first question is: Very few science stories go viral like Spud Cell did. There wasn’t one New York Times article. There were three. I’m wondering what that’s like. In that first week, what did you make of it? Did you expect this would get as much attention as it did?

Kate Adamala (01:12) I definitely didn’t expect as much attention. It was very humbling that something we’ve done generates so much attention, most of it positive. That was very surprising, but also very happy-making. It was amazing to be able to talk about it. I always want to talk about my work, and it’s really interesting to find so many different people wanting to talk about it because it makes me think about it in different ways, from different perspectives. It was definitely weird.

Niko McCarty (01:46) What were some of the weird things that happened?

Kate Adamala (01:48) There wasn’t a single weird thing. Most of my life is pretty quiet; it’s the normal work of a scientist. Then suddenly, most of your day is taking calls and doing interviews, and that’s not the job I signed up for. That was definitely a new experience.

Niko McCarty (02:08) When these journalists came, you told everybody that this thing is not alive, right? It’s a synthetic cell that eats, grows, and divides, but it stops dividing after about five generations. You were tamping down on the hype that other people were writing about. What would be required for you to say this thing is actually alive—at the risk of upsetting other scientists, probably?

Kate Adamala (02:41) It has to keep going. It’s that simple to me. Life, to me, is defined by its robustness, by its ability to keep going—like the famous quote from Jurassic Park: “Life finds a way.” A Spud Cell could not find its way out of a straight tube. It’s just not robust. It’s capable of undergoing all those processes that we normally associate with life, but not continuously, constantly.

To declare this alive, it needs to be able to keep going, keep replicating as long as there are resources. I cannot say indefinitely because no life replicates indefinitely, but with an abundance of resources and no limiting factors like pH or the size of the vessel, it should keep replicating. The Spud Cell cannot do that right now.

Niko McCarty (03:32) What makes you confident that we can engineer robustness when the robustness of the natural world seems to have been emergently shaped over long timescales by evolution, which is pulling on many levers at once in this very multiplex, combinatorial fashion? Doesn’t it seem impossible that we could engineer robustness that matches evolution?

Kate Adamala (03:55) That might be the pride and overconfidence of a scientist, but I think we can do it. When I first started in this field, many people were saying that even making growing, replicating cells from chemicals is impossible, that you need some mystery spark of life, some mystery component we don’t understand. And we just did it. We put together molecules, and they’re growing and replicating.

Niko McCarty (04:16) Wait, you’re saying people believed in vitalism when you—

Kate Adamala (04:20) Totally. There are many biologists who think there’s something unique about biological life, that there’s some life force—who knows what—that is impossible to replicate in the lab. I always thought that it’s chemicals. Chemicals are doing it, so we can put together chemicals in the right way. Until I’m proven wrong, I’ll keep believing that robustness is also possible to engineer.

Niko McCarty (04:45) To clarify, you think, theoretically, it is possible to build a synthetic cell that is as robust as a natural cell if we just hit on the right combination?

Kate Adamala (04:56) Yes, I think so.

Niko McCarty (04:57) You see it as a big search problem in chemical space.

Kate Adamala (05:01) A huge combinatorial chemical problem, which makes it fun. But yes, I think we totally can. I don’t think there’s anything special about natural life in terms of its ability to solve problems and adapt to different environments.

05:16 — How SpudCell works

Niko McCarty (05:16) How long did it take you to make Spud Cell once you decided, “We’re going to try to put these components together and make a dividing organism”?

Kate Adamala (05:25) It was a trial-and-error process. This particular project’s been going on for about five years in my lab. We had a lot of ideas that failed, a lot of ideas that sort of worked and led us to other ideas, until eventually one of them worked. It wasn’t that one day I woke up, wrote up a plan, and we went and did it. It was all failed experiments that informed the next round of also-failed experiments, and so on until something started working. It was more of a Roomba-on-the-floor process. We didn’t go in a straight line; we just browsed.

Niko McCarty (05:59) What is Spud Cell? What are its key components, and how do they work together to give rise to this thing?

Kate Adamala (06:06) Spud Cell is a phospholipid liposome, so it has a membrane just like most natural cells. That membrane is very simple. It’s a two-component membrane. But we do know of a living cell that has a two-component membrane and can live fine with that. It’s not totally crazy; it’s not too simple.

Inside that membrane, we have a genome. That genome is 90 kilobase pairs. It’s tiny by natural cell standards. It’s encoded on several different plasmids, so it’s a multipartite genome. That genome gets replicated by a rolling-circle viral polymerase. We make copies of the genome inside the Spud Cell.

That genome expresses proteins. We have ribosomes, tRNA, all the enzymes that charge tRNAs, all the amino acids, all the small molecules. We have transcription and translation.

That translation produces proteins. One of the proteins it produces is, I think, my favorite protein in the whole world: alpha-hemolysin. It’s a bacterial toxin that creates a pore in the membrane. It’s a very dumb pore. It’s not a gated channel; it’s just a hole in the membrane. But the beauty of it is that it self-assembles. Because we make that alpha-hemolysin, we can have Spud Cells take up nutrients from the media by a concentration gradient and remove waste from the inside of the cell, again by diffusion.

The really cool thing is that alpha-hemolysin is pretty permissive to modifications. We’ve modified it by putting a small domain on one part of it that presents itself on a membrane. That domain is our feeding tag. It recruits feeder liposomes. The Spud Cell is a perfect heterotroph. It makes nothing; it doesn’t biosynthesize anything.

Obviously, then you run into the problem of how you add lipids, how you grow, how you make new ribosomes. We don’t. We feed it all. That membrane protein presents a tag on the membrane that attracts feeder liposomes. Those feeder liposomes bring in ribosomes, tRNA, and other nutrients that are too big to get in through the pore. They also bring lipids. Spud Cell does not make its own lipids; it has to eat all the lipids. Those lipids come in from feeder liposomes.

The beauty of the system is that, because the feeding happens as a result of that protein being displayed on the surface of the cell, and the protein is encoded by the genome, if the genome encodes for strong expression of that protein, the cell will grow faster because it’s going to have more of that feeder protein on the surface. If that feeder protein is encoded at lower levels, the cell is going to grow slower.

Then comes division, which is also induced by a tag—a different tag—on the surface of the membrane protein. We have two populations of those membrane proteins. One is for feeding; one is for division. That division, because it’s dependent on a membrane protein tag, depends on how much of that membrane protein you make. If you encode that division protein at higher levels, the cell can divide into more daughter cells.

It’s always daughter cells. They never talk about son cells in the literature, so we talk about daughter cells too. We always make those daughter cells, and the number of them depends on how much of the protein you make. That couples the genotype of the cell to the phenotype, which means how big they grow, how much they can eat, and how much offspring they can produce. That’s it. That’s all there is in a Spud Cell. It’s the simplest possible metabolism.

Niko McCarty (09:50) There’s a lot to unpack there. My first question is: Inside this lipid bubble you’ve made, you’re surrounding something called the PURE system. This is more than 20 years old. In the early 2000s, this Japanese—2001.

Kate Adamala (10:10) I think.

Niko McCarty (10:12) This Japanese team showed that they could reconstitute transcription and translation in a test tube. They figured out all the proteins and cofactors responsible, put those into a mixture, and they sell that mixture. You can buy it online. It’s called the PURE system.

It doesn’t seem so crazy to me to go from that to shrouding PURE in a lipid bubble and adding a few things. You’ve added a couple of clever proteins like alpha-hemolysin and these tags. I know progress takes a long time. I’m not poo-pooing progress, but why did it take 25 years for someone to get this to work if we already had some of the key components for a very long time?

Kate Adamala (10:58) I think part of the reason is that the description I just gave you was very simplified. For example, we’re not using the PURE system—the God-given, original PURE system. We’re using a heavily modified PURE system to make it compatible. It turns out that if you’re trying to replicate a genome and express proteins from that genome at the same time, you really need two different buffer compositions. To make it work in one pot, you’ve got to do a lot of optimization. Some people have already done some of that optimization, but we’re building on that.

Then the yields of expression from different parts of the genome have to be fine-tuned for this to work. There are a lot of annoying details that we give you now as, “This is the concentration to use.” But it took forever and many curse words to figure out what those actual conditions are.

Niko McCarty (11:54) But those things need different conditions. If different buffers are required for genome replication and protein expression, how do bacterial cells do both of those things in the same cell?

Kate Adamala (12:07) They do it in a smarter way than we do. They compartmentalize everything.

Niko McCarty (12:12) But this genome is just sitting there in the cytoplasm.

Kate Adamala (12:14) Most people don’t realize that even bacterial cytoplasm is very compartmentalized. It’s not a freely mixing bag or an aqueous bag. It’s something called deep eutectic, which means it’s not really a freely miscible water solution. It’s very, very dense—almost gel-like.

Water activity in a cytoplasm is pretty low. In theory, it is an aqueous solution, but really, everything has its place. Everything is stuck in different phases; everything has its own environment to be in. We did figure it out eventually—the Spud Cell does use one buffer—but it took a lot of figuring out.

When people originally started doing genome replication and then started doing translation, they thought, “We need different buffers.” But through a lot of trial and error—a lot of error—you finally found out that it’s possible to find one condition under which all of this works. It just took forever.

13:14 — Recycling as a major bottleneck

Niko McCarty (13:14) You went through the components. I think one way to understand how these components work is to start thinking about bottlenecks. I often like to think about problems as troubleshooting, and in troubleshooting something, I learn about the components. Why do you think Spud Cell stops dividing after five divisions? It sounds like it’s many problems.

Kate Adamala (13:41) To me, it’s one problem. Just to be clear, we published the five generations as a stopping point because that’s when we decided, “Okay, we’re going to characterize the hell out of it.” It can go on for a few more, but the yields go down significantly. It’s not like it goes for five generations and then drops down and stops completely. You could get a couple more cycles out of it, but the yields would keep dropping precipitously after each step.

I think the biggest problem is that a Spud Cell does not break anything down. It makes proteins and mRNA, but it doesn’t have degradation pathways other than chemical hydrolysis. You keep accumulating trash: ribosomes that lose activity eventually, tRNAs that get broken one way or another because there’s some chemical hydrolysis error in them. There is nothing to recycle them, nothing to break them down and put those monomers back into circulation.

In a given volume of cytoplasm, with each increasing generation, you have more and more waste—freeloading passengers that sit there and do nothing. That decreases the yield per volume. That’s why I think the biggest bottleneck in a Spud Cell right now is ribogenesis: making new ribosomes. And part of ribogenesis—there’s no good word for it—is “ribo breakdown,” breaking down nonfunctional parts, including ribosomes. Taking out the trash and making new parts. This is a huge bottleneck that I think, once it’s solved, will let it keep going.

Niko McCarty (15:23) This is maybe a foolish question, but instead of breaking things down and recycling them, at least in the early part, wouldn’t it be easier to tag proteins somehow so they sequester on one side of the cell, and then you bleb the waste away every couple of generations? You take out that part, because you’re feeding in new parts all the time.

How do you plan to implement this? How difficult is it to get rid of waste versus recycling broken proteins, which seems harder than just getting rid of waste?

Kate Adamala (15:57) Both breaking down waste and getting the waste out in a little bleb have the same problem: How do you recognize waste? How do you recognize that this particular ribosome has been sitting around and doing nothing for the last two hours? It’s a really big problem that, in natural biology, takes a lot of time and money to solve. There are a lot of chaperones, a lot of proteins that do this quality control all the time. We don’t have any of that in a Spud Cell.

The biggest bottleneck is: How do you figure out who’s the waste and who’s still a productive member of society? I don’t have good ideas right now to solve it. I have some stupid ideas that we might be testing, but I don’t have anything that I know works. That’s definitely the biggest problem.

It’s really easy to break down proteins and RNA. There are nucleases and proteases that I can tag and direct to a specific target. Once I know what the target is, I can break that target down pretty cleanly. But I don’t know what the target is. That’s the problem.

Niko McCarty (17:05) Are you saying that the problem is a lack of basic understanding of biology? Why can’t we just take the known components that do this in cells? What happens when you try to put those into Spud Cells?

Kate Adamala (17:17) There are too many of them. Biology has too many steps. If you start trying to reconstitute all of the natural proofreading processes, you end up with a hugely bloated metabolism, which we’re trying to avoid here. My goal is to try to hack it, like we’ve done with all the other processes—to simplify it. That’s really difficult. That’s turning out to be a huge bottleneck.

Niko McCarty (17:42) But if evolution did not hit on a minimal solution to recycling waste, what makes you confident that you can hit on a minimal solution to recycle that waste?

Kate Adamala (17:52) So far, we got lucky with other things. Our genome replication system is one protein, and cells use huge protein complexes that allow for a lot more regulation but are also much more complex. Same with growth, same with replication. So far, we’ve managed to hack every one of those processes. I think waste removal is also hackable.

Again, that’s my belief. That’s not a fact, because we haven’t done it yet. But I do believe we’ll be able to find a simple way to identify waste. We just need to find it.

Niko McCarty (18:29) It almost feels like if you keep finding hacky solutions, which is possible in principle, you might arrive at a solution in the end that is not robust. It only works in the lab, and then it has none of the applications that the news media was reporting.

Kate Adamala (18:43) Yes. That’s the biggest challenge right now: The solutions we’re looking for cannot just be a one-off trick that works under very specific conditions. They have to be generalizable. That is the biggest challenge. You just identified that it’s not just the challenge of how to make ribosomes or how to break them down. It’s a challenge of how to do that repeatedly under various different conditions.

Niko McCarty (19:09) Seems like a very hard problem.

Kate Adamala (19:11) It’s a hard problem, but so far, every hard problem we’ve encountered that seemed super hard, we managed to solve one way or another. I wouldn’t be doing it if I didn’t believe it’s solvable.

19:24 — Genome inheritance and cell division

Niko McCarty (19:24) You mentioned that one big bottleneck is recycling or getting waste out because it accumulates with each division. Another problem you write about in the preprint is that you have this 90-kilobase-pair genome. It’s partitioned across many different plasmids or minichromosomes—I’m not sure what you’re calling them. Every time Spud Cell divides, they randomly segregate. There’s some probability during each generation that the daughter cell did not inherit the full genome. How big a bottleneck is that, and how are you thinking about solving that problem?

Kate Adamala (20:02) That’s a very easy problem to solve. The only reason it’s partitioned into small plasmids right now is for ease of working with it, because we were trying to figure out what genes are necessary. Adding and removing genes is much simpler when it’s a plasmid that you can normally clone in bacteria.

There is a way to make a megaplasmid, basically the thing that John Glass does for his mycoplasma-based cells. We already know how to encapsulate the megaplasmid in a liposome. Once we’ve settled on, “This is the genome that I want to persist,” we can put it all on a megaplasmid. I cannot say it’s trivial, because it’s still a technical project to do it, but we know how to do it.

The thing is, those megaplasmids are technically difficult to work with. You need to clone them in yeast. You cannot just amplify them in bacteria like any normal plasmid. That’s why we were using the multipartite genome at the optimization stage: I can clone a new gene or mutate that gene within a day. With a megaplasmid, it is about two weeks to make the whole thing. That’s purely a technical difficulty. That’s not a scientific problem that I don’t know how to solve.

Niko McCarty (21:15) As you add components to Spud Cell to make it more robust or to recycle waste, the genome will get bigger and bigger.

Kate Adamala (21:21) Yes.

Niko McCarty (21:22) Do you have any sense of how big the genome will be at the end of the—

Kate Adamala (21:29) I do, actually, thanks to George Church and Tony Forster. A few years ago, they published a paper where they speculated about the minimal size of an independently living, autonomous cell built from scratch, with the PURE system as an assumption. They speculated that about 120 kilobase pairs is going to be the size. It’s a very well-written paper. They reason through each function and how many genes would be assigned to each function. It all makes sense to me, so I agree with that.

Niko McCarty (22:03) When they wrote that paper, which I’ll link in the video description, they presumably were not thinking about all these hacky workarounds you’re finding. They couldn’t have assumed that the solution to cell division was to express alpha-hemolysin with a tag. They were probably making that genome size assumption based on existing biology.

Kate Adamala (22:26) It’s an upper boundary.

Niko McCarty (22:28) You see it as an upper boundary. You’re already at 90 kb. The upper boundary is 120.

Kate Adamala (22:34) We’re at 90 kb right now. The size of the genome is mostly driven by the fact that it’s eight different plasmids. Each of those plasmids brings its own junk—all the parts a plasmid needs to be a plasmid: the antibiotic resistance, the bacterial origin of replication. The actual protein-coding part of that plasmid is usually about a quarter of the plasmid itself. Once you put them all onto one plasmid, that is going to become much smaller.

Niko McCarty (23:04) How big?

Kate Adamala (23:05) About half the size. The megaplasmid that encodes all the genes we now have in a Spud Cell is going to be about 40 to 45 kbp.

Niko McCarty (23:15) Assuming you can put these onto a single megaplasmid, get that into Spud Cell, and it continues to work as expected—going back to this idea of robustness—now you have a single megaplasmid that you need to partition during division. That means you need to express many copies if it’s randomly segregating. If you only have two copies, there’s a very high probability that one daughter doesn’t get a copy. Even with four or five copies, you’re going to suffer the perils of noise in random division. You might need to express lots of copies, but then that throws off the energy balance and all these other balances.

Kate Adamala (24:00) Which is why random partitioning is not going to fly. I want a cytoskeleton. I want a very simple cytoskeleton. There’s already been great work from people like Petra Schwille and Marileen Dogterom that demonstrated that an actin-based cytoskeleton can be constructed inside a liposome. I want to anchor my plasmid somehow and then partition it into daughter cells in a more controlled way.

Niko McCarty (24:31) Can you say anything else? How do you think this is going to work?

Kate Adamala (24:36) We can have a plasmid with a DNA sequence that’s recognized by a DNA-binding protein, and then have some sort of tendril of actin, which we already know can be formed inside synthetic cells. That tendril of actin can go into the membrane. It doesn’t have to be perfect. You don’t have to have poles like in a natural cell. You just have to have two or three of those tendrils that go somewhere into the membrane.

The most likely scenario, just geometrically, is that they’re going to be going in opposite directions because of steric hindrance. When a cell starts dividing, they’re going to pull apart, and they might influence the geometry of the segregation. It won’t be perfect, but it’s going to be much better than this smoothie we have right now, where things partition randomly.

Niko McCarty (25:25) That’s also related to cell division. The solution you hit on for cell division is very strange. It’s really not how cells divide at all. It’s bulky crowding of proteins that repel each other and lead to a sort of schism. One thing I noticed in the preprint is that you have these proteins sticking out of your cell, and these proteins are grabbing onto the bulky molecules that cause cell division. My understanding is that that interaction is permanent.

Kate Adamala (25:59) It is. Like every interaction, it has a k-on and k-off, but it’s pretty tight binding.

Niko McCarty (26:06) What that means is that after the first division, the daughter cells already have these proteins attached to them. Does this make them divide faster? By the time you get to the fifth generation, is the cell just studded with all these things and completely incapable of dividing for that reason? How do you get them off?

Kate Adamala (26:25) These are two independent problems we’re looking at. We’re looking at one of those problems, the persistence of the bulkiness, with a modeling collaborator.

This division process is pretty well characterized. People act like this is the first time they hear about it, but it’s been really well characterized by Reinhard Lipowsky and others working on it for over a decade now. It was never genetically encoded before, but as a physicochemical process, they’ve done quite a bit of membrane dynamics to figure out how it works. There are some models of it. We’re following those models now to see how many of them there are and how that affects the daughter cell’s geometry.

There are two competing theories. One is that it makes it harder for them to divide because the new alpha-hemolysin is being made and the membrane’s already crowded, so it’s harder to go in. But then you also are possibly pre-poised to divide because you’re already coming in with a lot of those bulky proteins. One or both might be true.

That also circles back to this huge problem we have, which is waste removal. Ideally, I would love to start each generation with a clean slate. I would love to find a way to recycle those alpha-hemolysin proteins: have some sort of protease that goes into the membrane and either cuts the FLAG tag that holds that bulky protein or cuts the whole thing away. Right now, we don’t have that. That’s definitely limiting the number of generations.

When we want to get to a higher number of generations, we have to solve that problem. We have to find a way to either completely remove those proteins or at least limit their number at each generation. It’s one of the problems we’re still working on.

Niko McCarty (28:15) With Spud Cell, you have waste accumulating inside the cell because ribosomes and tRNA stop working. You have waste accumulating outside the cell, which is all these proteins that have bound to Spud Cell. Are those two completely different recycling problems?

Kate Adamala (28:32) I don’t think so. The membrane problem is obviously a little more complicated because of the presence of the membrane. You have to figure out a way for this protease to poke its head up above the membrane to be able to cleave it. But the general problem is the same: How do you recognize what’s waste?

I’ve been thinking of scenarios like, what if, between each generation, you just hydrolyze all of your hydrocarbons? Would that be possible? Then start from scratch. That’s obviously resource-intensive. I don’t have a good solution yet. That’s one of the problems we’re working on. But to me, it definitely will not be continuously capable of division, so it won’t be able to claim aliveness until we solve that problem.

Niko McCarty (29:24) Energy intensiveness doesn’t matter, right? You just want one of these things to keep dividing. Theoretically, you can supply infinite energy for a Spud Cell.

Kate Adamala (29:35) Right now, yes, because it’s in the lab. But eventually, we do want applications for it. They cannot be hungry little caterpillars all the time. They have to find a way to metabolize things and make their own energy.

Niko McCarty (29:48) Natural bacteria divide using things like contractile rings, where they express proteins that form rings toward the middle of the cell, and the rings pinch down and cleave it in two. That system has been so well—

Kate Adamala (30:03) Yes.

Niko McCarty (30:04) Right? Every microbiology department in the US has somebody who studies bacterial division. But earlier in this interview, you said that you didn’t even try that. From the beginning, you were using this bulky alpha-hemolysin approach, which is so unnatural. Why did you not try the contractile ring?

Kate Adamala (30:24) That’s because a contractile ring has been an obvious choice for synthetic cell division, and other people, better microbiologists than me, tried already. There are a few labs that have been trying to build a protein-based division system for—not forever, but a very long time—and they still haven’t succeeded.

To me, that was, “I’m not going to compete with the people who have been doing it since I was in undergrad, and they still haven’t figured it out. I’m not smarter than them. I’m not going to figure it out where they failed.” I still don’t think it’s a lost cause. I think there is a way to figure it out. We just need more work on it.

Niko McCarty (30:59) When you say they have not succeeded, I’ve seen papers where people get liposomes to divide.

Kate Adamala (31:05) Not by genetically encoding the division. That’s the key. There are many ways to make a liposome divide, including protein-induced division. But to me, the key is that you have to couple phenotype to genotype. You have to genetically encode what makes you grow and divide, because otherwise you don’t have the selection pressure; you don’t really undergo a full cell cycle.

At the beginning, you asked why Spud Cell achieved what others couldn’t. People have made liposomes grow and divide before, but we’re the first who actually made it happen as a result of the metabolic activity of the cell itself. It’s the same with the protein-based division systems. People have made the blebbing work, but not as a result of internal protein expression.

Niko McCarty (31:57) It’s a fine line, though, right? You’re providing a genome, but you’re also providing tons of nutrients and everything else.

Kate Adamala (32:04) Wait.

Niko McCarty (32:05) It’s kind of like moving the goalposts.

Kate Adamala (32:07) Totally. That’s why I don’t know what alive means. On one hand, I don’t think it’s alive because it doesn’t keep going, and I have to provide everything, including ribosomes. But then every life form needs something provided to it. There is no life form that’s going to completely live on its own.

Even if you have, for example, HeLa cells in a dish, they’re alive. No one questions that. But if you forget to come on a Saturday to passage them, on Monday they’re going to be dead because you didn’t provide for them. Life always needs to be provided for. That’s why I don’t have a good definition of the limit of what we should provide.

Niko McCarty (32:50) Going back to the contractile ring and people who have been trying to make cells that divide with the contractile approach, that system is so well studied. As someone who has looked at that field—you might caveat and say, “I’m not an expert”—why do you think that’s not working? Is this a problem of unknown unknowns? Are there other factors involved that we don’t know about? What is it? Why is it so difficult to get this to work?

Kate Adamala (33:25) Obligatory: I’m not an expert. I think the protein cytoskeleton is very close to working. I don’t think it’s an unknown-unknown problem. I think it’s an engineering problem, a stoichiometry problem. Some other groups, for example Cees Dekker, have shown that they are getting really close to totally autonomous, multiple cycles of blebbing. It’s just pushing it through this final finish line.

That’s why, when people ask me what’s the future of Spud Cells and whether they’re going to be dividing by this protein crowding model, I keep saying, “I don’t think so.” I think in the next-generation tater tot cells, we’re going to be able to replace this protein crowding system with something better, some actin-cytoskeleton-based division. I think we’re really close.

34:19 — Bottom-up vs. top-down cells

Niko McCarty (34:19) What would falsify this program for you? Your goal is to get continuous divisions. Let’s say it’s 2030, and it’s still only seven divisions. Would you ever admit defeat and say, “Actually, maybe it’s not possible to create a synthetic cell that divides in perpetuity”?

Kate Adamala (34:46) Two answers to that. One is that I think we’re going to have a perpetually dividing cell by 2030, because we already have a cell that can grow and divide, and now we can have more selection pressure on it.

But if we get to a few years from now—and not just me; I’m not the only lab working on it, there are other labs working on this technology—and we all fail as a field, then that means there is an unknown unknown. There’s something about a living cell. At that point, I would say the bottom-up approach has not proven viable, and we should focus on the top-down approach, the Craig Venter approach, the minimal cell approach.

That cell is growing and dividing. We still don’t understand it fully. But that’s more of a tractable engineering problem because that’s a limited set of components, and we just need to figure out what they are. I hope we don’t get there, but if we do, and if it really looks like we can’t hack it from the bottom up, then I would switch to putting all the resources into the top-down approach.

Niko McCarty (35:53) What else needs to be done in the top-down approach? I haven’t carefully followed the J. Craig Venter Institute’s progress. They’ve been working on it for so many years. All I ever see in the media is that it has 400-and-something genes, and they still don’t know the function of X number of these genes. But surely there are deeper problems limiting it. What else is there?

Kate Adamala (36:15) That’s the biggest problem: What are those essential genes of unknown function? Right now, they already have a defined medium for that cell. It’s no longer a kitchen sink. They can use chemically defined media to grow it. They can minimize the lipidome. That’s what I meant earlier when I mentioned there is a living cell that uses two lipids in its membrane. That’s the Craig Venter cell grown under certain conditions. It can have a membrane made out of two lipids.

Niko McCarty (36:45) This has been published?

Kate Adamala (36:46) Yes, it’s been published. I’ll send you the link. It’s work out of James Saenz’s lab. It’s really minimal. The biggest bottleneck keeping it from being fully understandable and engineerable is the fact that there are several dozen genes we know are essential, but we don’t know what they do. We have not even the faintest idea what the hell those proteins do. They have to be there because if you knock them out, the cell gets dead, but that’s all we know about them. Figuring out the functions of those is what stands between that cell being fully knowable and where we are right now.

Niko McCarty (37:23) But after throwing tens or hundreds of millions of dollars at a problem—and maybe I’m taking too aggressive an approach on this question—why is it so hard to figure out the function of proteins? Why can we not grow these things in a hundred little flasks, give them to people, and do some super-resolution microscopy? Well, they do. So what’s the holdup?

Kate Adamala (37:47) That’s the unknown unknown. I don’t know. I’ve been very closely following that field, and I’m friends with a lot of people who have been in it for a long time. It’s a very difficult problem, partially because biology has only two states: live or dead.

That’s the beauty of a Spud Cell: It can be mostly dead; it can be a little bit dead. You can figure out a lot of functions in a suboptimal way. Right now, the Spud Cell divides very slowly. If you break something, it’s not going to completely stop dividing; it’s just going to divide slower. I can still figure out what went wrong because my system is still working. It’s experimentally possible.

With the minimal cell, if you break something, the cell gets dead, and you have no cell. That’s the problem with living biology in general: To figure something out, you have to do it on a cell that’s still viable, because if it stops being viable, you don’t have a cell anymore. I think that bottleneck, the fact that we do not have a slightly dead biology, makes it much harder to work with because you have to keep it viable in order to work on it.

That’s why I think this gray area where a Spud Cell sits right now, where it’s not really dead but it’s not yet alive, is a very comfortable place to be in. I can break a lot of essential things. I can mess with the fundamentals of metabolism. With the minimal cell, they’re stuck with either studying a fixed, dead cell or trying to figure it out on a living metabolism without making it dead. I think that’s just technically really challenging.

Niko McCarty (39:30) Isn’t it so sad that so many of the tools of molecular biology require that we kill cells, extract their molecules, and study their molecules? Even then, almost everything we study is in isolation. We study RNA on its own; we study proteins on their own. We sequence the genome only to later realize that the genome has a particular 3D shape, a topology that also really matters for how the genes are expressed.

Kate Adamala (39:59) Mm-hmm.

Niko McCarty (40:00) Isn’t this also a rallying cry that if we actually want to understand biology, we need methods that enable us to study it across space and time on living samples?

Kate Adamala (40:11) Or somewhat living samples. That’s the thing to me. I think that’s why it’s so hard to study living biology: You have to keep it alive. If we had a state in which we could study it, kind of like a breadboard for biology, where we could see if a particular circuit is ticking without turning the whole thing on, then at least some of those problems might become easier.

40:38 — What synthetic cells could be used for

Niko McCarty (40:38) That’s a very good segue, because the next thing I was going to ask you is: What is the point of Spud Cell? The argument I really understood was that Spud Cell could be a very interesting platform, like you said, to test things. I want to add a new genetic circuit in my cell: Does it work or not? You’ve stripped away all the complexity, so it’s good for troubleshooting. But the question I always had was: If you made a circuit work in Spud Cell, does that have anything to do with making that circuit work in an actual organism?

Kate Adamala (41:11) No, it doesn’t.

Niko McCarty (41:13) You’re saying that’s not the point. It’s that we would actually move over large swathes of the biotech industry.

Kate Adamala (41:19) Large swathes of it. Not all of it, but big swathes of it.

Niko McCarty (41:22) We would have bioreactors not of CHO cells making drugs, but of Spud Cells making drugs.

Kate Adamala (41:28) Probably not the Spud Cell in its current iteration, because it’s just—

Niko McCarty (41:31) Something very robust.

Kate Adamala (41:33) Once Spud Cell becomes more robust, once it can actually replicate under perfect conditions indefinitely, then it can become a biomanufacturing strain.

Niko McCarty (41:44) Can you say what you think some of the likeliest applications would be?

Kate Adamala (41:48) Catalysis that’s not possible with natural proteins. If you need an industrial process that requires very high pressure and temperature, those are very difficult to turn into a bioprocess right now because of those physicochemical limitations. But we know that, for example, if you add fluorinated molecules—basically make Teflon proteins—they can withstand much higher temperatures. We can only do that at a small scale right now because you have to do it in vitro, in a PURE system in a test tube. There’s no other way to introduce those fluorinated amino acids, for example. They’re not compatible with being introduced naturally.

Enzymes that require more chemistry than natural life can give you—that’s one huge field. I think now that the technology is catching up, that field might explode, because people doing protein design were working with what’s available to them. New proteins were being designed with the building blocks that are available. We already know from in vitro studies that there are multiple hundreds—I don’t know the current number, but it’s over 300—of different noncanonical amino acids that can be translated even by a natural ribosome. That huge chemical diversity doesn’t exist in biology right now. We’re not designing proteins taking advantage of that chemistry.

That’s why I firmly believe we can replace the petrochemical industry with bioprocesses. Not with bioprocesses using modern life as we know it, with the 22 puny little amino acids with low chemical diversity—sorry, biology—but with the actual huge diversity that’s available, that’s possible. Bioproduction, especially of molecules that are very far outside the constraints of natural biology, is going to be a big application.

Another one is drugs and biomolecules that are bioorthogonal, that are not made out of natural building blocks. You can, for example, inject a drug into your system, and your proteases don’t chew it up immediately. That’s going to be a big one.

Another one is that right now, Spud Cell needs to be fed everything, and it’s incredibly picky about its nutrients. But because it’s a platform we hope to bioengineer, I want to be able to engineer it so it can eat anything. I want to go from a toddler to a teenager that eats anything you put in front of it.

That enables a globally distributed bioeconomy, because different places, different climates on Earth, can grow different types of biomass. That’s a limiting factor right now in the development of the bioeconomy, because those bioreactors need to ferment a very particular set of biomass. You have this bioreactor that can ferment biomass from soy but not from prairie grass. If you can build a Spud Cell that has enzymes that can chew up a lot of different food sources, then they can ferment anything. That’s another area.

Obviously, we’re far in the future. We’re not talking about a five-year horizon. But by 2050, I think we’ll get there. If we don’t, we’re in trouble as a civilization. We have no choice.

Niko McCarty (45:08) It’s interesting that when I asked you what the point of Spud Cell is, you focused on the technological applications. You did not say what is normally the rallying cry for synthetic biology, which is, “We are building this to understand life”—which honestly has always struck me as a more tenuous position.

I can accept your premise that, yes, if there were a robust Spud Cell, we could engineer with very interesting new chemistries, and we wouldn’t have to worry about crosstalk and all these other factors. But building life to understand it—Spud Cell is so different from natural biology. Does it really tell us anything about natural biology, or does it tell us more about chemistry?

Kate Adamala (45:52) I think it tells us more about chemistry. If you take a pathway from biology and study it in isolation in a Spud Cell, then you can study that particular pathway much better. In that way, it does teach you a little bit about a natural pathway, because you can look at it separately from all the other pathways. But the Spud Cell does not teach us how a complex natural cell works. It’s just too different. Because of all the hacks we did along the way, and more hacks we’re hoping to do, I don’t think it’s going to be a reliable model that teaches me how my own cell works.

46:29 — Evolution and complexity of synthetic cells

Niko McCarty (46:29) Another thing I was wondering: You’re building Spud Cell, and it will improve over time. Let’s say it’s 2030. You have a Spud Cell, and it can still only divide ten times. It’s still not robust. You all get together as a group and say, “I know how to make this more robust. We’re going to evolve it. We’re going to put it in a bioreactor and pressure it, and—”

Kate Adamala (46:52) We’re going to do that way before 2030.

Niko McCarty (46:55) Okay, but then the question is: Hasn’t that destroyed the whole point of Spud Cell? As soon as you let evolution act on this, you don’t understand the components anymore.

Kate Adamala (47:06) That’s the problem with natural evolution, because you’re starting with something you already don’t understand. With a Spud Cell, I can sequence the hell out of it at every generation.

Niko McCarty (47:15) We can do mass spec on all the components.

Kate Adamala (47:18) And I know I have a baseline that I fully understand. Every new thing that pops up, every point mutation, every new functionality will be compared to the baseline. If I don’t understand it, then I throw it out the window, because I’m not going to have something I don’t understand in a system that’s designed to be understandable.

Every tiny improvement, every way to deal with a selection pressure, is going to be compared to the baseline we now understand. It’s not like adding an unknown brick to something you already don’t understand. It’s going to be a baseline platform we do understand. Evolution might create unexpected solutions, but they won’t be unknown solutions.

Niko McCarty (48:06) That’s really interesting to think about: Spud Cell as a benchmark to trace evolution. That’s—

Kate Adamala (48:12) Mm-hmm. That’s why I feel—

Niko McCarty (48:14) Very interesting.

Kate Adamala (48:15) That’s why it’s key to me to couple genotype to phenotype. I think we’re not as smart as evolution. I really think evolvability is a key function that will enable us to improve Spud Cells faster. Now we have a genome that’s coupled to the outcome, to the number of offspring, so now we actually can evolve it.

Niko McCarty (48:36) You could build some kind of assay of robustness. You measure robustness, take Spud Cell, and evolve it. Every few generations, you sequence it and do mass spec. The thesis is that you could actually map what makes things robust under different— That would be very interesting.

Kate Adamala (49:02) Right now, nobody’s doing that because the Spud Cells still need to keep replicating. That minor detail of it not replicating fast enough. We’ll start doing it the moment I have continuous replication. That’s the goal. I have several collaborators who are tippy-tapping right there in their labs now, waiting for it to happen so they can start doing it. But until we can stop worrying about housekeeping—waste removal and the number of cycles—we cannot do that yet. That’s why the first problem is to make it more robust.

Niko McCarty (49:38) Do you think evolution might be required to make it divide longer?

Kate Adamala (49:44) I hope not, because it’s chicken and egg then. You can’t really have evolvability unless you can go through multiple generations. I think we can hack it to the point where it will be replicating. It might not be very efficient at it; that’s why I say “hack it to that point.” But once we have multiple cycles of growth and replication, evolution can improve on it for sure. I think evolution is much better than me. Evolution is much faster, much more efficient than artificial design, than directed evolution that humans do.

Niko McCarty (50:20) This is a difficult question to pose, but bear with me for a second. There are all these knobs that evolution can pull on, even within a single molecule. There was just a paper in Cell where they were studying a particular virus, and they said the infectivity of this virus—how infectious it is—is not only a function of its sequence. It’s an RNA virus, but that RNA, when it enters the cell, folds into a particular shape that also plays regulatory roles in the infected cell’s genome. In other words, the structure and the sequence of the RNA are both playing functions for the virus.

If you look at the evolution of genomes, historically we just sequenced genomes and that was good enough. But we now know that certain types of cancers in human cells emerge not from sequence errors, but because the way chromosomes physically package in relation to each other is off. Clearly, there are these very combinatorial knobs that evolution is pulling on.

The vitalism that people speak of, which I don’t believe in either—I subscribe to the view that life is just chemicals computing—but maybe there is something to this belief that if you strip things down too much, you reduce the levers evolution can pull on, and you just won’t ever match the robustness or sophistication of natural life. I’m wondering if you have takes on that or subscribe to that.

Kate Adamala (52:03) I don’t mind complexity. I love complexity as long as I can understand it. I think both positions can be correct. Life can be an effect of molecules, just physicochemical behavior of molecules, but it has to be complex enough to be robust.

For example, a modern airplane is incredibly complex. I don’t think there’s a single person right now who can understand every single system of a modern airliner and explain every single subsystem, all the avionics, all the aerodynamics of the airframe, everything. There’s not a single person who can control all of that. But that doesn’t mean an airplane is a magical being. It’s still designed, engineered, and you can explain every function of it. It just became so complex. That’s how I think of life—

Niko McCarty (53:07) But the difference is that we can study airplanes because we can see all the components. We can put the transistors under a microscope and see them. But biology—if you were to evolve Spud Cell and it hits on a solution to the division problem, let’s say, and then you study its molecules, do mass spec, and sequence the genome, you’re not actually capturing the topology of the genome, the structures of those molecules.

Kate Adamala (53:32) That’s why you need to slice it and put it on a cryo-EM.

Niko McCarty (53:35) But what if evolution hits on solutions that our measurement tools are not equipped to even detect?

Kate Adamala (53:42) That’s a really interesting problem. I think we need a better measurement tool, and that measurement tool has to be computational. That’s another thing I think Spud Cell might enable: building a true digital twin of a cell. Now that we’re starting with something simpler, we have a collaborator who is building an atomistic-resolution model of Spud Cell. Because we can. I can tell you what the concentrations are and where those molecules are. You can make a model right now. As we build complexity, that model can grow.

With a natural cell, we don’t have that model because we don’t know where everything goes. But I think the problem you raised is really important: What if the evolutionary process finds a solution that we can’t backtrack, that we can’t figure out? I think computational tools will help a lot with that.

Niko McCarty (54:35) But how long can you run an all-atom simulation of Spud Cell? Even though Spud Cell is very simple, you probably can’t run that for very long at all, right?

Kate Adamala (54:44) Yep. My goal is seconds, which is very long by those standards, but I was told it might be possible.

54:51 — Why Kate changed her mind about mirror life

Niko McCarty (54:51) I want to go to mirror life now.

Kate Adamala (54:54) I don’t. I don’t want to make mirror life.

Niko McCarty (54:56) No, I know. When that big report came out, we covered it at Asimov Press. I wrote a long piece with Finn Morehouse about the report. If somebody were able to make a cell that has the opposite handedness of natural life, meaning all of its molecules had been replaced with their mirrored versions, that mirrored cell could evade the human immune system. It could evade viruses in the environment and proliferate out of control; there would be few things to keep it in check. Most scientists acknowledge that this is a very distant thing. We are nowhere near able to build such an organism.

But I want to go back to 2019. There was a conference, some kind of workshop event, where synthetic biologists got together. You were there; John Glass was there. Everybody looked around and was like, “Yeah, we should make mirror life. This would be amazing.” Why did you think at the time that it would be interesting or useful?

Kate Adamala (56:09) Because I didn’t know enough. I didn’t know enough to know better.

Niko McCarty (56:11) But what were people saying at the event? What was going to be the functional purpose of mirror life?

Kate Adamala (56:17) All the benefits we thought about getting from mirror life are the reasons we now think it shouldn’t be made. We thought it would be stealth to the immune system. Great, it would make a fantastic drug. It would not be possible to infect it with natural phages. Great, it would make a fantastic bioreactor.

We didn’t think about it from the risk-analysis point of view. We only thought from the benefit point of view. Only after we started talking to evolutionary biologists, ecologists, and experts in immunobiology did we realize that all those things we considered features are actually bugs. If it’s stealth to the immune system, it doesn’t just make a perfect drug; it also would make a perfect pathogen. If it’s not possible to eat it or infect it, then if it leaks out of your magic bioreactor, it’s just going to spread in the environment.

Niko McCarty (57:17) Who were the people who first approached you to tell you, “Hey, you should not do this”? Even the NSF funded it. It wasn’t just scientists in a room. Big funding agencies also gave money. Probably hundreds of people were aware of this.

Kate Adamala (57:32) Yes. It was public—NSF grants are public—so we were never hiding that we were doing—

Niko McCarty (57:37) Of course, but I just mean that to get an NSF grant, dozens or hundreds had to say this is a great idea. Who first approached you and was like, “Hey, I want to talk to you about this”?

Kate Adamala (57:48) It was very gradual. It was over a couple of years—really, over a year—of talking in the biosafety and security community about that idea. There was not a single person who came and said, “Hey, I think this is a bad idea.” People started genuinely asking questions. Not in this Midwestern ironic way of “I’m just asking questions,” which means “I’m criticizing you.” No, no. They were genuinely asking questions about, “How do you make sure it doesn’t take over in the environment once it leaks out?” It was coming from curiosity, but also concerned curiosity.

We, the people working on it, realized we actually did not have a plan for that. So, okay, we have to come up with a plan for that. Then we started thinking: How would we prevent persistent infection or persistent replication in the environment? We started realizing that we couldn’t think of a way to do that.

Then we started going out to more people, asking ecologists, for example, “Hey, I have this problem. What if I make an organism that can do this? How would you limit that organism?” Those people could not come up with a way to limit it. Then we started realizing that we have a problem and we should not be enabling that technology.

Niko McCarty (59:12) What’s interesting is that after the big mirror life debate kicked off a couple of years ago, people went and looked in the history books. Louis Pasteur said this would be a problem. In the 1800s, Louis Pasteur was like, “If we made mirror-image life forms, it could be really bad.” In the 1990s—or in 1990—these Purdue chemists wrote a letter and warned about this. There was a Wired article in 2010 saying that mirrored organisms would wipe out the planet. George Church’s book with Ed Regis, Regenesis, devotes a huge share to mirror life.

It’s so confusing to me, the timeline of why, in 2019, despite even a very mainstream Wired article, people were like, “This is an amazing idea.” Again, I’m more techno-optimistic than fearful. I acknowledge that mirror life is so far away. I’m just trying to understand the timeline: How could it have been that these scientists in a room were not aware of this?

Kate Adamala (1:00:24) Now we’re aware of it. This Wired article, for example—I think it’s a great article, but I had no idea it existed until we started digging into the background on safety, when we started considering its safety. I think it became more mainstream as the problem became mainstream. It wasn’t an article everyone was aware of.

Obviously, I did read George’s book, but it was not—like, when we first started thinking about it, we thought, “Okay, we can put safeguards into the organism.” It took a long time to put all those things together. I had no idea Pasteur wrote about it until I was way into writing the Science paper. With all due respect to Louis Pasteur, I of course never read his original writings.

Niko McCarty (1:01:17) It was from a lecture, actually, so even harder to find.

Kate Adamala (1:01:20) Now we know those background materials exist, but they weren’t prominent. I think that shows a short—

1:01:27 — Why life evolved a single handedness

Niko McCarty (1:01:27) When the debate starts, people look for evidence. It’s selective bias. I’ve read books and articles that state as a fact that there’s no reason life adopted the handedness it did. It’s just a fluke of evolution that the cell that evolved to the present day and gave rise to all living organisms adopted those chiralities in its molecules. It chose those handednesses.

Kate Adamala (1:01:59) I agree with that, actually.

Niko McCarty (1:02:01) My question is: How do we actually know that? You’re making the assumption that it could be reversed and that there are no emergent properties or utility of the existing handedness. Also, why would only one handedness survive? It also insinuates that the evolution of life is rare. I’m curious what you make of it.

Kate Adamala (1:02:28) That’s a four-part question.

Niko McCarty (1:02:31) The question was terribly posed.

Kate Adamala (1:02:33) It’s actually a really good question. It’s four really good questions. Let me answer the second question first. The reason we need one chirality is that life has to be interoperable. Life has to be able to eat other life. A pretty strong hypothesis I support is that if you want a biosphere to work, everything in that biosphere has to be interoperable. If you die, you have to decompose to elements that other cells can eat. If a predator comes at you, that predator has to be able to eat you. You have to be able to eat food. Everything eats everything else, so one chirality is kind of necessary.

On the metabolic level, you have to be homochiral, because otherwise the proteins would just not work. You cannot have mixed chirality because you would have too many different stereoisomers, different possible configurations, different possible molecules that wouldn’t physically interact with each other. That’s the reason life had to pick one.

How do we know it probably doesn’t matter? Because we have a lot of evidence that opposite-chirality proteins and nucleic acids interact with each other just like the normal-chirality ones. That’s why we think it could have happened either way.

The third question in your string was: Does it imply that the evolution of life is really rare? I don’t think so. I think the universe might be lousy with life. Obviously, there’s the Fermi paradox, and I think that’s a very different discussion. Intelligent life might be very rare, but life as a phenomenon is probably very easy, given how quickly it started on Earth and how easy it seems to be. My small lab made molecules into a lifelike system. There’s nothing magical to me in how molecules make themselves into life if you give them enough time and the right conditions.

I don’t think life is a particularly difficult process to get started. We have n equals one because we only have one on Earth, but it started so quickly. The moment Earth had liquid water and a crust, life started. Either aliens came with a salt shaker, or life spontaneously started really quickly—and I take the easier solution: Life started quickly.

Niko McCarty (1:04:58) If it started quickly, it could start often.

Kate Adamala (1:05:01) It could start often, and it could pick a chirality at random.

Niko McCarty (1:05:05) We’ve just never detected it, or it peters out, or it’s outcompeted in its environment. That’s your thesis.

Kate Adamala (1:05:11) That’s my thesis exactly.

1:05:12 — SpudCell as an enabler for mirrored life

Niko McCarty (1:05:12) Another point is: Why are people even having this debate? We haven’t made a synthetic cell, let alone a synthetic mirrored cell, which is even harder. To play devil’s advocate, what is even the point of worrying about this?

Kate Adamala (1:05:30) Because now we can actually do something about it. People wanted to do it. We were funded to do it; others were funded to do it. If you get too close to it, there’s no stopping it. It’s really hard to stop scientific progress when it takes one or two bad actors to get there.

Right now, we’re so early that we could stop this whole field in its tracks by creating peer pressure: All the experts in the field don’t work on it. We’re not building the capacity that gets us closer to it. Now is actually the only time we can stop it. Once we get too close to making a mirror cell, a single bad actor can do it. Right now, we’re too far from it, so we can stop it as a field.

Niko McCarty (1:06:15) One thing I’ve always wondered about this is: If we had the technology to create mirror life, that insinuates that we’re very advanced in biotechnology. If we could make a mirrored organism, surely we could make a mirrored phage.

Kate Adamala (1:06:30) That’s a very safe assumption.

Niko McCarty (1:06:31) Mirrored antibiotics and mirrored—so is it actually as big a threat as people think it would be?

Kate Adamala (1:06:39) That’s a debate. That’s the argument against us being against mirror life that makes the most sense to me: If we can make mirror life, we can also make something that eats it and infects it.

But on the off chance that we’re both right about the dangers of mirror life and wrong about our ability to make a phage for it and release it globally—if those two are true, then we’re totally screwed. There’s nothing about mirror life that would be worth taking that risk. That’s it to me. It’s not like mirror life is completely worthless. It would have applications, but they’re not worth taking that risk. I’m not 100 percent sure we’re right about all those dangers, but I would rather not find out.

Niko McCarty (1:07:37) Personally, I struggle to envision exactly what the point of making a mirrored organism would be. I totally get a mirrored ribosome to make peptide therapeutics. That’s obvious. What do the proponents of making actual, full mirrored cells say are the applications?

Kate Adamala (1:07:55) Biofermentation: You cannot contaminate a bioreactor if it’s made out of mirror life. And drugs, medicine, cell therapies. Imagine a cell therapy where you have a cell you can inject into a patient that has all the properties of a cell. It can detect metabolites, it can produce a drug, and it’s stealth to the immune system. These are the two practical arguments, and—

Niko McCarty (1:08:20) Those are all things you’re saying we can make other ways. It just might be more expensive.

Kate Adamala (1:08:23) Exactly. Which is why I’m now on board for not making mirror cells, because I think the risks are much bigger than the possible benefits.

Niko McCarty (1:08:34) There are two ways to make a mirrored cell. There’s the top-down way, which seems extraordinarily difficult. I don’t even know that this would work, but maybe you could take a minimal cell like the JCVI cell and replace components one at a time, or swap in just the central dogma in a mirrored form. But because of the compatibility issue you talked about earlier, I don’t see how that would work. The only other way is to make it from the bottom up. Isn’t Spud Cell, in a way, the enabling technology to build mirror life?

Kate Adamala (1:09:14) Not really, because the biggest bottleneck in making mirror life is not the actual assembly. It’s the sheer amount of stuff you need that we cannot make right now. Spud Cell is difficult enough to do when all my components come pretty much free from E. coli. All my ribosomes, all the other proteins, everything I use to make a Spud Cell comes from E. coli. It takes pennies to grow it, and obviously it’s a pain in the lower back to purify it, but it’s doable. I can have buckets of it.

With mirror life, you would have to chemically synthesize every single thing because we don’t have a way to make that. That’s a huge technological barrier, and that’s the barrier we’re hoping never to cross. That’s the key enabling technology. That’s why I say a replicating, autonomous mirror ribosome would be my red line, because at that point you can start making all the components that will eventually become a cell.

Niko McCarty (1:10:14) But isn’t it a fear at all that if you know all the components exactly, you’re publishing a recipe where people could just recreate the—okay, so it’s expensive.

Kate Adamala (1:10:26) It’s more than expensive. It’s just not technically possible right now to make that amount of protein. Even state-actor-level resources would not be enough. We just don’t have a way to make that.

Niko McCarty (1:10:38) But what are the things we can’t make? Is it long proteins that are mirrored, too expensive, or—

Kate Adamala (1:10:44) Long proteins, it’s—

Niko McCarty (1:10:46) Because you have to make them using solid-phase synthesis.

Kate Adamala (1:10:48) Then stitch them together and have mirror chaperones that help them fold in the correct way. Resource-wise and technology-wise, we don’t have a chemical synthesis technology right now that would allow us to make those giant proteins. The biggest things we’ve made were those single instances of mirror polymerases. These are not that big an enzyme.

Niko McCarty (1:11:14) You have to make a high concentration of these.

Kate Adamala (1:11:17) High concentration, high amounts of all of them. Then you have to have all the nucleotides to make the mirror nucleic acids. Logistically, you would have to retool your whole scientific enterprise toward it.

Niko McCarty (1:11:32) Is that something you’re monitoring? You said your red line would be a mirrored ribosome that can make copies of itself. Are there other things where you’re like, “We would pause the Spud Cell work if we saw X”? If mirrored solid-phase synthesis becomes really good, what is your line for saying, “Okay, we’re not going to do this”?

Kate Adamala (1:11:53) I think mirror solid-phase synthesis would not give me pause. That’s why we’re pushing to codify what has become a sort of community-enforced red line into law before that technology becomes possible. So then we can—

Niko McCarty (1:12:13) Yeah.

Kate Adamala (1:12:13) We’re trying to frame that conversation. At least that’s my goal. It’s a diverse community. Some people don’t think it should be codified into law. But I would feel much safer if there were something like a Biological Weapons Convention, something that has similar power to the BWC but is about mirror life.

For example, right now no country can come out and say, “I’m making a virulent, weaponizable smallpox,” because international law literally tells you it’s illegal. I would love it if mirror life research got to that point. We’re not there yet. We might never get there, but that’s my goal. That’s the answer to your—

Niko McCarty (1:12:55) Don’t states still make smallpox? There are still bioweapons programs.

Kate Adamala (1:12:57) They do. They don’t admit to it.

Niko McCarty (1:13:02) Yeah, but okay, you know.

Kate Adamala (1:13:04) But also, it’s super cheap and easy to make smallpox, especially if you’re a state actor. That’s the good thing about mirror life now: We’re stopping the development of the technology before it becomes cheap and easy. It’s also a terrible weapon, because bioweapons only make sense if you can immunize your own population against them.

Niko McCarty (1:13:23) What, you mean mirrored?

Kate Adamala (1:13:24) Yeah, mirror life would be a terrible bioweapon.

Niko McCarty (1:13:26) But isn’t it a good ecological weapon?

Kate Adamala (1:13:29) Yes, but you cannot limit it. An adversary—

Niko McCarty (1:13:33) You would also be destroyed.

Kate Adamala (1:13:35) Would also be destroyed. That’s not a state-level weapon. That’s an individual-terrorist-level weapon, and those don’t have the resources.

Niko McCarty (1:13:43) Interesting argument. It’s interesting for me to examine how you justify working on things. Again, I’m a techno-optimist. I am a biotechnologist, but I live in San Francisco, so I’m sort of washed up into the discourse. It’s interesting to talk to someone like you who’s like, “We should stop mirror life,” and then, by the way, we’re making a synthetic cell platform that is the enabling technology.

Kate Adamala (1:14:16) That’s the thing. I don’t see it as an enabling technology, because I think the technical problems toward making mirror life—and I say that as someone who tried to make it, so I know how much it sucks to try to make it—are way beyond what normal-stereochemistry synthetic biology is facing. They’re just huge resource problems.

1:14:38 — Biotic

Niko McCarty (1:14:38) I think that’s a good segue to Biotic. After the Spud Cell preprint came out, you announced there’s going to be this big nonprofit initiative. The goal is to raise $100 million. Is that public? The goal is public.

Kate Adamala (1:14:52) The goal is public. The goal is to enable a global bioeconomy. Do the math.

Niko McCarty (1:14:59) There’s not a lot about Biotic that I’ve seen. I want to understand: What is Biotic actually doing? Is it a grant-making organization that’s trying to bring more people into the field? Is it going to be a building where you’re centralizing efforts? What is the goal here?

Kate Adamala (1:15:18) It’s a grant-giving organization. Right now, our goal is not to bring more people into the field; it’s to focus people who are already in the field. Synthetic cell engineering is an amazing, very diverse field. That’s the beauty of it: People have different definitions of what a synthetic cell is. But that makes the efforts rather unfocused. Progress is being made, but not fast enough.

We as a society really are struggling right now. The world is burning. The world is having a lot of societal problems that can be traced back to literally the way we’re moving atoms, the way our industry works. We think many of those problems could be solved by enabling a bioeconomy. We also believe it will not be possible to enable a bioeconomy unless we have a better way of moving atoms beyond the natural cell. That’s why we think synthetic cells have to happen for the world to become sustainable, for civilization as we know it to survive and be sustainable.

That has to happen soon—soon enough to matter. We’re talking about 2050 as the deadline when we have to start seeing things change. Right now, the efforts in the synthetic cell field are not focused enough. They’re not focused on a single interoperable platform.

At its core, what Biotic wants to do is fund the development of a robust synthetic cell that can create all those applications. All the people who work on developing modules and technologies that enable that synthetic cell have to agree on a standard, on a single platform. For example: This is the translation system we’re all using. This is the genome design and architecture we’re all using.

Those types of standards could be created by the community, with “Pretty please use those standards.” That could be one way of enforcing it. But a much more efficient way would be to fund research and say, “We’ll fund your research. We love what you’re doing for this field. We want you to keep doing it. Just please make your work interoperable with our platform, and we’ll pay you to do that.”

That’s what Biotic is. It’s a grant-giving organization, but the grants aren’t a free-for-all for anyone who wants to make synthetic cells. Those are grants for people who will make useful modules for synthetic cells that will work with all the other modules developed within that ecosystem.

It’s like saying we want to make apps for a Linux operating system, and those apps have to work on a Linux kernel. We’ll pay you to write that app, but only if I can boot it up on my computer. That’s an analogy.

Niko McCarty (1:18:11) Who decides what to fund and what the priorities are?

Kate Adamala (1:18:16) Right now, it’s the Biotic board that decides. As we grow, we’re hoping to build an international scientific advisory board, as the decisions will get tougher as we get more granular.

Right now, the guiding principle is that the modules you make have to operate on a Spud Cell. The only reason for that is that Spud Cell exists. There’s no other synthetic cell platform that grows and replicates. If someone makes a better one, and it’s compatible, we reserve the right to move. We’re not “Spud Cell is the one and only platform” people. We’re “Whatever works is the one and only platform” people.

Right now, we want to win by winning. We have a working prototype, so we’re asking people to contribute to making that working prototype better and better and better until it stops being a prototype. The standard now is: Make modules that will make this one thing that exists better.

Niko McCarty (1:19:19) Have you already shown that Spud Cell works in the hands of others? How reproducible is it? What other labs have made Spud Cells in their labs?

Kate Adamala (1:19:29) We’ve reproduced Spud Cells in Drew Endy’s lab. Right now, other people have reproduced other elements of Spud Cell. It’s been a little over a month since we made the whole thing public, so I don’t know of any other lab that has reproduced the whole thing. Elements of Spud Cell have been reproduced in other labs, but as far as I know, not the whole thing.

Niko McCarty (1:19:55) Even if Biotic raises $100 million, what is the plan to keep it sustained through 2050, especially when interest starts to die down? It’s one thing to fundraise when you’re backed by all the hype. It gets progressively harder and harder to fundraise as the progress becomes more incremental.

Kate Adamala (1:20:14) Hopefully, we will not be fundraising by 2050. The goal is to make it a useful technology. The only victory condition for Biotic is to make it into a commercially useful technology that enables biotechnology, that enables a bioeconomy. That’s going to bring in money.

We say this is an open platform for research, but we did file IP on it. It is not a platform that any company can use for free. We will license that and all the future improvements to any company that wants to make it into a useful technology. The royalties they pay back to Biotic will support further Biotic research. That’s the victory condition.

That’s why we have to succeed. Biotic is not designed as an ongoing, indefinite research operation. It has to start producing viable products that will bring in money, and that money will sustain it going forward.

Niko McCarty (1:21:09) Is the idea ever that you would spin out companies? Only licensing the technology may be interesting, but it’s very uncertain. Would you ever consider, “Hey, we did something very promising. We’re just going to spin out our own company”?

Kate Adamala (1:21:25) We do, and Biotic has this plan that once we start having viable product possibilities, we’re going to spin out Biotic Ventures. Think of it as something internal, almost like a family office run out of Biotic. We as Biotic founders will not spin out companies on our own, but we will support people spinning out companies, and their revenue will come back to the research ecosystem. That ties to our victory condition of making a useful product.

1:21:58 — China and 2050 goals

Niko McCarty (1:21:58) When this announcement came out, there was a lot of discourse about China: “We have to do this, and it’s so much better than China. China has its own synthetic cell effort, and we’re so far ahead.” What’s the deal with China? Who cares about China’s synthetic cell effort? Why not be collaborative? Why does it seem so antagonistic?

Kate Adamala (1:22:23) It really bugs me that there is this competitive narrative to it. It’s obviously tied to geopolitics.

Niko McCarty (1:22:29) But in Drew Endy’s Hoover Institution interview, a big chunk is about China and how we’re ahead of China. I’m like, “What does this have to do with geopolitics?”

Kate Adamala (1:22:39) It’s enforced by the political structure that governs the results. If Biotic is the first to produce a functioning synthetic cell that becomes technologically important and powers a bioeconomy, that bioeconomy is going to run by the rules of the country where Biotic is, or the system that supports Biotic. That’s going to be an open ecosystem.

If the first functioning, commercially viable synthetic cell comes out of research supported by a totalitarian government, then access to it and the ways it’s going to be used and deployed are going to be very different. That’s the biggest difference.

Niko McCarty (1:23:26) Are there Chinese labs that are really good that Biotic would fund but cannot fund because they’re in China?

Kate Adamala (1:23:33) Right now, we’re not funding any work in China, but there is a huge institute in Shenzhen led by Chenli Liu that has amazing expertise in synthetic cell engineering. In an ideal world, I want to work with them, not against them.

Niko McCarty (1:23:46) I visited the Shenzhen Institute, visited him, and saw his lab. They have two floors of robotic labs and very impressive scale.

Kate Adamala (1:23:56) Yes. We all would be much better off if we were able to collaborate. The sad reality is the geopolitics, and you can’t get around that.

Niko McCarty (1:24:07) But if Spud Cell is nonprofit, open-sourcing its technology, and publicly revealing things, isn’t the timeline for China to catch up to whatever you publish about a month? Considering how advanced they are—I was just there a couple of months ago, and I was even more impressed on the second visit.

Kate Adamala (1:24:28) Yes, but that’s why we filed IP on it. We filed protective IP. The Spud Cell is—

Niko McCarty (1:24:33) But that doesn’t matter, right?

Kate Adamala (1:24:36) It does and it doesn’t. We still have to pretend we believe in the law.

Niko McCarty (1:24:41) Assuming this works and Biotic becomes a self-sustaining organization, what is the end goal or the success case in 2050?

Kate Adamala (1:24:52) It has to work, because by 2050 we have to make all the products sustainably, and we have to be truly engineering biology. We’ll have a cell that people can work with, and bioengineering will become more democratized. I know it sounds like a utopia right now, but I really think it’s doable that this technology can become a universal jump-start to biological evolution.

People have been talking about the 21st century being the century of biology, and to me that is necessary. We have to engineer cells in an intentional and accessible way. That’s going to happen by 2050 or we’ve failed. I don’t like failing, so it’s not going to happen.

Niko McCarty (1:25:45) Awesome. Kate, thank you so much.

Kate Adamala (1:25:47) Thank you.

When we paused Asimov Press six months ago, our goal was to assess both what we had built and the role that writers can play in shaping scientific progress. After doing so, we came to believe more than ever that words are the best starting point to manifest ideas in the world. A well-constructed story acts as a “force multiplier”: it raises awareness for an idea while orienting the field around bottlenecks.

Writing can be uniquely catalytic. Brian Wang wrote a piece for the press about broad-spectrum prophylactics and then became a program director at ARIA, an R&D agency in the UK, where he launched a $50 million program to build them; Wendi Yan wrote about the discovery of artemisinin, the first antimalarial drug, which became the impetus for launching the Research Revival Fund; and Liyam Chitayat argued that mitochondria are alive in an essay that reached millions of readers, growing interest in developing new tools to engineer them. Making these ideas intelligible in print was, for each of them, one of the first steps toward advancing it elsewhere.

This is why we’re excited to share that Asimov Press is returning with a renewed focus. Our goal is to use the journal to grow ideas in the real world. Rather than solely publish pieces in the hope they are seen and spur change, we are moving to a thesis-driven model for scientists and funders by working directly with Renaissance Philanthropy. We’ll still work with authors to sharpen their ideas, but will also follow up with microgrants, workshops, dinners, or assistance with fundraising as needed.

Starting soon, original articles will appear in your email inbox. (If you received this message, you’re already subscribed. We recommend updating your Substack preferences to get only the types of emails you want.) We’ll be hosting a launch party in San Francisco on September 17. A limited number of seats are available, and you can RSVP here.

Subscribe now

Our forthcoming articles will emphasize deep reporting. Every reported piece we publish will be an original investigation or detail how a specific problem in biotechnology should be tackled. Essays might explain how a breakthrough was made and what bottlenecks are stymying progress. Each will be fact-checked and copy-edited, as always, and we will issue corrections when, inevitably, we get something wrong.

We’re also launching two podcasts and will host regular events such as live lectures and intimate dinners on narrow issues, from clinical trial reform to technologies for saving coral reefs. We will continue to print beautiful books, including anthologies and originals. (Our first original book, a coffee-table-sized hardcover on the history of the research laboratory, is currently being printed in Italy.)

The team is expanding to meet this ambition. Brady Huggett, who spent fifteen years as an editor at Nature Biotechnology and most recently was features editor at The Transmitter, a neuroscience magazine, is joining as editor-in-chief. Brady previously created the award-winning “First Rounders” podcast for Nature Biotechnology and also created “Hope Lies in Dreams,” a multi-part series about the history of antisense technology and the making of the first drug for spinal muscular atrophy.

Niko McCarty will continue his role as founding editor, while also joining Renaissance Philanthropy as a fellow. There, he will help biotechnologists develop and launch time-bound, thesis-driven philanthropic funds based on articles published in Asimov Press.

We’re at a strange and interesting moment in history. Interest in biotechnology is surging, but people can’t agree whether we’ll soon cure all diseases or build uniquely dangerous pathogens. With serious reporting that goes beyond the daily discourse, we think we can shape how science improves our future.

We hope you’ll join us! We’re actively seeking writers and artists and would welcome pitches on efforts to cure or prevent a disease, make life multiplanetary, improve animal welfare, augment biosecurity safeguards, make plant engineering more efficient, and just about anything else. Please send pitches to editors@asimov.com.

Asimov Press is the most important thing we've ever worked on, and something we'd like to keep building for decades. If you’d like to support our work as a donor or advertiser, please get in touch. Thank you to Asimov, Renaissance Philanthropy, Edison Scientific, Fulcrum, and Likith Govindaiah for making this relaunch possible.

This essay will appear in our forthcoming book, “Making the Modern Laboratory.”

Published research papers are far from literal accounts of the process of scientific discovery. In contemporary scientific practice, once publishable results are obtained, the actual path taken to reach them becomes more or less irrelevant. Dead ends and false trails are omitted, and out of the messy process of raw research emerges a coherent narrative following clean, linear lines of argument.

But in the space between the hands-on, physical reality of experimental science and the structured narratives fit for printed journals, sits a special genre of scientific writing: lab notebooks. They are the closest witness to “science in the making” (short of live video recordings, which only became available at scale recently).

Historically, scientists recorded ideas and experiments in their lab notebooks with a very restricted audience in mind, sometimes just their colleagues within a research group. For this reason, though some are distinguished by a more literary style and read almost like diaries, most of these records are highly abbreviated and undecipherable to outsiders.

The origins of lab notebooks in experimental science can be traced back to the Renaissance humanist practices1 of copying excerpts from texts to create repositories of proverbs, quotations and miscellaneous facts in personal, thematically organized “commonplace” notebooks. In grammar schools, students were encouraged to develop their notetaking skills, collecting extracts from classical Latin authors. Natural philosophers such as Robert Boyle, John Aubrey, John Ray, and Robert Hooke adopted and repurposed these practices, making meticulous records of their own empirical investigations, while also keeping traditional commonplace books.

The naturalist John Ray’s Collection of English Proverbs (1670) was based on copious notebooks of proverbs extracted from printed catalogues, his own observations of “familiar discourse” and contribution sent to him by “learned and intelligent persons.” Some of the proverbs were accompanied by Ray’s own empirical observations contradicting the proverbs’ claims. For example, the proverb…

If the grass grows in Janiveer, it grows the worse sor’t all year

…is followed by Ray’s qualifier:

There is no general rule without some exception: for in the year 1677 the winter was so mild, that the pastures were very green in January, yet was there scarce ever known a more plentiful crop of hay then the summer following.2

The rise of early modern science was thus deeply influenced by humanist inquiry. Notebooks were used in both traditional and novel ways, as memory aids and as records of information to be communicated later. In Notebooks, English Virtuosi, and Early Modern Science, historian of science Richard Yeo writes:

In the seventeenth century there was a conviction that, to a large extent, copious knowledge could be reliably stored and manipulated in memory. However, during the Scientific Revolution a contrary view was emerging: namely, that the advancement of natural knowledge entailed a reconfiguration of the balance between memory and other ways of storing information. It was accepted that the empirical sciences demanded large quantities of detailed information that needed to be recorded with precision, and kept as durable records to be shared and communicated.3

Isaac Newton’s famous Waste Book,4 currently kept at Cambridge University Library, is a rare example of a physical continuity between the two cultures of notetaking: humanist and scientific.5 It starts out as a commonplace book of excerpted scriptural commentary collected by his stepfather, reverend Barnabas Smith. In 1664, on a visit home in Lincolnshire, Newton found the deceased Smith’s partially used commonplace book and began adding his own prolific and inventive notes on mathematical problems and derivations and sketches of physical experiments.

In contrast to his stepfather, Newton didn’t just collect facts and excerpts: he used them as seeds of his own theoretical explorations. This way, the Waste Book served as an extension of his mind, rather than merely a memory aid, later becoming the foundation of his magnum opus Principia Mathematica. Throughout his life, Newton kept returning to the Waste Book again and again, and the notebook that reached us is quite decrepit from such abundant use.

Newton’s later notebooks, from the 1670s to the 1690s, document his optical investigations in a series of mostly unbound notes.6 These epitomize the gap between private and public research records. Newton seemingly didn’t intend the notebooks to be a lasting record of his experiments as barely any raw data survives, except for some of his late experiments on diffraction. It appears that Newton discarded most of his raw experimental records after completing and writing up each study. In his experiments on thin films, the colors of thick plates, and diffraction, he proceeded from a hypothesis expressed as a mathematical model, to experimental design, to deducing general laws, then back to new drafts.

Newton’s “hypothesis-driven” (his term) experiments on colored circles in thin films are described in his notes under the title “Of y^e coloured circles twixt two contiguous glasses,” likely from 1671. Newton’s rings, as they are now known, are concentric, alternating bright and dark circles formed in the gap between a spherical lens and a flat glass surface, which are caused by the interference of light. Newton first wrote down a series of propositions about the properties of the colored circles, deduced by postulating the existence of hypothetical entities — light corpuscules. The first such proposition on the colored circles reads:

Prop 1. That their areas are in arithmeticall proportion, & soe thicknesse of interjected [film.] Or the spaces rays pass through twixt circle & circ[l]e are in arithm prop[ortion].

He then recorded the measurements of the diameters of the concentric circles and showed that their squares (and therefore, the areas of the circles) increase by a constant quantity — that is, they make up an arithmetic progression, just as stated in the first proposition.

In these and subsequent experiments, Newton made use of averages, a practice almost unheard of in seventeenth century experimental physics, though already in use in astronomy and navigation. The historian of science Richard S. Westfall noted how Newton elevated “quantitative science to a wholly new level of precision … He boldly transported the precision of the heavens into mundane physics … ”

In 1704, Newton published the results of these investigations in his monumental Opticks — in a highly polished form, however, omitting his workings through physical models and the relentless pursuit of precise measurements that populate his research notes. Like Galileo, Newton believed that mathematics was a source of greater certainty than natural philosophy and that natural laws were best expressed in a mathematical language. But his raw experimental data didn’t perfectly align with those laws, even though he managed to achieve remarkably high precision for his time (within 1 to 2 percent). Most intermediate steps of his research thus remain hidden from the readers of Opticks.

Newton also left extended commentary on a famous alchemical text, Introitus apertus ad occlusum regis palatium (An Open Entrance to the Closed Palace of the King).7 This book is attributed to George Starkey,8 a colonial American alchemist who moved from New England to London at age 22 and worked under the tutelage of Robert Boyle. The book is written in a veiled and heavily symbolic language featuring fiery dragons, rabid dogs, and Diana’s doves — traditional alchemical cover-names referring to specific chemical substances. This florid imagery, however, stands in stark contrast to Starkey’s private “chymical” notebooks,9 which are considered models of scholarly clarity. In their laboratories, alchemists seem to have preferred dry recipes with precise annotations, keeping the spectacle and symbolism for public presentation.

Starkey, a Harvard graduate, made extensive use of his scholastic training in documenting his alchemical experiments. Throughout his notebooks, recurring tags mark sequential steps in each experiment: Processus conjecturalis (conjectural process), Conclusio probabilis (probable conclusion), Quaere (search), Observatio (observation), Animadversio (animadversion, criticism), igne refutata (refuted by fire (!), that is, rejected by empirical testing). These are the kinds of annotations that he inherited from the educational culture of early Harvard.

At the center of Starkey’s investigations was, of course, the Philosopher’s Stone, or the Great Bezoar, the legendary and elusive alchemical substance that could turn any “base” metal like lead or copper into a precious one like gold or silver (it additionally was believed to serve as the elixir of life, granting eternal youth and immortality). In his Magnum Opus — in the original alchemical sense of the actual process of creating the Philosopher’s Stone — Starkey worked persistently to achieve higher efficiency in obtaining its supposed precursors, with recurring concerns about the cost of reagents he used. In spirit, Starkey’s work is quite close to modern pharmaceutical and industrial chemistry, and his notebooks attest to his clear-headedness and pragmatism as a practicing (al)chemist.

Two centuries later, historians of science observe at least two distinct lab notebook styles emerging: narrative and numerical. These are best illustrated by the notebooks of two pioneering English physicists, Michael Faraday and James Joule.

Michael Faraday started out as a bookbinder and then worked as a “chemical assistant” and amanuensis to Sir Humphry Davy at the Royal Institution of London.10 After Davy’s death in 1831, Faraday took over as the director of one of the most well-equipped laboratories in Europe, dedicating himself to the study of electromagnetism.

Faraday kept a detailed, narrative lab diary as a series of volumes he bound himself, spanning 42 years — 1820 to 1862.11 They contain records of about 30,000 experiments, both successful and unsuccessful.12 Entries between August 25, 1832, and March 6, 1860, are numbered from 1 to 16041. Each record includes a date, a description of the experimental setup, and results. Helpfully for historians of science, Faraday had the habit of marking with a vertical line the paragraphs in the lab diaries that made it into published papers, often unchanged. To some of the lab notes, he would add his interpretations of the results, new ideas to pursue, or a sign of excitement (as an exclamation mark). For raw ideas and speculations, he kept separate “idea books,” mostly dating from before 1830, at the beginning of his career.

Faraday’s copious notes served as compensation for his famously faulty memory. Even so, he repeated a previously completed experiment that he had apparently forgotten about more than once. (Such amusing occurrences of cryptomnesia are not uncommon among scientists).13

There are some signs that the notebook entries were made with some distance from the immediate action in the lab: Faraday’s handwriting is very neat, there are few corrections, and no chemical stains indicate contact with lab events. As the historian of science H. Otto Sibum puts it, many entries look like “the diary of a Victorian gentleman, written at the conclusion of an exciting day.”14 But Faraday also diligently recorded experimental failures and inaccurate measurements, so the notes appear to reflect the raw reality of his lab investigations.

When his lab diary eventually became too expansive, he added directories and indices to track its contents. This systematic and detailed approach to notetaking could be traced to Faraday’s past engagements as a bookbinder and businessman, as well as to his early quantitative chemical research, all of which required meticulous record keeping.

Though his own lab notes were most likely taken with some delay after execution of the experiments, Faraday encouraged his students to be expedient in notetaking. In one of the earliest experimental manuals for students, Chemical Manipulation, being Instructions to Students in Chemistry, on the Methods of Performing Experiments of Demonstration or of Research, with Accuracy and Success, he writes:

The Laboratory notebook, intended to receive the account of the results of experiments, should always be at hand, as should also pen and ink. All the results worthy of record should be entered at the time the experiments are made, whilst the things themselves are under the eye, and can be re-examined if doubt or difficulty arise. The practice of delaying to note until the end of a train of experiments or to the conclusion of the day, is a bad one, as it then becomes difficult accurately to remember that succession of events. There is a probability also that some important point which may suggest itself during writing, cannot be ascertained by reference to experiment, because of its occurrence to the mind at too late a period.

Faraday was keen on establishing notebooks as a consistent and reliable research practice but, alas, his manual didn’t reach a wide audience at the time. It took a long time before lab notebook practices were standardized.

In contrast, another English physicist, James Prescott Joule, practiced a more quantitatively-oriented, or numerical, way of keeping research notebooks. His major contributions to physics include the mechanical theory of heat and the heat effects of electricity (the SI unit of work, Joule, is named after him).15 Unlike Faraday’s, Joule’s lab notebook entries (from 1843–1858, and over 400 pages in total) seem to have been created in real time as he was taking measurements. They are very terse, containing mostly numerical records and calculations, with little commentary on experimental design. The metadata in each entry usually includes only the date, weather conditions, and a brief description of the experiment’s purpose.

Curiously, Joule had prior experience of working in the brewing business, and his biographers suspect that the accounting skills needed to run such a business and ensure quality control might have shaped his habit of numerical record-keeping in later scientific experiments. Indeed, Joule’s notebooks bear a striking resemblance to brewers’ excise books:

Before putting any water upon his malt for brewing, the brewer is to enter in an excise book or paper, the date of such entry, the quantity of malt intended to be used, and the date of the brewing … 16

That Joule’s notebooks remain mostly silent about the details of measurements suggests he kept them for himself alone. Still, his style wasn’t completely idiosyncratic but indicative of a broader methodological change unfolding in scientific practice at the time.

The nineteenth century saw a tangible improvement in the precision of scientific measurements, and a corresponding shift in judgement where dry numbers came to be trusted more than subjective, narrative descriptions of fallible, all-too-human scientists. Likewise, with the rise of “mechanical objectivity,” photographic images started displacing artistic drawings as illustrations for scientific texts. Some scientists, like the French physiologist and chronophotographer Étienne-Jules Marey, went so far as to declare images to be “the language of the phenomena themselves” and advocated for replacing language with photographs and polygraphs in scientific texts.17

The newly invented instruments like kymographs (tracing spatial position in time) and barographs (tracing atmospheric pressure readings) recorded their own data by generating paper traces as a new type of lab documentation. The lab notebooks increasingly became a place to index and annotate instrument-generated records, along with tabular data and more standardized forms of experiment annotation.

Another shift took place in lab organization, marked by a growth in both lab size and the complexity of coordinated lab operations. The scientific career of the Russian physiologist Ivan Pavlov is an illustration of how lab notetaking practices evolved in response to these changes.18

Pavlov enjoyed a long and prolific life in science. In the 1870s and 80s, he worked in the physiology of digestion and blood circulation, defending his doctoral dissertation on the nerves of the heart in 1883. Next he switched his research focus to digestive physiology, where his work on conditional reflexes (now known as Pavlovian conditioning) in dogs earned him the Nobel Prize in Physiology and Medicine in 1904.

He started his scientific career as a “workshop physiologist,” an independent investigator working in labs at the Veterinary Institute in Saint-Petersburg and later in Breslau (now Wrocław in Poland). During this time, Pavlov designed and conducted his own experiments and analyzed and wrote up his investigations. His lab notebook from this period is a large, thick volume, written in his own hand and reflecting his lab activities: experimental protocols, comments, sketches, and first drafts of research articles.

But in 1891, when Pavlov was appointed as the director of the Physiology Division of the newly established Institute of Experimental Medicine in Saint-Petersburg, he became a “factory physiologist” — the head of a large, hierarchically organized lab.19 He was now in charge of many lab assistants and students (“pracititioners”) whom he regarded as his “skilled hands,” almost like extensions of his own body, conducting and keeping records of experiments in pursuit of his own research agenda. Each practitioner was assigned a research question and a subject dog to experiment on.

As lab head, Pavlov introduced stringent notetaking protocols, with each lab notebook following the fate of a particular dog’s surgeries and treatments. The lab notebooks remained in the lab, where Pavlov could always access them. He instructed his practitioners to record procedure descriptions, notes on the dogs’ behavior, and quantitative data. Practitioners were to abstain from adding their own interpretations, leaving that task to Pavlov himself. He would communicate his analysis of experimental data in lab meetings and more casual conversations with colleagues and, eventually, in published articles.

Though Pavlov didn’t maintain his own notebook, entirely relying on his prodigious memory (and the lab notes of his students), in later years he started to delegate some of his thinking to pocket calendar books repurposed as personal notebooks. His archives contain five such notebooks, dating from 1909 to 1918 and from the late 1920s and 30s, when his research interests shifted to higher nervous activity (that is, the activity of the central nervous system). In addition to addresses, reminders, political comments, and philosophical musings, these eclectic notebooks contain notes on research happenings in his lab, ideas for new experiments, and outlines of articles:

“How will reflexes of time change under the influence of exciter substances: caffeine and so forth?”

“An interesting episode with Kal’m, that impudent and aggressive dog.”

“We consider all so-called psychic activity to be a function of the brain mass, of a defined mechanism, that is, of an object conceived spatially. But how can one place in this mechanism an activity that is conceived psychologically, that is, non-spatially [?]”

“I do not know what exactly we have done, in what way we have broken through, but it is clear to me that there now exists a union of thought, a mixing and unification of the ideas of all participants in the intellectual work [of the laboratory].”

“Some thoughts and dreams about the current war [World War I]: And the example of Germany and England in this war shows that the idea of a world government is not a true resolution of the land question, but rather a human weakness, originating, so to speak, from the inertia of human nature.”

Pavlov’s was one of the large, almost factory-style laboratories that revolutionized the social and material conditions of scientific research from the late nineteenth century onward. Similar in ambition and scale were those led by the chemist Justus von Liebig, microbiologists Robert Koch and Louis Pasteur, and immunologist Paul Ehrlich. These labs were expensive to maintain, had a purpose-designed workspace, clear division of labor, and an additional layer of lab management, which, among other things, took care of the reliable research record keeping in the lab.

In the twentieth century, scientific institutions continued scaling up, as did the pressure for standardization and reproducibility in science communications, including lab notekeeping.20 With the onset of the digital era, scientific data started moving from physical to digital formats that required large memory storage. Electronic lab notebooks (ELNs) emerged to address these changes, yet their history, in fact, goes back much farther than one might think.

One of the first published records of using computers for lab notekeeping was a 1958 paper titled “An Electronic Computer as a Research Assistant.”21 It lists several applications for using computers in lab work: mathematical calculations, copying and storage of large volumes of data, and data analysis and interpretation. These were tasks that entailed “computation volume or complexity, which otherwise would have meant thousands of man-hours for calculation.” The article also mentions “routine report preparation” by computers based on paper-based lab records. Lab notebooks thus evolved from paper notebooks to computer-assisted report generation, followed by digitized laboratory databases and, finally, ELNs themselves.

In the 1980s, a chemistry professor at Virginia Polytechnic Institute, Dr. Raymond Dessy, started advocating for the development of ELNs. In 1985, RS/1, a version of an ELN repurposed from a data analysis and statistical software system, was developed by BBN (Bolt, Beranek and Newman, of the ARPANET fame).22 Dessy later created another ELN prototype from scratch in 1994.

Interestingly, ELNs were first enthusiastically welcomed and adopted by the pharmaceutical industry, whereas their acceptance by academic communities took much longer. By 1997, several pharmaceutical and chemical companies supported a new consortium called Collaborative Electronic Notebook Systems Association (CENSA) which worked with scientific software and hardware vendors to assist with the development of ELNs that met the scientific and regulatory needs of the member organizations.

The University of Oregon introduced one of the first web-based ELNs, Virtual Notebook Environment (ViNE), in 1998. By the 2010s, a range of universities started offering institutional ELN subscriptions, but academic adoption as a whole is still patchy. This state of affairs, however, is likely to change in response to the 2024 NIH IRP Electronic Lab Notebook Policy which mandates researchers to “use only electronic resources to document new and ongoing research.”

ELNs facilitate lab record-keeping by enabling version control, timestamping, search function, hierarchical organization of information, the ability to point to external databases and to manipulate diverse data types (numerical, images, and sequences, among others). One may ask, then, why academia has been so reluctant to adopt them.

Besides the general friction towards adopting a new technology, it could be argued that handwriting is more flexible than typing and more conducive to thinking as a result: one can write how and wherever one wants and draw diagrams and sketches alongside it. ELN templates are more rigid and only allow linear text (though it can be richly formatted). The freedom of a blank sheet of paper cannot be surpassed by the already structured space of an empty digital template. Indeed, drawing and writing have historically remained as valuable, and perhaps indispensable, research techniques in their own right.

When pressing for the adoption of ELNs, the emphasis is on standardization, reproducibility, and regulatory compliance — concepts far from lab notebooks’ original use as a space for working through research questions. Perhaps paper notebooks will remain as equivalents of the waste book used by the bookkeepers of yore, while ELNs will serve as ledgers where final, more organized notes on experimental procedures will be recorded.

Ulkar Aghayeva is a science writer and a columnist at Asimov Press. She also writes about science history on her blog Measure for Measure and about music history and cognition on The Bass Line.

Cite: Aghayeva, U. “A Brief History of Lab Notebooks.” Asimov Press (2026). DOI: 10.62211/52wg-76ye

This essay will appear in our forthcoming book, “Making the Modern Laboratory.”

When protein sequencing was invented in the late 1950s, biologists found themselves faced with the enormous task of managing and analyzing long strings of apparently random amino acids. The difficulty was that most humans can only remember a string of random values around seven items long, which is 67 times shorter than the average protein sequence. Fortunately, scientists’ growing need for sequencing coincided with the development of computers, which helped to make sense of the overwhelming influx of data.

Although the term “bioinformatics” was coined in 1970 by Dutch theoretical biologists Paulien Hogeweg and Ben Hesper, the first bona fide bioinformatician was a quantum chemist, Margaret Belle (née Oakley) Dayhoff. Born in 1925 in Philadelphia to Ruth Clark, a high school math teacher, and Kenneth W. Oakley, a small business owner, Margaret was academically gifted and flourished in the sciences, a notable achievement for a woman in her day.

Dayhoff completed her PhD at Columbia University in just three years, receiving her diploma in 1948. She was one of the first to use computers to solve quantum chemistry problems, a process she described in her thesis, “Punched Card Calculation of Resonance Energies.”1 In the decade or so that followed, Dayhoff focused on raising her children, interrupting this professional hiatus only briefly to take a postdoctoral position in computational chemistry from 1957-1959 at the University of Maryland.

When she tried to resume full-time work in 1960, Dayhoff was surprised to have her funding application rejected by the NIH, which cited her “time off.” This led her to work with an acquaintance of her husband, Robert Ledley, a biophysicist who had just established the National Biomedical Research Foundation (NBRF), an organization in Silver Spring, Maryland dedicated to promoting the use of computers in biomedical research.

Ledley was an early believer in “computerizing” biology and medicine. He explained his motivation as a hunch that computers would one day be “analogous to the staff of a laboratory” and that each program would have “a function to perform just as a laboratory has people with each a job to perform: cleaning people, technicians, senior research workers, a librarian, a machinist, etc. The programmer and the protein chemist have been upgraded to the chief of the computer staff.” Though we take this idea for granted, it was unusual in Ledley’s day. Many biologists at the time rejected computers, and some were outright hostile. Dayhoff later recalled one biochemist who refused to share data or collaborate, stating “I am not a theorizer.”

Still, Ledley and Dayhoff pursued this opportunity to merge biology and computation. Combining their expertise, the two endeavored to create a computer program that could assemble full protein sequences based on results from the Edman Degradation reaction, a sequencing technique developed in 1950. The Edman Degradation reaction was limited in that it could only be used to sequence strings of about 60 amino acids before the reaction stalled out. To handle longer proteins, these shorter strings were broken up and sequenced as a set of overlapping peptide fragments. Assembling the full sequence required a researcher to compare the fragments one by one to piece them together.

Dayhoff’s program was published in 1962. Named COMPROTEIN, it freed scientists from the drudgery of protein alignment. When tested against ribonuclease, a sequence that had taken scientists months to solve manually, the program correctly arrived at the sequence in a matter of minutes.

The ability to automate Edman Degradation reactions with programs like COMPROTEIN led to a rapid leap in the number of sequenced proteins. In 1965, Dayhoff, Richard Eck, and colleagues published the Atlas of Protein Sequence and Structure, the first ever biological sequence database. Notably, the Atlas also contained the first published use of the single-letter amino acid abbreviations (Tryptophan: Y, Glycine: G, Lysine: K, etc.), still employed today.2 By the end of the decade, the collection contained about 1,000 full protein sequences.

These protein sequences allowed scientists to interrogate evolution and phylogeny like never before. While building phylogenetic trees, a branching graph that organizes creatures by how closely related they are, dates back to Charles Darwin himself, access to molecular data and computation allowed scientists to compare organisms based on sequence data rather than observation alone. The Atlas data led to the fundamental realization that sequence similarity was proportional to evolutionary relatedness: as organisms accrue mutations, they separate into distinct species. Closely related organisms will share more protein sequences, having had less time to drift apart from one another genetically.

The first sequence-based phylogenetic trees relied upon closely related sequences that were easy to align and compare visually. In 1966, Dayhoff and Eck published one of the first computer-deduced phylogenies from molecular sequences for a highly conserved iron-sulfur protein called ferredoxin. Dayhoff presented the phylogenetic tree to the public in Scientific American in 1969, writing, “Each protein sequence that is established, each evolutionary mechanism that is illuminated, each major innovation in phylogenetic history that is revealed will improve our understanding of the history of life.”

However, this strategy quickly proved insufficient for more distantly related sequences. This pushed scientists to develop multiple sequence alignment (MSA) algorithms that could compare three or more sequences and handle differences in sequence length arising from mutations like insertions or deletions. Their challenge was to create functional algorithms that were memory and time efficient for the hardware of the era. After alignment, they also needed to find a way to calculate sequence similarity to reflect how these mutations correlated to evolutionary relatedness.

In 1970, Saul Needleman and Christian Wunsch presented a dynamic programming approach that could align two sequences of different lengths. They represented these differences by using gaps to shift the frame of each sequence to accommodate insertions or deletions and maximize the number of amino acids that matched one another. However, their algorithm was computationally too slow to apply to multiple sequences.3

Dayhoff and Eck continued to work on this challenge. In 1978, they published a mathematical framework for calculating sequence similarity after the completion of alignments, a method conceptually similar to the algorithms used today. This “similarity score” was more biologically relevant than previous attempts, capturing evolutionary distance by the number of mutations between two sequences. And finally, in 1987, Da-Fei Feng and Russell F. Doolittle published the first truly practical approach to MSA. They used a “progressive sequence alignment” that initially performed a Needleman-Wunsch alignment for all sequence pairs, then extracted similarity scores for each alignment and built a phylogenetic tree based on the comparison of those scores.4

Even during the early days of bioinformatics, Dayhoff and contemporaries realized that these programs could one day be applied to genes once DNA sequencing was developed. Ironically, the man who would finally unlock ribonucleic acid sequencing, Frederick Sanger, was self-admittedly “rather reluctant” to adopt computers. For Sanger, sequence data was an enjoyable puzzle, and employing computers might “take [away] some of the pleasure” that he got from “looking through the sequences and seeing what could be made of them.”

By the 1970s, though, Sanger had accepted that computation was necessary to handle increasingly lengthy sequences. As his group at the Laboratory of Molecular Biology in Cambridge pursued the sequence of the ΦX174 virus, data was split among nine different researchers’ lab notebooks. A British-Canadian biochemist, Michael Smith, was tasked with collecting and organizing these sequences. Since most of Sanger’s group lacked computational experience, Smith recruited his brother-in-law, Duncan McCallum, who routinely used computers to process administrative data in the management division of the chemical multinational Ciba-Geigy. The pair wrote the first programs to analyze DNA sequence data. Their programs were also used to compile the first full DNA genome sequence, ΦX174, for Sanger’s seminal 1977 publication, considered the beginning of the DNA sequencing era.

When Smith returned to British Columbia, Sanger turned to LMB colleague Rodger Staden to help continue the computational work. Staden, a mathematical physicist by training, expanded and adapted McCallum and Smith’s original programs into the “Staden Package,” able to run on minicomputers rather than mainframes. Staden and colleagues continued to develop this package until 2005, and their software is still available to download.

Our DNA and protein sequence databases grew alongside our ability to manipulate genes with techniques like the polymerase chain reaction and restriction enzyme cloning. As these developments required easier ways to share sequences and manage cloning experiments, it was fortunate that the internet emerged at just this time, in 1969. The first wave of ready-to-use microcomputers followed less than a decade later, released to the public in 1977.

Biologists were among the first to use the internet to share information. Soon, several online databases were established, including the Protein Data Bank (PDB) in 1971, and the Protein Information Resource (PIR) in 1984, led by none other than Dayhoff herself at the NBRF. (Regrettably, she passed away shortly before the project’s completion.)

Also in 1984, the University of Wisconsin Genetics Computer Group published the eponymous “GCG” or “Wisconsin Package,” which initially contained 33 command-line tools for the manipulation of DNA, RNA, and protein sequences. While the Staden Package had focused on DNA assembly, the Wisconsin Package included programs for a variety of tasks, like sequence alignment, identifying protein-coding regions in DNA (called “open reading frames,” or ORFs), translating DNA to the corresponding protein sequence, and finding restriction enzyme cut sites for cloning.

Bioinformatics software continued to evolve steadily over the next 40 years, though this stage was mostly defined by improvements in speed, scale, and user experience. Commercial entities, like Geneious, SnapGene, and Benchling, began to offer programs that dramatically reduced the computer expertise needed for simple bioinformatics tasks like sequence alignment and DNA manipulation, core skills for the modern molecular biologist. However, the field has also welcomed numerous public and open-source programs, such as BLAST and BioPython.

It is hard to believe that in under 70 years we have gone from a handful of painstakingly gathered protein sequences to databases like Genbank, RefSeq, UniProt, and PDB that boast billions of unique sequences. These structural datasets, gathered by hundreds of scientists over decades, were crucial for the development of AlphaFold, an AI model for predicting 3D structure from amino acid sequences that won the 2024 Nobel Prize in Chemistry. Much as the Atlas helped to unlock a deeper understanding of phylogeny and evolution in 1965, these sequences are ushering in a new era of bioinformatics driven by machine learning.

Subscribe now

Ella Watkins-Dulaney holds a PhD in bioengineering from the California Institute of Technology. She is now a sci-comm freelancer and the Art Director for Asimov Press.

Cite: Watkins-Dulaney, E. “A Brief History of Bioinformatics Software.” Asimov Press (2026). DOI: 10.62211/72hw-48jh

To determine which is faster, natural or artificial selection, one might select examples to compare: while wolves and wild jackals branched off from a common ancestor 3.5 million years ago (natural selection), arctic foxes were domesticated within a couple of decades (artificial selection). The oldest moth species appeared in the fossil record around 200 million years ago (natural selection), while the peppered moth changed from a light-colored to a primarily dark-colored one in just 47 years (artificial selection).

But this becomes harder when we consider questions like: Is COVID well-adapted to human hosts? What animal or plant has had as much time to adapt to a new environment as COVID has to humans? Is the persistence of dominant genetic disorders best explained by a lack of purification time (i.e. that humans haven’t lived in their current environment long enough to purify out harmful alleles that might have once been adaptive) or something else?

Very quickly, two major issues arise. First, “years” seems an inaccurate unit for expressing evolutionary rates because evolution considers changes between one generation and the next. In absolute terms, dog, moth, and COVID-19 generations occur over different spans of time; yet from the perspective of evolution, all three should probably be treated similarly.

A second important consideration is that “the time it takes to evolve [X]” is a parameter with large, inherent error bars. It’s difficult to measure rates of evolutionary change accurately, and biological evolution is an inherently stochastic process, contingent on events like a cell accidentally skipping a base in DNA replication or a tortoise, swept away in a storm, happening to land on an island full of edible ferns. Without such random events, an evolutionary process that took a million years might have taken half or twice as long.1

Still, we can get a feel for how long different kinds of evolutionary events take. Let’s standardize generation time and imagine one generation happens per second. Choose an organism and start a stopwatch. Each second, replace the individual in your head with its offspring. Imagine that offspring a tad different from its parents. Do this for thirty seconds, then a minute. How does that span feel? Can you picture evolution happening?

Let’s calibrate this clock against human history:

• At one generation2 per second, the industrial revolution began ten seconds ago. There were fewer than a billion humans, the vast majority of whom were farmers.

• If we go back seven minutes further, humans have just begun implementing large-scale agriculture.

• Go back half an hour and behaviorally modern humans appeared. They possessed paints, composite tools, art, burial, complex hunting and fishing methods, and other common features of modern “primitive” civilizations.

Now let’s look at evolution itself, starting with some familiar examples and expanding to cover much of the eukaryotic tree of life. Then we’ll contrast these with examples of artificial selection to get a feel for their relative speeds.

Natural Selection

Human Evolution

• Anatomically modern humans appeared about three hours ago. These are hominids who are essentially physiologically indistinguishable from contemporary humans.

• Ancestors of Homo sapiens diverged from the other great apes (specifically, from the ancestors of modern gorillas and chimpanzees) some three and a half days ago. These were still primarily quadripedal apes.

• The dinosaurs died out and mammalian diversity began to explode about one hundred days ago.3

Human-Pathogenic Viruses

Using the time from initial infection of a host to the successful propagation to the next host as “one generation”:4

Flu

• Human influenza viruses experience approximately 19 minutes of evolution each season.

• The influenza virus responsible for the devastating 1918 epidemic that killed some 1-5% of the world’s population still exists, having evolved over some 33 hours to become today’s H1N1 seasonal flu virus.

SARS-CoV-2

Under some simplifying assumptions5 we can estimate the time it took to evolve the major branches of SARS-CoV-2:

• The Alpha strain was first detected 42 seconds after the first reported COVID-19 case. It peaked as a proportion of global cases at 72 seconds, then was gradually replaced and effectively went extinct after about 110 seconds.

• Delta appeared about 66 seconds into the pandemic. It was a parallel lineage, not related to Alpha, which became the dominant strain worldwide after 2 minutes and rendered Alpha functionally extinct by the 2.25 minute mark.

• The ancestor of all Omicron lineages probably first appeared 67 seconds into the pandemic. It was not derived from either Alpha or Delta, and wasn’t particularly successful itself, but it did give rise to both Omicron variant 21K and Omicron variant 21L (both first detected at 3 minutes). 21K displaced Delta and peaked at about 3.15 minutes and was displaced in turn by 21L another 10 seconds later.

• In total, SARS-CoV-2 has existed in humans for approximately 10 minutes.

Evolution of Major Clades

• The time to go from single-celled organisms to multi-cellular organisms is tough to estimate — single-celled organisms have generation times as fast as tens of minutes or as slow as decades. Assuming the most historically-important single-celled organisms were like today’s marine phytoplankton,6 multicellularity took 26,000 years to first evolve.

• To understand the evolutionary age of the vertebrates, we might use generation times taken from one of the oldest extant vertebrate lineages — sharks. One study of shark reproductive parameters found average generation times for 18 shark species ranging from 3 years to 42 years, which implies a metaphoric evolutionary age of 140 days to 5.5 years.

• Fast-maturing plants, reproducing annually, have been around for 15 years of evolution. Flowering grasses and other annuals have been around for 4.4 years, while typical flowering trees have evolved for only 1.5 years.7

• What about fungi? The fastest-growing eukaryotes known to science (as of 2009) is an ascomycetous yeast called Kluyveromyces marxianus, which grows a bit faster than one generation per non-metaphorical hour. This sets an upper bound of the metaphorical age of all fungi at about 375,000 years (fungi are quite ancient). Other single-celled fungi like yeasts and slime molds have probably evolved for a few years; more familiar mushroom-producing fungi may have “only” been around for 40-odd years.

Artificial Selection

So far we’ve only covered examples of natural selection. But anything that selectively influences which organisms do and don’t reproduce can force an evolutionary change. Humans do this all the time by selectively breeding animals and plants for traits we like.

How does the speed of artificial selection compare to that of natural selection? We can start with a classic textbook example of evolution that bridges the two: the evolution of the peppered moth under unintentional selection pressures caused by the industrial revolution.

Before the industrial revolution, peppered moths living in Manchester, England were light-colored moths with a dark, splotchy pattern perfect for blending into forests. During the industrial revolution, coal soot and sulfur dioxide emissions from the city blackened the trunks of those trees, making peppered moths stand out. Under intense bird predation, the peppered moth switched from dark splotches on a light background to a totally dark coloration that allowed them to hide on soot-blackened trees. This rapid evolution took about 47 seconds.

Agriculture

The longest-running experiment in artificial selection is the domestication of crop plants.8 Crops are a particularly rich example of evolution because domesticated plants tend to pick up the same cluster of traits — non-shattering seed hulls, enlarged seed or fruit, shorter height, synchronized flowering — so the collection of all agricultural crops can be viewed as replicates in a single giant experiment in artificial selection.

To get a sense of how crop domestication works, let’s look at the recent evolutionary history of a crop for which we have particularly complete evidence: barley. Barley was among the most utilized crops in early agricultural old-world societies, grown across nearly all of Eurasia and North Africa. Critically, it is one for which we have an abundance of archaeological evidence. It seems to have been an unusually slow crop to cultivate, so the evolutionary times given here for barley are more representative of the relative pace of domestication than of its absolute time.

A brief and vastly simplified9 history of the domestication of barley,10 in our metaphorical time, would be that:

• The oldest evidence of cultivation of phenotypically wild barley appeared approximately 20,000 years ago, and serves as our zero second mark.

• The first recognizably cultivated barley appears at around 2.7 hours, after almost 10,000 generations. These seeds look a lot like wild barley, just with bigger grains.

• The first recognizably cultivated barley slowly spreads throughout the fertile crescent over the next 50 minutes.

• Meanwhile, barley gains a “non-shattering hull” just 7 minutes after first displaying a larger grain. This variety has a more solid seed husk, less prone to spontaneously breaking open and prematurely scattering its seeds before harvest.

• The first 6-row barley appears just 10 minutes after the first non-shattering hulls. Wild barley bears two seeds at each repeating subunit; this variety bears six instead, greatly increasing yield.11 (We are now about 3 hours into domestication).

• A second, independent domestication event occurs in Asia around 3 hours and 12 minutes in.

• The first hulless barley appears in Iran just a few minutes later, at the 3 hour and 15 minute mark. “Hulless” is something of a misnomer as this barley still has a hull. However, it is much, much easier to remove. Both hulless and hulled barley will continue to be used for different purposes up through the present day.

• A “dense 6-row” barley appears at about 4 hours and seven minutes, bearing more seeds per ear.

• At about 4 hours and 20 minutes, Hulless barley begins to appear much more broadly across Europe.

• Barley cultivation continues with shifting balances and mixes of existing varieties until close to the present day, around five and a half hours in.

• In the last minute of barley cultivation, humans have modified barley to grow a shorter stalk (less prone to breaking under heavy seed load) with greater seed-to-vegetation weight ratio, and increased resistance to several important diseases.

Barley domestication teaches us that the process occurs in many steps, spread out over thousands of generations; both wild and cultivated varieties were used virtually unmodified for thousands of generations before undergoing multiple major changes over just a few hundred. New traits take almost as long to spread and take over cultivated populations as they take to arise in the first place. Individual domestication events seem to require on the order of thousands of generations, not hundreds and rarely tens of thousands.

But the real value of agriculture as a case study in evolution is the number of replicates it provides. Ideally, to get the best feel for the timescale of plant evolution under artificial selection, we would re-visualize domestication for each of the thousands of species humans have domesticated as crops.

Luckily for us, Meyer, DuVal, and Jensen have collected evidence of domestication for 203 food crops from around the world into a handy spreadsheet. Limiting ourselves to crops with high-quality evidence, we can see that:

• The median crop becomes domesticated after about 24 minutes of (metaphorical-time) exploitation by humans, while mean domestication time is a bit longer, at 33 minutes.

• The fastest-domesticated crop is the common bean, for which the archaeological evidence for first exploitation and first domestication are indistinguishable (though this may say more about the quality of our evidence than the friendliness of the common bean).

• The most recalcitrant crop to eventually yield to domestication is, ironically enough, barley, at its rather lengthy 2.7 hours between first exploitation and eventual domestication.

• Annuals — crops with a one-year lifecycle — take somewhat longer to domesticate, with a median domestication time of 33 minutes.

• Perennials — crops that live more than one year — domesticate more quickly, with a median domestication time of just over 9 minutes.

We can do a lot with this data. For example, if we split up domesticated crops by their gross morphology,12 it appears that trees (and possibly cactuses and succulents) may take fewer generations to domesticate than grasses, vines, and other herbaceous crops.13

Specifically, on average:

• Trees take about 10.5 minutes to domesticate.

• Cactuses and succulents take 8 minutes to domesticate (but N=2).

• Grasses take a bit over an hour to domesticate.

• Vines take about 40 minutes to domesticate, while herbaceous plants take about 35 minutes to domesticate.

Domesticated Mammals

The evolution of the dog from wolf-like ancestors bridges a gap between crop domestication and pure artificial selection. It began as an unintentional and gradual process, but advances in selective breeding (and in the scientific understanding of evolution itself) led to a recent acceleration in dog “speciation.” Its history is also complex, speculative, and contentious, but we have enough combined genetic and archaeological evidence to broadly estimate that:

• The ancestors of dogs split from modern wolves anywhere from 2.5 hours ago to 50 minutes ago by wolf generations,14 or 2-5.5 hours ago by dog generations.15

• The earliest recognizably domesticated dog remains date to approximately 2 hours ago.

• Most modern dog breeds — especially hunting breeds like greyhounds, schnauzers, terriers, beagles, and bloodhounds — are thought to have split from a common ancestor only 1-2 minutes ago.

We also have much more intentional, directed examples of mammal domestication. In 1952, Soviet geneticist Dmitry Belyayev and his graduate student Lyudmila Trut began a multi-decade experiment to attempt to replicate the domestication of dogs. Instead of wolves, Belyayev chose silver foxes as the base species for domestication, starting with foxes bought from commercial fur farms.

Generation after generation, Belyayev and Trut separated out the young foxes who responded most positively to human handlers, as well as the foxes who responded most aggressively. By breeding the most sociable foxes together, the scientists obtained increasingly friendly, loyal, and loving individuals; conversely, the antisocial-selected line quickly evolved heightened aggression responses.

The domestic fox experiment continues to this day. Some key observations are that while the researchers initially categorized their foxes into three “classes” based on their level of tolerance to approach by humans, foxes soon appeared that actively sought human attention, requiring a fourth “elite” category.

• The first four of these “elite” foxes were born after 6 seconds.

• After 30 seconds, fully half of the foxes were born elite.

• After 42 seconds, almost three quarters of friendly foxes were elite.

• As early as 20 seconds into the experiment, some of the domesticated foxes started to be born with dog-like physiological traits, including floppy ears, piebald coat patterns, shortened or upward-curled tails, and shorter legs.16

• Domestication was also associated with a dramatic decrease in corticosteroid production, which the experimenters hypothesize is responsible for the foxes lack of fear response towards humans. By 30 seconds, the domestic foxes had half the baseline cortisol blood concentration of un-selected controls (a ratio which hasn’t changed much since).

The same experiment has been repeated in other species:

• River otters showed early signs of domestication after one or two generations, but breeding them proved too difficult to sustain and their breeding was cut short.

• Wild Norwegian rats (“laboratory rats”) adapted so quickly over their first 35 seconds of breeding that they maxed out the researchers’ initial behavioral scale and forced them to define a second one.17

• American minks began showing active interest in human handling after 18 seconds of selection.

Evolution in the Lab

The most direct and well-quantified examples of human-induced evolution come from laboratory experiments on single-celled bacteria and protists. Fast-growing, single-celled organisms make amazing test platforms for evolution, due to their ease of genetic manipulation and characterization and incredibly fast reproduction. One test tube, incubated overnight, can easily hold tens of billions of E. coli, representing millions of de novo genomic mutations. On the other hand, bacteria reproduce asexually, without sexual recombination, which makes even huge populations of bacteria less genetically diverse than, say, a small tribe of monkeys.18

With those features in mind, let’s look at how scientists have measured — and forced — evolution in microorganisms. In a famous long-term evolution experiment in the lab of Richard Lenski, twelve parallel lineages of laboratory-bred E. coli have been successively grown in simple, glucose-limiting media for over 35 real-world years and 80,000 generations — a bit over 22 hours of metaphorical time. A few highlights from that experiment, in metaphorical timescales are that:

• The E. coli evolved very quickly at first, then slowed considerably (but never stopped). In the first half-hour of evolution, each of the 12 strains increased their fitness in their new environment by about 40 percent.

• One of the first real surprises of the experiment was the sudden appearance of extreme mutation rates in 6 out of 12 strains. These “hypermutator” strains are estimated to be about 1 percent less fit than their more stable ancestors when they first appear (because they are much more likely to replicate genomes with harmful mutations), but were able to evolve more quickly and therefore were able to outcompete their ancestors in the long term.

• The hypermutators took over their respective populations at about 40 minutes, 45 minutes, 1.4 hours, 2.4 hours, 7.5 hours, and 9.7 hours into the experiment.

• One — and only one — of the twelve E. coli strains evolved the ability to eat citrate from their growth media after about 31,000 generations, or 8.6 (metaphorical) hours of adaptation to lab conditions. This trait was dependent on a set of background mutations on acetate metabolism, which arose somewhere around 7 (metaphorical) hours into the experiment.19

• These first citrate-eaters were pretty messy eaters that spewed carbon-rich metabolites20 as a by-product of citrate metabolism. Within about 15 minutes (900 generations) of their appearance, another non-citrate-metabolizing lineage evolved a greatly-enhanced ability to consume those by-products.

• The citrate-metabolizing and by-product-metabolizing strains co-existed for about 2.7 hours of metaphorical time, until the citrate-eating E. coli got good enough at consuming their own waste products to out-compete the secondary strains and drive them to extinction.

• The E. coli evolved to grow steadily larger over time, roughly doubling in cell volume in the first metaphorical half hour of the experiment. They continued to grow in size over the rest of the experiment, but more slowly, expanding to 2.5x their ancestors’ original volume after about 2.7 hours and 3.5x the ancestral volume after 14 hours.

Other examples of long-term evolution experiments abound, too. In 1998, for example, Boraas, Seale, and Boxhorn forced (or at least strongly encouraged) the single-celled green algae Vulgaris chlorella to evolve to grow in multi-celled clumps by exposing them to heavy predation by a flagellate predator. Under this intense selection for larger, harder-to-eat forms, V. chlorella evolved into eight-celled clumps in 10-20 (metaphorical) seconds.

And in a more recent example, four scientists at the University of Minnesota used an even simpler, more ingenious method to select for multicellularity — they propagated media containing Saccharomyces cerevisiae (baker’s yeast) from the bottoms of test tubes over many generations.21 This artificially selected for yeast growing in many-celled “snowflakes” that could sink quickly.22 In our metaphorical time scale, out of ten parallel replicates of evolved yeast, one population evolved noticeable numbers of such snowflake yeast in about 1.5 minutes, half evolved them by 2.5 minutes, and all ten were “multicellular” in 6-7 minutes. The same experiment has been replicated a number of times on a number of different single-celled eukaryotes, with different rates of evolution:

• 70 seconds for Kluyveromyces lactis (another budding yeast that diverged from S. cerevisiae about 100 million non-metaphorical years ago).

• 2.5 minutes for Schizosaccharomyces pombe (a fission yeast).

• Between 1.5 and 7 minutes for Sphaeroforma arctica (a single-celled protist relatively closely-related to animals).

• In slightly over 5 minutes, one out of twelve replicates of Chlamydomonas rheinhardtii (a single-celled green algae closely-related to plants) evolved multicellular clumping.

• Trichormus variabilis (a cyanobacteria) increased in size (through a combination of larger cell size and clumping) by about 3-fold over 2.5 minutes and >30-fold over 6 minutes.

Feeling Evolutionary Time

We’ve argued in previous articles that metaphor helps with quantitative understanding of the sizes and speeds of molecular biology. But metaphor really becomes useful when considering evolutionary time. Relating the number of generations required for evolutionary events to everyday time scales helps us see patterns in what would otherwise seem like totally incomparable evolutionary facts.

Our examples have shown us that intense artificial selection can drive changes far faster than natural selection. “Natural” domestication of both plants and mammals seems to act on the metaphorical scale of hours (with some exceptions, like trees and cacti!), but intense selective breeding can create dozens of new dog breeds in minutes, or even recapitulate dog-like domestication in foxes in tens of seconds. The intensity of selection is just as, or more, important than the size of the difference produced, when it comes to how long it takes to evolve a change.

We have also seen that bacteria evolve relatively slowly. Half a dozen different single-celled eukaryotic species were capable of evolving primitive multicellularity in a matter of metaphorical minutes; E. coli, in contrast, adapted to experimental conditions more on the order of tens of minutes or hours—even though they could draw on absolutely massive population sizes! Speculatively, this may be because eukaryotes use sex (and other powerful DNA-recombination tricks) to generate lots of genetic diversity even from relatively small populations.

And we’ve learned that evolutionary speeds are trends, not exact rates. Several examples (plant domestication, hypermutators in E. coli, multicellularity in single-celled eukaryotes) reveal that roughly the same change occurs multiple times in different populations or species. When this happens, the speeds of the individual examples vary, but only about 3- to 5-fold around the median speed.

We’ve seen that the speed of evolution is complicated. It depends on the size of the evolving population, the strength and stringency of selection, and how the species reproduces—and how quickly. The “one generation per second” quantitative metaphor can’t eliminate all of that complexity, but it at least puts species with radically different life cycles on an even footing, which helps clarify where and when other factors like population size or replication fidelity come into play.

More importantly, using a single scale of “one second per generation” lets us place concrete evolutionary examples onto a more intuitive reference frame. The next time you learn that a weird animal lived so many millions of years ago, or a bacteria evolved antibiotic resistance in so many days, you can place it on that same common scale — simply divide the total time by the length of one generation, convert that number of seconds into something convenient like minutes or days, and you’ll be able to compare it directly to any of the examples given in this brief introduction.

Hopefully, this technique has given you a better sense of how evolution works.

Subscribe now

Samuel Clamons is a bioinformatics scientist at Illumina, Inc. with a PhD in Bioengineering and training in applied mathematics and computer science. Outside of his day job, he writes science fiction and researches theoretical questions in biology at Asimov Press.

Many thanks to Dr. Scott Biering and Dr. Alistair Russell of UCSD for their consultations on viral reproduction and evolution. Header image by Ella Watkins-Dulaney. Image Credits: M. Garde and Tkgd2007.

Cite: Clamons, S. “Metaphors for Biology: Evolution.” Asimov Press (2026). DOI: 10.62211/29je-83ut

Asimov Press is going on hiatus. While a few more articles will appear over the next month — and our stunning hardcover book, Making the Modern Laboratory, will come out this summer — operations will pause in April. All articles will remain freely accessible online.

As we prepare to pause, I want to celebrate what we’ve accomplished. In June 2023, Asimov Press was just a vague idea. (I remember posting the announcement from my phone in the Amsterdam metro, feeling a deep sense of fear about what we were starting.) A few months later, I brought on Xander Balwit. The two of us were its only editors and only full-time staff. We launched with about 7,000 subscribers, ported over from my personal blog, Codon. Our first article appeared in January 2024.

While I had attended journalism school and thought I knew about publishing, I soon discovered that literally everything — from managing people, to recruiting and paying writers, to fact-checking stories, printing books, and much else — doesn’t play out as it’s taught in school. By trial and error, Xander and I had to learn how to develop an eye for writers with exceptional promise, separate good pitches from vague ones, figure out which parts of a draft were most compelling, and make sure stories were told deeply enough to satisfy scientists, while remaining accessible to all “ambitious” readers.

Devon Balwit served as Copy Editor, helping us to convey these complex ideas in readable prose. In 2025, we hired Ella Watkins-Dulaney as our Art Director, and she promptly improved our header images and style. In the last several months, she has stepped into other roles as well, helping fact-check and compile the latest book. Ulkar Aghayeva composed and recorded the music for our podcasts, and also wrote several excellent articles for our pages. We’re also immensely grateful to the dozens of fact-checkers, artists, proofreaders, and writers who have contributed to maintaining a high bar of excellence at Asimov Press.

We’ve come a long way since our inception. Today, we have about 42,000 subscribers between Substack and our website. The press has published 149 original articles, reaching about half a million readers each month.

These articles have ranged from short to long, and from technical reports to history to speculative fiction. Some personal favorites have been The Penicillin Myth, about how Alexander Fleming’s discovery story may have been partly fabricated, How Nature Became a ‘Prestige’ Journal, What Makes an Experiment Beautiful?, and The Making of a Gene Circuit.

While many of our pieces were widely discussed, a handful made an even more tangible impact in the world. Our two recent essays on clinical trials — Clinic-in-the-Loop and AI Won’t Automatically Accelerate Clinical Trials — have been widely discussed in policy circles, as has the essay on China’s Clinical Trial Boom. The essay Mitochondria Are Alive became a bit of a rallying cry for funders considering new tools to engineer these organelles. An essay on artemisinin’s discovery was cited in discussions that led to the Research Revival Fund, a new nonprofit seeking to restore “neglected, illegible, or prematurely dismissed research.”

The Press has published and sold out of two anthologies. When we first tried our hand at printing books, we assumed we would lose a lot of money as printing is expensive, and around 96 percent of all books sell fewer than 1,000 copies. But in both cases, we sold thousands of copies, recouped our costs, and donated the profit to charities. I’m especially proud of our second book, the first commercially available book published in both print and DNA. It was mentioned in The New Yorker and featured in WIRED magazine. Thanks again to the companies that made it possible.

Our decision to pause is not related to funding. Asimov has supported us for the last two years, and we’ve received generous grants from Astera Institute and Stripe. But new projects have called away Xander and I, making it a good time to recalibrate.

For me, the greatest part of writing and editing is drawing closer to ideas and the people who generate them. Essays are a forcing function toward both. Great writers often find an exciting idea and decide to pursue it, only to confront a gap that nothing on the internet can fill — a question too deep or too specific to answer alone. And so the work begins: emailing and calling people, visiting where they work, hosting dinners, and sometimes flying around the world to understand it. The search for great ideas, and a deep understanding of them, is why I’ll keep writing, and helping others do so, for the rest of my life.

When Asimov Press reemerges, I know that those who lead it will be just as curious about the history and future of science as we have been. Thank you for taking this journey with us.

Until next time,

— Niko

In On Exactitude in Science, the writer Jorge Luis Borges imagines an empire so devoted to cartography that its mapmakers draw a map as large and detailed as the empire itself. “In the Deserts of the West, still today, there are Tattered Ruins of that Map,” Borges writes, “inhabited by Animals and Beggars.” Borges’s map is a parable for knowledge, and one of its lessons is that too much detail can quickly become impractical — a map at that scale would be perfect but useless.

But with today’s AI systems, one might wonder if such a map is so absurd after all. Computers and the Internet have already helped us to digitize much of human knowledge, and AI enables us to scan it quickly and easily. For instance, large language models are trained on trillions of words spanning much of recorded human knowledge. In biology, systems like AlphaFold learn from large databases to predict a protein’s folded structure from its amino acid sequence.

This means that, in some domains, something resembling Borges’s life-sized map has become extremely useful. And given the rate of progress on this front, it may seem like advancing science now simply requires building ever larger and more navigable versions of such AI systems, effectively mapping every field.

A lack of practicality, however, was never the sole flaw of Borges’s map. The deeper problem is that adding detail only gives you more of the same kind of information — more roads, more mountains, more villages — when what you might need is a completely different schematic.

Consider the map of the London Underground. Until 1933, the map plotted stations at geographically accurate locations in London. But this made central London, where most stations clustered, an unreadable tangle, while the outer suburbs, devoid of relevant data, took up most of the space. The draughtsman Harry Beck solved this inefficiency by abandoning geographic accuracy and instead redrawing the network as a circuit diagram of colored lines and evenly spaced stations.

A scientific paradigm can also be thought of as a kind of map, but unlike Beck, scientists do not usually know in advance what their maps will be used for. Instead, new paradigms are driven by the desire to explain complex phenomena with a simple and unified set of principles. Such principles tend to have knock-on implications that stretch far beyond the phenomena that inspired them.

For instance, by the mid-nineteenth century, electricity and magnetism were described by a patchwork of separately discovered laws, each explaining a different phenomenon. The physicist James Clerk Maxwell simplified the field by replacing this patchwork with four short equations. But they also implied the existence of electromagnetic waves that could travel through space, including low-frequency waves no one had yet detected. These waves eventually became the basis for radio.

Current AI, by contrast, is not set up to do this. It excels at prediction within existing frameworks, but paradigm shifts require replacing these with simpler alternatives whose implications haven’t yet been explored. A computational system trained on electromagnetic measurements may have predicted these results perfectly, but would never have found radio.

Seen in that light, even as AI becomes more central to scientific work, we risk falling into what one might call hypernormal science, where we get ever better at prediction within current models, alongside a weakening capacity to ask completely new categories of questions. Much like Borges’s empire of cartographers, we risk confusing more detail for a true understanding of the territory.

To avoid this kind of myopia, we must deliberately build AI that helps us devise new conceptual vocabularies. In other words, we must build visionary machines rather than merely predictive ones.

How Paradigms Work

Before exploring how to build such visionary AI, it helps to look more closely at how paradigm shifts in science actually happen. Science usually progresses by adding facts within an existing paradigm, which functions like a rulebook for a field. But over time, evidence accumulates that an existing paradigm cannot explain, requiring a new one.

One might expect that a new paradigm would immediately replace the old one as it better explains the facts. Instead, they tend to gain preeminence only after becoming useful for new applications.

One example is the development of special relativity. In the late nineteenth century, physicists could describe light with wave equations. Because every familiar wave (like sound or water) seemed to need a material carrier, the scientific consensus was that light must also travel through an invisible medium, dubbed the luminiferous ether. The academic establishment was profoundly attached to this concept; Lord Kelvin, the elder statesman of British physics, even declared the ether was the only thing in physics we could be absolutely certain existed.

Albert A. Michelson and Edward W. Morley reasoned that if the ether existed, the Earth’s motion would create an “ether wind,” making light traveling along that wind move at a slightly different effective speed than light traveling across it. Michelson and Morley sent light along two perpendicular paths and expected that because of this speed difference, one beam would come back slightly later than the other. But their experiment revealed no detectable difference.

This result didn’t immediately convince the academic community to abandon the concept of ether. Many physicists (including Michelson) instead adopted an interim position, that the ether’s effects must be hidden in some way. Most prominently, Hendrik Lorentz proposed that the ether existed, but that objects moving through it would shorten in the direction of travel, canceling the expected signal.

An alternative paradigm was ultimately offered by Albert Einstein, then a 26-year-old patent clerk in Bern, Switzerland. His theory of special relativity posited two principles: that the laws of physics are the same in every uniformly moving frame, and the speed of light in a vacuum is the same for all such observers. Through these principles, Einstein was attempting to introduce “a simple and consistent theory,” by which “the introduction of a ‘luminiferous ether’ will prove to be superfluous.”

Initially, Lorentz’s and Einstein’s theories both explained the known experimental data similarly well. But Einstein’s theory proved far more fruitful over time. If light’s speed were genuinely constant, then space and time could not be absolute. This eventually led to the demonstration that mass and energy had to be equivalent, as per Einstein’s famous equation, E=mc², which now underpins technologies from nuclear power to medical imaging.

A paradigm can take hold even if incomplete, provided its core idea is sufficiently useful. For instance, Charles Darwin’s theory of natural selection offered a single principle that could explain the diversity of living species even though it still lacked an explanation for how traits actually passed from parent to offspring. In the late 1860s, he posited the missing mechanism, the erroneous notion of “pangenesis.” In it, he speculated that every cell in the body sheds tiny particles called “gemmules” that collect in the reproductive organs and transmit traits to offspring. Despite this error, Darwin’s core vision survived, and spread amongst biologists, before genetics supplied the necessary, physical mechanisms.

Thus, Einstein and Darwin were both able to generate simple and elegant theories that could make predictions beyond current evidence, even when some details were missing or wrong. In both cases, their decisive advantage was not technical skill within the paradigm, but rather a willingness to step outside it: Einstein benefited from being an outsider to the academic establishment, as this freed him from attachment to the idea of the ether,1 while Darwin appropriated concepts from Charles Lyell’s geology and resource competition from Thomas Malthus’s economics.

Current AI Training Risks Hypernormal Science

If paradigm shifts require stepping outside the prevailing logic, we should ask whether current AI is set up to do this.

Consider an early attempt. In the late 1970s, computer scientist Douglas Lenat built the Automated Mathematician, a program designed to discover not just new facts but entire mathematical concepts. It would start with simple ideas, combine and vary them, and keep the results that seemed interesting. It appeared to work, reportedly rediscovering prime numbers and Goldbach’s conjecture. But its creativity turned out to be limited, because many of the concepts it “discovered” were already implicit in the way mathematics was written inside the program.

While today’s AI has vastly more power than the Automated Mathematician, a similar constraint applies. Most machine-learning systems are trained by minimizing prediction error against a dataset whose inputs and labels are defined in advance. This makes them very good at predicting current data, but locks them into the conceptual vocabulary of the data they learn from.

Consider medicine before germ theory. In the mid-nineteenth century, doctors thought that illness was caused by noxious air, and kept meticulous records accordingly. The physician William Farr mapped cholera deaths across London and found they correlated strongly with low elevation, which he thought was because noxious vapors accumulated in low-lying areas. He was actually picking up a real signal: low-lying districts were closer to the contaminated Thames River. But because his data was organized around air quality, he could not find the true cause.

An AI trained on Farr’s records could have found even subtler correlations, and would have been genuinely useful for predicting which neighborhoods would be hit hardest in the next outbreak. But it would not be able to derive the concept of a waterborne microorganism, as this was not a variable anyone had yet recorded. It took researchers like Louis Pasteur and Robert Koch, working with microscopes and culture dishes rather than statistical registers, to establish germ theory and open the door to antibiotics, antiseptic surgery, and modern public health.

In 2023, Google DeepMind used a graph neural network called GNoME to predict the stability of crystal structures at an enormous scale, discovering 2.2 million new materials. But the vast majority were substitutions within already-known structure types, for instance swapping one element for a neighboring one on the periodic table. The system optimized impressively for thermodynamic stability relative to known structures, but could not venture far from these.

Some newer AI models, known as foundation models, sidestep this problem partly by learning directly from raw data rather than human-curated labels. For instance, the protein generation model ESM3 was trained on protein sequences and structures, and managed to design a novel fluorescent protein highly different from those found in nature.

While impressive, generating novel proteins in this way is analogous to filling in unexplored spaces on a map, but not to creating new maps entirely. ESM3 does not, for instance, ask whether an amino acid sequence is the right level of description, or whether some other organizing principle might unify protein behavior with phenomena outside biology.

When researchers trained a foundation model on ten million simulated solar systems, it learned to predict planetary orbits with high accuracy, but acquired no representation of gravity. Instead, it had assembled a patchwork of statistical regularities that happened to produce correct trajectories. Of course, it is possible that such models already contain, buried in their weights, patterns that point beyond current theories; but extracting them would require deliberate study in its own right.

This grows in importance as scientists use AI more often in their work. On the far end of this spectrum, researchers have begun to develop “AI scientists,” end-to-end pipelines that aim to execute the whole scientific workflow by chaining together literature search, idea generation, code writing, experiment execution, and paper drafting. If these systems are successful, they might be able to work much faster than human scientists, and thereby may constitute the majority of scientific work that gets done.

But these systems have to evaluate the quality of the new ideas they generate, and it is hard to do so without reference to the existing paradigm. When the system proposes a new hypothesis or experiment, the only available proxy for what constitutes a good idea is consistency with existing science. This often involves passing simulated peer review, aligning with established results, and looking like a plausible contribution to the field. A genuinely novel reframing would likely score poorly on all of these measures, for the same reasons that paradigm-shifting work has always faced resistance from reviewers trained in the paradigm it aims to replace.

Simply put, optimizing performance on extant benchmarks makes it difficult for alternatives to emerge. And this, in turn, risks hypernormal science.

Consider a thought experiment in genetics. For centuries, breeders kept detailed records of which animals were mated, what their offspring looked like, and which traits appeared in which family lines. An AI trained on such data could learn to predict what size, color, or yield the offspring of any given two parents might have, which would surely prove immensely useful.

But such a predictor would never discover the gene as a discrete unit of inheritance, nor DNA as its carrier. Without that insight, while farmers may have been thrilled with their better breeding predictions, we would not have had the ability to create genetically modified organisms or targeted gene therapies. Indeed, a prolonged state of hypernormal science is dangerous precisely because its practical implications would not be immediately obvious.

There are early indications that this is already happening. A recent study of 41 million research papers found that scientists using AI publish more and receive more citations, but collectively, AI-augmented research covers around five percent less topical ground. This appears to be because AI gravitates toward problems rich in existing data, where the current paradigm is most established. The result is that AI induces authors to converge on known solutions, rather than search for new ones.

The Bitter Lesson for Science

In a 2019 essay entitled The Bitter Lesson, the computer scientist Richard Sutton observed that methods which try to build in human knowledge consistently lose out, over time, to methods that simply scale search and learning. “We want AI agents that can discover like we can,” he writes, “not which contain what we have discovered.” This applies naturally to paradigm shifts, which by definition move beyond existing knowledge. It might seem, then, that the path to paradigm-shifting AI is to get out of the way and let computation run.

Indeed, open-ended systems like AlphaZero (which learned chess entirely through self-play) are capable of creating both powerful and highly original results. Starting from nothing but the rules, it played millions of games against itself, generating strategies and discarding the ones that lost. Within a day, it was playing at a superhuman level, and did so while playing extremely original moves. The grandmaster Peter Heine Nielsen compared it to a superior species landing on Earth and showing us how to play chess.

Science might seem like a more complex version of the same problem, requiring more powerful computation to solve. In some senses, though, the opposite is true. Chess has simple rules, but a winning strategy is extraordinarily complex. In science, the winning paradigms can be extraordinarily simple; special relativity only has two postulates. But without the benefit of hindsight, there is no clear way to pick a winning paradigm.

To design AI for disruptive science, we would need to understand what “rules” make one paradigm better than another, and build systems that optimize for these. This turns out to be a harder problem than scaling compute. The answer cannot simply be experimental success, since experiments are slow and do not always reliably distinguish between paradigms (as was the case with Lorentz and Einstein). And there are other plausible candidates, but none yet offer a sufficient formulation.

One rule might be that good paradigms are simple. There are early attempts to make AI optimize for this. For example, in physics, symbolic regression systems such as AI Feynman try to discover the simplest equation that explains the data, instead of doing a black-box mapping. On benchmarks drawn from the Feynman Lectures, the method discovered all 100 test equations, while prior software found only 71. One can even formalize a drive towards simple theories using the Minimum Description Length principle, which effectively penalizes unnecessary complexity.2

But these systems currently work on clean data with pre-selected variables, and only search one dataset at a time. So far, they have been tested on rediscovering equations we already know, but haven’t yet demonstrated that they can find new ones.

Another rule may be that good paradigms draw effective analogies. Feynman devoted a chapter of his Lectures to the observation that heat flow, fluid flow, diffusion, and electrostatics all share the same equations, treating this as a deep fact about nature. This makes intuitive sense: an idea that already works in two domains has a better chance of working in a third.

The most obvious kind of analogy is across disciplines, as when Darwin borrowed the logic of competitive scarcity from economics and applied it to biology. In principle, AI could search for these connections at a scale no individual researcher could match, trawling across fields for ideas that seem structurally similar. Early systems have been built that find functional analogies across large databases of patents and product descriptions.

But some analogies are not between written theories, but rather between an idea and a sensory intuition. At sixteen, Einstein imagined riding alongside a beam of light and asked what he would see. Maxwell’s equations modelled light as a wave. But if this were true, if one travelled at the speed of light, the wave would appear frozen, hanging motionless in space. To Einstein, this picture felt physically wrong; this visceral discomfort ultimately inspired him to develop special relativity.3

Some research programs are trying to ground AI reasoning in physical experience. There are early attempts at multimodal architectures that can jointly process vision, language, and action. Self-driving laboratories couple AI to robotic instruments that manipulate real materials, which could in principle ground abstract reasoning in physical feedback (though most operate within a single experimental domain). But these are early efforts, and the gap between a robotic arm pipetting reagents and multisensory human experience is vast.

In the meantime, the fastest path to getting effective analogies may be leveraging both human and AI abilities together. Humans have breadth across modes — we see, hear, touch, move through space, and read, all at once — which grounds our capacity for structural analogy. AI has depth within modes, having processed far more than any person ever could. If AI can help researchers learn faster across disciplinary boundaries, perhaps by disciplined use of LLMs, that alone could accelerate discovery.

The deeper problem is that we do not have a good formal understanding of how paradigm shifts actually happen. Simplicity and analogy alone do not appear to be a complete description. J.J. Thomson’s “plum pudding” model imagined atoms as spheres of positive charge with electrons scattered through them. It was simple, matched what was known at the time, and was comically analogical, but soon proved completely wrong.

For now, then, the Bitter Lesson for Science may be that scientific acceleration will not happen by default until we understand science itself better.4 But if we can identify the conditions that produce paradigm shifts, we can start to engineer them.

A Paean for Metascience

Understanding and codifying science is no small task. Since Bacon’s Novum Organum in 1620, we have had a reasonable account of how science works in theory — observe, form hypotheses, test, and revise. But paradigm shifts seem to depend just as much on conditions as on method: who gets to do science, what they are rewarded for, and how freely ideas can cross between fields, for instance.

We can learn some things from history. Bell Labs, Xerox PARC, and the early Laboratory of Molecular Biology at Cambridge all produced extraordinary concentrations of paradigm-shifting work, mostly because they were small groups with enough institutional protection to pursue ideas that looked unproductive by conventional measures. There are clear parallels, in fact, to the kind of independent self-play that made AlphaZero successful. But there are limited numbers of historical paradigm shifts to study, and we have barely begun to explore the design space.

Here, AI itself may be able to help. We have never been able to run controlled experiments on scientific institutions; it is impossible to create labs that differ in only one respect and compare the results. But we could run AI agents in parallel populations under different research conditions, and analyze the results in detail. In this sense, AI scientists may give metascience its first model organism.

For instance, one could test how group structure shapes discovery: do small, isolated teams produce more conceptual reorganization than large, well-connected ones? Do flat hierarchies outperform rigid ones? One could run AI agent populations that vary these factors independently and measure the results — something that is impractical to do with real institutions, where size, hierarchy, and communication patterns are all entangled.

Some of the optimism around AI for science rests on the intuition that if we build systems with much stronger general reasoning, discovery will follow as a near-automatic consequence. But there is no guarantee that this will happen by default. Many technologies have promised radical scientific acceleration, and so far, not completely fulfilled their promise.

The Internet, for example, made knowledge much easier to search, and in theory may have led to much faster science. Online collaboration has certainly sometimes enabled better scientific work, such as in the Polymath Project. But this revolution has largely not emerged at scale, mostly because deeper inefficiencies in how we organize science (e.g., career incentives) remain a bottleneck. There is even evidence that online journals actually narrowed what researchers read and cite, because scientists search or follow hyperlinks rather than browsing journals, thus potentially accelerating consensus rather than expanding the space of ideas explored.

AI could repeat this pattern at a larger scale — generating faster results within the existing paradigm, while the structural conditions for disruptive science remain unchanged or worsen. There is no reason to expect this design problem to sort itself out on its own. But if we treat AI for disruptive science as a deliberate research program, we have a better chance of building the capabilities that paradigm shifts require. And to do that, we will have to understand how to design science itself.

Subscribe now

Alvin Djajadikerta is the CEO of Evidentia Labs and a founding researcher at Science Works. He holds a PhD in Molecular Neuroscience from the University of Cambridge.

Cite: Djajadikerta, Alvin. “Designing AI for Disruptive Science.” Asimov Press (2026). DOI: 10.62211/29ej-27et

By Rachel Dutton

The human immune system is, in one sense, a detection mechanism. It has evolved, over millions of years, to scan the body for molecular signals that tell it whether to attack or stand down. Most of these signals come from pathogens, damaged cells, or the body’s own hormones. But in 2019, a lab in Germany published a finding that pointed to a much stranger source: one of the signals sensed by the immune system is found in sauerkraut.

When people eat sauerkraut, a molecule called phenyllactic acid (D-PLA) — found in fermented foods — enters their bloodstream and activates a receptor, known as HCA3, on immune cells, triggering an anti-inflammatory response. In addition to lactic acid, phenyllactic acid is one of many compounds produced by lactic acid bacteria during the fermentation of sauerkraut and related fermented foods. Prior to this study, other molecules had been found to bind HCA3, but D-PLA was a hundredfold more potent than any of them.

This discovery advances our understanding of how fermented foods can reduce inflammation and positively affect human health. But more striking is what it suggests about hominid physiology. Although HCA3 is part of a larger family of receptors broadly conserved across eukaryotes, HCA3 is only present in humans and other great apes like chimpanzees and gorillas — and not even in other mammals. It is a recent addition to the genome, appearing only a few million years ago. Its existence seems to suggest that our immune system evolved to recognize the microbial metabolites from fermented foods.

We tend to think of fermented foods as something humans invented and then chose to eat. But, increasingly, scientific evidence suggests the causality runs the other way. Fermented foods appear to have helped shape human biology itself, and our bodies may have been built, in part, to expect them. The case for this runs from changes in hominid gut anatomy millions of years ago to the HCA3 receptor, to a growing body of research linking fermented food consumption to immune function and gut health. And it raises an uncomfortable question about what happened when the Western food system, in the name of safety and efficiency, quietly removed these foods from our diets in the nineteenth and twentieth centuries.

Early Fermentation

Fermented foods are the result of the controlled growth of communities of microbes. At their core, they are the products of the interaction of these microbes and whatever food they consume, whether cabbage or cucumber. While this process varies from food to food, fermentation typically involves managing environmental variables such as oxygen, temperature, and salinity. In contrast to food preservation methods like canning or even pickling, which are designed to prevent microbial growth, fermentation harnesses the capacity of naturally occurring microbes found on fresh food or in the environment to outcompete spoilage organisms.

Most archaeological evidence from pottery shards suggests that fermented food production is at least 7,000 to 10,000 years old. This timeframe coincides with the major transition to agrarian lifestyles, which would have reliably produced surpluses of food and the subsequent need for preservation methods.1 While this explanation satisfies most scholars, there is reason to believe that fermentation may be far older.

For one, it’s a very simple process to trigger. Some foods even ferment spontaneously. In the case of alcoholic drinks, like beer, wine, or mead, ubiquitous yeast species, which are naturally found on grapes and other fruit skins, rapidly use sugar as a food source for reproduction, producing ethanol as a byproduct. The same results occur when ripe fruit falls to the ground and its sugar is exposed to the environment, or when honey is diluted with water.

Other processes require only minimal intervention. For example, submerging fresh food in liquid or burying it creates a low-oxygen environment that encourages the growth of acid-producing bacteria that preserve the food by what is called “lactic acid fermentation.” This technique produces dill pickles, sauerkraut, and kimchi. While salt is often added as an additional intervention against unwanted microbes, it’s not required as it isn’t the primary driver of the fermentation.

Another argument for an earlier origin for fermented foods is that they are found across nearly all human cultures. While the number of fermented foods in the modern, Western diet is fairly limited (cheese, yogurt, bread, chocolate, coffee, beer, wine, kimchi, and kombucha) hundreds more are eaten around the world, from fermented shark in Greenland to a seemingly limitless variety of fermented soy beans in Asia. This diversity is a testament to how humans gradually mastered this ancient practice and modified it to suit new environments as they moved out of Africa.

Around 14 million years ago, our hominid ancestors were arboreal species whose diet would have been primarily based on fresh fruits picked from the trees they lived in. When ripe fruit fell to the ground and underwent spontaneous fermentation, it would have been toxic to our ancient ancestors due to its high concentration of ethanol. Their bodies as yet had no efficient way to break down ethanol.

But then, about 10 million years ago, a mutation arose in the genome of the common ancestor of humans, gorillas, and chimpanzees. This mutation, a single amino acid change in the enzyme Alcohol Dehydrogenase 4 (ADH4), enabled it to break down and detoxify ethanol with 40x higher efficiency. The capacity to consume this energy-rich but previously dangerous fruit may even have driven our transition from an arboreal lifestyle to a terrestrial one. What’s more, this ability to tolerate ethanol may have been what allowed our ancestors to diversify their diet and survive while lineages without this mutation went extinct.

The fossil record shows that a major shift in hominid anatomy occurred around 2 million years ago, when hominids developed a smaller rib cage and larger skull. At the same time, another major change took place in their intestines. Compared to our closest relatives, humans have a digestive tract that is 40 percent shorter. This decrease was thought to be driven by the external processing of our food, which reduced the time and energy involved in chewing and digesting. Anthropologist Richard Wrangham argues that the technological innovations of controlling fire and cooking food led to this major change, and that the excess energy we got from cooked food, in turn, supported the evolution of a larger brain.

However, two recent studies, by biological anthropologist Katie Amato in 2021 and evolutionary biologist Erin Hecht in 2023, suggest that these anatomical changes may have been driven by human use of fermentation even before humans began to cook. By allowing microbial species to ferment and break down complex carbohydrates and other macromolecules in foods, we may have turned over certain parts of an otherwise energy-intensive digestive process to microbes in a form of “external digestion.” This use of fermentation to pre-digest food, intentional or not, may have served as a predecessor to cooking, providing the extra calories needed to support the evolution of a larger brain.

Another benefit of fermentation is that it offered access to foods which, previously, would have been toxic. As our ancestors came down from the trees and needed new ways to fill their stomachs, the tubers of many plants and grasses offered an appealing, ready source of calories. Tubers contain large deposits of starch. Root vegetables, such as potatoes, yams, and carrots, are our modern-day, highly domesticated equivalents. But the wild tubers of our ancestors’ time were hard to chew, and some contained low levels of toxins. Varieties of cassava, for example, contain compounds that release cyanide when ingested. After just a few days of fermentation, however, microbes destroy these dangerous molecules and make the food safe to eat.2

Ultimately, given the simplicity of the fermentation process, its provision of new food sources, and its role in decreasing the need for extensive chewing and digestion, fermentation could have had an outsized impact on human evolution. It may, in fact, have helped make us human.

Peril and Promise

Perhaps the most striking thing about fermentation is that humans figured out how to control the growth of microbial species long before we understood what “microbes” were.

No one had seen microbial life until the 17th century, when Dutch scientist Antonie Van Leeuwenhoek used a handmade microscope to reveal small, motile forms he described as “animalcules.” It wasn’t until the late 1800s, however, that the French chemist Louis Pasteur demonstrated the role of microbes through his studies of fermented foods. His work on the spoilage of beer and wine formed the basis of the “germ theory” of disease. If microbes could be the causative agents of spoilage in food, Pasteur reasoned, maybe they could also be the causative agents of disease.

For his part, Pasteur was not strictly “anti-microbe.” He found them enthralling, investigating the differences between lactic fermentation (yogurt, pickles, and sauerkraut), butyric fermentation (butter, cheese, and milk), and acetic fermentation (kombucha, sourdough, sour beer), as well as developing the categories “aerobic” and “anaerobic” (which refer to whether microbes require oxygen to survive).

Even so, Pasteur’s work identifying microbes responsible for putrescence and disease has had the greatest impact on our food system. At the beginning of the twentieth century, automated machinery expanded canning capacity from roughly 10 cans a day to 1,000. Variation in the quality of the sealed lids, temperature treatments, and exposure to contaminants through handling meant a greater potential for microbial illnesses like botulism.

After three deadly botulism outbreaks in 1919 linked to California-distributed olives, the canning industry turned to Pasteur-inspired “bacteriologists” to design better canning systems and restore public confidence. While the resulting practices — including steam sterilization, the marking of batches, and traceable can coding — were at first voluntary, the Cannery Act of 1925 mandated statewide compliance.

By the mid-1900s, microbial research into food had come to focus on how to keep organisms out of it. Innovations in heat, pressure, refrigeration and anti-microbial agents to extend shelf life and decrease potential contamination with food-borne pathogens formed the core of academic and industrial research. As our methods of food production shifted, so did our diets. Americans largely moved away from the consumption of traditionally fermented foods, with one major exception.

That exception began with a single lecture. The Russian zoologist, Ilya Mechnikoff, based at the Pasteur Institute in Paris, was well known in the early 1900s for his scientific discoveries involving the immune system, for which he would win the Nobel Prize. But towards the end of his career, Metchnikoff became fascinated with the idea that aging was just another disease awaiting a cure.

In his 1908 book, The Prolongation of Life, Metchnikoff proposed that this disease was caused by "putrefying" microbes that lived within the human gut, or, as he put it, “chronic poisoning from an abundant intestinal flora.” His suggested cure came in an unexpected form — yogurt. Mechnikoff reasoned that just as the acid in sour milk prevented the growth of spoilage organisms, it might limit the growth of “spoilage” microbes in the gut.

After speaking with a student after one lecture, Metchnikoff learned of large numbers of centenarians in Bulgaria (one of whom he credulously reported as being “158” years of age).3 Metchnikoff hypothesized that the consumption of large quantities of yogurt in Bulgaria prevented aging through the action of lactic acid on gut bacteria. After obtaining a sample of the “Bulgarian bacillus” (Lactobacillus delbrückii subsp. bulgaricus) from yogurt, he set out to study the effects on the gut. His lab work showed that the presence of the yogurt microbe and the lactic acid it produced slowed the growth of intestinal microbes.

Metchnikoff thus advised the consumption of yogurt to promote healthy intestinal balance, writing: “A reader who has little knowledge of such matters may be surprised by my recommendation to absorb large quantities of microbes, as the general belief is that microbes are all harmful. This belief, however, is erroneous. There are many useful microbes, amongst which the lactic bacilli have an honourable place.”

Metchnikoff initially presented his yogurt hypothesis at a public lecture in 1904 in Paris, titled “Old Age.” The lecture went the early 1900s equivalent of viral. Newspapers ran stories with headlines like “Drink Sour Milk and Live to be 180 Years Old” (Evansville Courier and Press) and “Sour Milk is Elixir: Secret of Long Life is Discovered by Prof. Metchnikoff” (Chicago Daily Tribune). “Within months of Metchnikoff’s lecture, milk-souring germs had blossomed into an international business. Pharmacies throughout Europe and the United States were offering Bulgarian cultures in the form of tablets, powders, and bouillons — to be consumed as is or used,” writes Luba Vikhanski in her book Immunity. It was, in effect, the first “superfood.”4

“Probiotic” products quickly followed, containing dehydrated “lactic bacilli,” that anyone could use to sour their own milk in the way prescribed by Metchnikoff. But despite interest in yogurt and probiotics as health foods, most 20th century microbiologists remained focused on “bad actors,” the microbes and pathogens involved in disease. However, this changed toward the end of the century with advances in sequencing technology and chemical analysis techniques like mass spectrometry, which allowed them to probe microbial genetics and metabolism more closely and with more nuance.5

Environmental DNA sequencing revealed the diversity of microbial species across habitats, including in and on the human body as well as in our food. Understanding what these microbes are doing, and how, has become one of the challenges of modern microbiology.

To Eat Microbes

In response to this challenge, a lab at Stanford University, led by Erica and Justin Sonnenburg, has been doing seminal work on the human gut microbiome and how species within it function and interact. Recently, the Sonnenburgs have turned their focus on diet-microbiome interactions to fermented foods. A clinical trial, which the Sonnenburgs set up in collaboration with nutrition scientist Christopher Gardner in 2021, offered the clearest evidence to date that fermented foods are essential for gut health. In the trial, 36 healthy adults spent 10 weeks eating a diet high in such foods.

At the end of the study, they showed increases in gut microbiome diversity, which is generally associated with gut health. (The participants had new microbial species in their guts, which did not come from the fermented foods. It seems, rather, that fermented foods somehow make the existing gut ecosystem more receptive to incorporating new strains.) Perhaps an even more impactful outcome was that the researchers also found widespread decreases in inflammatory markers in the blood of these participants.

Given that low levels of inflammation sustained over time, referred to as chronic inflammation, is believed to be implicated in many diseases, the ability to decrease it though simple dietary changes would be welcome. However, the six servings of fermented foods a day consumed by the study participants is far higher than that in most U.S. diets. While yogurt and cheese are commonplace, with an average of 13.8lbs and 42.3lbs consumed per person per year respectively, this represents about a serving of fermented dairy a day. Fermented vegetables also comprise less than a serving per day. The impressive 387 million pounds of sauerkraut Americans consume per year equates to only 1.5lb per person, or about 0.06 servings a day.6 There has been a resurgence in interest in fermented foods, from high-end chefs experimenting with new types of fermentations to home fermenters caught up in the pandemic sourdough craze, but we have a long way to go before consumption reaches levels needed for clinically meaningful outcomes.

Even larger gaps remain in our understanding of how fermented foods actually drive health benefits. Beyond inflammation, fermented food consumption has been associated with a wide range of positive health outcomes, from gut health to mental health. As compelling as the Sonnenburg study is, it doesn’t yet reveal the mechanism(s) by which fermented foods trigger changes in human physiology.

While the 2019 sauerkraut study points to the importance of microbial metabolites like phenyllactic acid, if we co-evolved with fermented foods over millions of years, this example of a metabolite-receptor pairing is likely just the tip of the iceberg. As in many other areas of microbiome research, we need to move beyond correlation to causation. Doing so will require a much deeper understanding of the molecular composition of fermented foods, how they are connected to human biology, and more precise clinical studies.

Scientists are beginning to recognize the potential diversity of bioactive metabolites in fermented foods, but most studies focus on a single food type at a time and use different assays of bioactivity from lab to lab. To be able to effectively map the complex interactions between fermented food microbes, metabolites, and human biology, we need larger, consolidated datasets.

To this end, the Microcosm Foods project has been assembling a first of its kind collection of fermented food data pairing metagenomics (mapping microbes), metabolomics (mapping metabolites), and transcriptomics (mapping changes in human immune cells upon exposure to fermented foods) across over 100 different foods. Open-source, systematically-acquired datasets such as these will help build the scientific foundation needed to untangle these complex interactions.

For most of human history, people harnessed microbial ecosystems to make fermented foods and reaped the benefits without understanding their mechanism. Then, in a period of roughly 100 years, the Western food system replaced them with sterile, shelf-stable alternatives.

Today, we are finally beginning to understand how fermented foods interact with our biology — through receptors like HCA3, through shifts in gut microbial diversity, and through inflammation pathways. Science is catching up to what fermentation has been doing all along.

Rachel Dutton is a microbiologist studying fermented foods, from their use as microbiome models to their impacts on health. She received a PhD in Microbiology and Molecular Genetics from Harvard University, and led academic labs at Harvard and UC San Diego. Rachel is currently a Resident at the Astera Institute and a Fellow in the Big If True Science program at Renaissance Philanthropy.

Credit: L. reuteri image by Anastasiia Dmytriv.

Cite: Dutton, R. “Culture Shift.” Asimov Press (2026). DOI: 10.62211/92yr-34qw

This essay will appear in our forthcoming book, “Making the Modern Laboratory.”

By Donna Vatnick

In the 1960s, David Chambers, a researcher at Deakin University in Australia, instructed teachers to give children a blank sheet of paper and ask them to draw a scientist. Chambers repeated this experiment many times over eleven years, collecting more than 4,800 drawings. The results were surprisingly consistent: white lab coat, glasses, beakers, mysterious machinery, someone saying “eureka!” The study has since been repeated dozens of times. While some details have changed, with beakers replaced by rockets, microscopes by vaccines, or men by women (sometimes), the scientist always wears a white lab coat.

The white lab coat, however, only came to symbolize scientists in the 20th century. Before that, cartoonists satirized chemists by portraying their craft as sorcery whose practitioners wore long dark robes, and painters drew naturalists in waistcoats and breeches against backdrops of plants and landscapes. It was really surgery, more than any other scientific discipline, that gave us the white laboratory coat. Today, scientists don a variety of multicolored, specialized protective equipment to suit the needs of their field, but the fact that children still inextricably link white lab coats to “scientists” says everything about how a simple garment came to exemplify a profession’s public image.

To understand how the white lab coat arose, we have to go back to Victorian England to examine not the scientists, but rather the fashion of that time.

In mid-19th-century England, so-called “gentlemen of science” dressed in dark frock coats. In his 1885 portrait, Louis Pasteur stood in his laboratory, rabies sample in hand, in a black frock coat, waistcoat, and black cravat. Charles Darwin, whose personal home was his “laboratory,” sported a similar style. John Snow, the epidemiologist best known for tracing the source of London’s cholera epidemic, inspected the Broad Street pump while dressed like a banker, in a multipiece suit and tie.

This style was influenced by Prince Albert, Queen Victoria’s consort, who elevated the frock coat into a fitted, double-breasted symbol of class in the 1850s. Though the Prince himself likely never fooled around with specimens and staining chemicals, the darkness of the coats he popularized benefited scientists, who didn’t want the stains of their labors visible. Frock coats were, after all, very tedious to wash in the days before laundry machines.

However, for blood-splattered Victorian surgeons, the situation was much worse; heavy and woolen, the frock coat absorbed just about any fluid. Stories of surgeons making their rounds in bloody black robes reeking of rot and sweat are most likely accurate. The esteemed surgeon Sir Frederick Treves (credited with saving the life of King Edward VII in 1902) reminisced in his 1923 book: “The surgeon operated in a slaughter-house-suggesting frock coat of black cloth. It was stiff with the blood and the filth of years. The more sodden it was, the more forcibly did it bear evidence to the surgeon’s prowess.”

Beyond being filthy, frock coats were also very uncomfortable. The sawing, stitching, and other manipulations of surgery were already messy, laborious work, but operating rooms grew even more “clammy and sodden” after Joseph Lister introduced his carbolic spray antiseptic. Lister himself stripped off his frock and “turned up his shirt sleeves” and “pinned an ordinary unsterilized huckaback towel over his waistcoat (for his own protection, not that of the patient).”

In fact, Lister never advocated for changing the fabric of surgical garments. His writings from 1867 assured his readers that using carbolic spray was enough to keep infection risks low. In any case, illustrations from the bacteriologist Watson Cheyne’s authoritative 1882 textbook on antiseptic surgery showed a group of surgeons working in full formal attire, with their outdoor coats and hats resting on a nearby windowsill.

These early practitioners of science suffered from the expectation that they would, at all times, appear genteel. Even while hard at work, they eschewed light fabrics, protective gear, and aprons in favor of a wardrobe that demarcated their prestige.

The real driver of white lab coats was the hygienist movement. The mid-19th century witnessed dramatic strides in public sanitation. Rivers of free-flowing feces and effluent gave way to managed sewage disposal systems. Public bathhouses opened, such as the 1842 warm fresh-water baths on Frederick Street in Liverpool, England, serving the “vast numbers of workmen” of the 500,000 that populated the borough. The introduction of hand-washing in obstetrics in 1847 began to save patients, with doctors in other fields following suit. Motorized washing machines were invented in 1851. The domestic use of soap in England nearly tripled between 1801 and 1861, and then almost doubled again by 1891, and suddenly, both being and looking clean became possible.1

Ice cream vendors, butchers, and bakers were also adopting white clothing as their professional uniforms to project the image of working under sanitary conditions. White made any breach of cleanliness immediately obvious, another way to reassure their newly cleanliness-obsessed customers of the shops’ safety and security.

By the 1880s, some surgeons were beginning to adopt this aesthetic of cleanliness in their practices, like Robert Lawson Tait, a gynecological surgeon who donned a white apron. Tait used soap and water on patient skin, boiled instruments, applied freshly “laundried” towels around wounds, and had a regimented hand-washing ritual before operating. His surgical attire was a visual signal meant to attract paying clientele.

Other surgeons took this fashion further. When Australian surgeon Alexander MacCormick became “the first man in Sydney to wear a white coat while operating” in the 1880s, his colleagues mocked him: “MacCormick was called the ‘Hokey Pokey Man,’ a reference to the popular confection called hokey-pokey, sold by ice cream vendors,” write historians Susan Hardy and Anthony Corones in Dressed to Heal: The Changing Semiotics of Surgical Dress. The joke backfired. Patients flocked to him, attracted by his gleaming cleanliness.

The white medical garment popularized during this period was everything the frock coat wasn’t: washable, lightweight, cheap, and disposable. Hospitals could send uniforms to commercial laundries with a quick turnaround. Textile mills churned out mass-produced cotton and linen garments after the Civil War. By contrast, frock coats were tailor-made from broadcloth, requiring tremendous time and effort.

Two famous portraits from the American painter Thomas Eakins capture the transition from black to white. “The Gross Clinic” from 1875 shows surgeon Samuel Gross and his assistants in black coats, blood on their hands, with a skylight letting in the noontime sun. The patient is a young man wearing his street socks. When this life-size painting was introduced at the 1876 United States Centennial Exposition in Philadelphia, it was relegated to a remote corner of the exhibit hall after having been initially rejected for display at all. Some scholars suspect that the painting was too “realistic” for a public that now expected visible hygiene.

“The Agnew Clinic,” painted in 1889, presents a stark contrast, with surgeon David Hayes Agnew and colleagues in shining white gowns using sterilized instruments under artificial light. The patient is draped in white sheets on a white table. Here, however, the painting is misleading because Agnew was remarkably resistant to adopting these practices. The public remembered Agnew as one of the several attending physicians to the beloved President James A. Garfield after he was shot in the back at a railroad station in DC in 1881. Agnew had poked and prodded at the President’s bullet wound without ever having washed his hands or sterilizing his probes. As the President developed abscesses, Dr. Agnew would drain them with dirty instruments.

The younger generations of American surgeons watched in horror, but even after this high-profile disaster (with evidence of the infected abscesses revealed when the president succumbed to his injury two months later), Agnew didn’t change his techniques. Just a year before this painting was made, a photograph shows Agnew in a clinic, wearing a nice, buttoned-up street coat just as Samuel Gross had a decade before.2

The new expectations of operating rooms didn’t immediately apply to those of the laboratory. White linens weren’t as practical for bench science as they were for clinical medicine. Industrial chemists, for example, were known to wear brownish — not white — coats and aprons, often made of leather, in German and British labs in the early 1900s. Wanting to hide the stains of her experimentation, famed physicist and chemist Marie Curie requested her wedding dress to be “practical and dark” so that she could afterwards wear it in the lab. This she did, until switching to black dresses upon the death of her husband in 1906.

Curie was not alone in preferring black for cleanliness. French surgeon and biologist Alexis Carrel worked out of a black-walled lab and wore black deliberately because it showed dust. Black also works better for contrast when using white mice and powders, as well as hiding any blood or pigment stains. Even today, some scientists prefer black labwear.

Despite such outliers, white lab coats have become synonymous with lab scientists, even if experts disagree on exactly when the white lab coat made its way into the lab. A 1902 photograph in a neurobiological lab in Berlin, for example, shows researchers wearing long, light colored coats while working at benches. Another from 1922 shows two industrial chemists wearing what appear to be white coats.

Of course, the coats in these black-and-white photographs could actually be light brown, or they could have been touched up in retrospect, with photo-editors assuming the coats were white. After all, many paintings and reproductions show scientists wearing white lab coats long before they would’ve actually worn them. For example, a 1930s lithograph from the “Teachers World” supplement shows an illustration of Pasteur mulling over test tubes in a white lab coat.

The standard garb of laboratory workers also began to shift to white during the turn of the 20th century, when the relationship between medicine and laboratory science became more entwined. Lab scientists likely adopted surgical fashion as a result. In 1898, Sir William Osler, a Canadian physician already known for writing some of the first clinical laboratory literature, introduced ward laboratories at Johns Hopkins Hospital. Gradually, surgical practice relied on laboratories for tools, microscopes, chemical assays, pathology reports, and bacterial cultures to diagnose patients.

After the American education reformer, Abraham Flexner, visited medical schools all over America, he published a report in 1910 calling for restructuring medical education by getting rid of private medical schools, standardizing admissions, laboratory access, and curricula. With this, hospitals transformed from small, charitable institutions for the urban poor to large, funded medical centers serving mixed-class populations. They projected an image of themselves as safe, clean places, with the white lab coat — worn by surgeons and lab scientists alike — as evidence of that promise.

This white coat also became a way for these professionals to signal their membership in their respective highly-trained “guilds,” separating themselves from non-scientists. When chemistry historian Peter Morris stumbled upon a 1950s photograph of schoolboys from his alma mater wearing white lab coats, he theorized that the coat was introduced “to inculcate an ‘esprit de corps’ in students and laboratory workers” to improve morale rather than protect clothing.

Social psychologists tested a similar theory in 2012. Students wearing white coats, which they believed were doctors’ coats, performed better on tests. Students wearing identical coats, which they thought were painters’ smocks, showed no improvement. While the researchers called this observed phenomenon “enclothed cognition,” the real takeaway was perhaps much more intuitive — clothing shapes the way we think about ourselves.

The powerful symbolism explains the lab coat’s persistence even as we question its functionality. In 2009, a UCLA researcher was transferring tert-butyl lithium, a chemical that ignites spontaneously in air, between tubes. She wore safety glasses and nitrile gloves but no lab coat. When the syringe malfunctioned, and the highly flammable chemical splashed on her synthetic sweater, the synthetic material, essentially a solid form of gasoline, caught fire immediately. A colleague tried to smother the flames with his own lab coat, but failed. She should have been rushed to the emergency shower, but in the panic, she was not. She died from her burns, which covered 40 percent of her body.

Could a lab coat have saved her? Some are skeptical, since many common lab coats are also made of flammable, synthetic materials. But a flame-resistant coat, which had been commercially available for years, certainly might have.

Indeed, the incident incited a widespread rethinking of PPE, which is often eschewed for being bulky and cumbersome. Unfortunately, the environment the UCLA researcher worked in and the “circumstances that led to her death were certainly not unique.” Laboratory safety standards often lag behind available technology, and a culture of safety and training is wanting, even today.

Despite such lassitude, the lab coat has become an object of innovation. In 2012, MIT launched an initiative to create lab coats that scientists actually wanted to wear, addressing not just design but laundry logistics and institutional culture. The MIT Media Lab took this a step further, declaring, “Media Lab researchers are not only scientists — we are also designers, tinkerers, philosophers and artists. We need a different coat!” Their philosophy is that PPE should not only protect scientists but also support the exact kind of work they do. A scientist carrying around motors needs specialized pockets; one that works with lasers needs a reflective coating.

Since then, material science has ushered in promising possibilities for industrial uniforms, including lab coats, though many of these breakthroughs are in the research phase as of 2025. Scientists are experimenting with sustainable, bioactive coatings like chitosan (from seashells) and silver nanoparticles — natural microbial fighters — as well as phosphorus-based flame retardants that are highly effective at low concentrations. Built layer-by-layer with nanoparticles such as silica and titanium oxide, the experimental textiles can self-clean when activated by short periods in the sun. While we have yet to scale these innovations, their potential to improve lab coat functionality and clothing in general is worth watching.

Today, despite a wider array of shapes, colors, sizes, and materials, the archetypal dress of the laboratory remains the white coat. This suggests that the broader challenge is not design, but adoption. How do scientific institutions create new symbols of identity that prioritize function over tradition? Looking at the history of the white coat, we can observe that the shift from black frocks to white coats took decades and required not only technological but cultural change. The next transition, from symbolic to specialized PPE and laboratory wear, requires a similar shift in imagination.

Subscribe now

Donna Vatnick writes essays that explore scientific discovery and its most passionate devotees. Before completing her MFA in nonfiction, she worked in molecular biology labs and coordinated clinical trials in Boston.

Acknowledgements: Thanks to Kate Goldkamp and Natasha Muhametzyanova for their research guidance, and Ethan Madore and Julia Zaltsman for valuable conversations. Header credit: Wellcome Collection.

Cite: Vatnick, D. “Why Lab Coats Are White.” Asimov Press (2026). DOI: 10.62211/62hw-98tk

Further Reading:

1. Chambers, D. W. (1983). Stereotypic images of the scientist: The draw‑a‑scientist test. Science Education, 67(2), 255–265. https://doi.org/10.1002/sce.3730670213

2. D’Addezio, G., & Besker, N. (2024, January). Science and scientists from children’s point of view: comparison and gender outlooks among 2011 and 2021 primary school student drawings. Frontiers in Education, 8. https://doi.org/10.3389/feduc.2023.1179179

3. Rehn, D. K. (2021, November 3). Poverty and the pursuit of the philosopher’s stone: Representation of alchemists in sixteenth‑century Netherlandish art. Dana K Rehn Blog. https://danakrehnblog.wordpress.com/2021/11/03/poverty‑and‑the‑pursuit‑of‑the‑philosophers‑stone/

4. Smithsonian Institution. (2020). Alexander von Humboldt’s influence on America. IMPACT, 6(2). https://www.si.edu/support/impact/humboldt

5. Hardy, S., & Corones, A. (2016). Dressed to heal: The changing semiotics of surgical dress. Fashion Theory: The Journal of Dress, Body & Culture, 20(1), 27–49. https://doi.org/10.1080/1362704X.2015.1077653

6. Lister, B. J. (2010). The Classic: On the antiseptic principle in the practice of surgery. Clinical Orthopaedics and Related Research, 468(8), 2012–2016. https://doi.org/10.1007/s11999‑010‑1320‑x

7. Cheyne, W. W. (1882). Antiseptic surgery: its principles, practice, history and results. Smith, Elder. https://archive.org/details/antisepticsurger00chey/page/70/mode/2up

8. Drysdale, C. (2025). What we find in the sewers. Asimov Press. https://www.asimov.press/p/sewers

9. Paul, S., Salunkhe, S., Sravanthi, K., & Mane, S. V. (2024). Pioneering Hand Hygiene: Ignaz Semmelweis and the Fight Against Puerperal Fever. Cureus, 16(10). https://doi.org/10.7759/cureus.71689

10. Aiello, A. E., Larson, E. L., & Sedlak, R. (2008). Hidden heroes of the health revolution Sanitation and personal hygiene. American Journal of Infection Control, 36(10), S128-S151. https://doi.org/10.1016/j.ajic.2008.09.008

11. Macintyre, I., & Hughes, S. (2024). Robert Lawson Tait (1845–1899): The true innovator of aseptic surgery?. Journal of Medical Biography, 32(1), 157-165. https://doi.org/10.1177/09677720221140085

12. Flannery, M. C. (1999). Dressing in Style? An Essay on the Lab Coat. The American Biology Teacher, 61(5), 380–383. https://doi.org/10.2307/4450702

13. Friedlaender, G. E., & Friedlaender, L. K. (2014). Art in Science: the Gross Clinic by Thomas Eakins. Clinical Orthopaedics and Related Research, 472(12), 3632–3636. https://doi.org/10.1007/s11999-014-3989-8

14. Millard, C. (2012). Destiny of the Republic: A Tale of Madness, Medicine and the Murder of a President. Penguin Random House.

By Dr. David S. Kim

Researchers have known that the pancreas secretes several peptide hormones critical to metabolic regulation, most notably insulin and glucagon, since the early 1900s. With the advent of recombinant peptide synthesis in the 1970s, scientists not only began working on isolating the gene responsible for insulin production but fabricating it outside the body. In 1978, scientists at Genentech turned bacteria into factories able to transform sugar into insulin by cutting the insulin gene from the human genome and pasting it into the genome of E. coli. This breakthrough allowed peptide drugs to be synthesized at mass scale, without needing to be harvested from animal pancreases, and gave birth to the biotechnology industry.

The role of glucagon and the gene encoding it, however, remained elusive. Then, in 1982, Joel Habener and colleagues at Massachusetts General Hospital (MGH) published a paper indicating not only that they had located the human glucagon gene, but that it actually encoded three separate peptides, two previously unknown. Those two unknowns were labelled as glucagon-like peptides: GLP-1 and GLP-2.

Svetlana Mojsov, a peptide chemist also working at MGH, managed to synthesize a truncated, but active, form of GLP-1. In 1987, her paper with Habener and Daniel Drucker showed that GLP-1 stimulates glucose-dependent insulin secretion in the pancreas, suppresses glucagon release, and slows gastric emptying, all of which stabilize blood sugar levels after eating.

While these effects initially made GLP-1 seem a promising candidate for drug development, the earliest clinical studies from 1992 and 1993 revealed that GLP-1’s half-life in the body was disappointingly short. A proteolytic enzyme, dipeptidyl peptidase-4 (DPP-4), binds to and destroys GLP-1 molecules in the blood in minutes.

Hope for boosting GLP-1’s half-life in the body came in the 1990s, when endocrinologist John Eng, at the Bronx Veteran Affairs Hospital, noticed something curious about the Gila monster while researching snake and lizard venom effects on the pancreas. Specifically, he was struck by how this desert lizard eats extremely infrequently, yet somehow maintains a stable blood sugar level for months.

Eng found that Gila monsters secreted a salivary venom peptide, now known as exendin-4. Structurally similar to GLP-1 but unaffected by DPP-4, this peptide lingers in the blood for hours rather than minutes. Exendin-4 thus became the molecular blueprint for exenatide, the first twice-daily injectable GLP-1 receptor agonist, approved by the FDA in 2005.

For many years after exenatide’s approval, GLP-1 receptor agonists were used primarily as diabetes medications. Practitioners and patients noticed, however, that taking such medicines produced weight loss as a side effect. During this time, researchers also found GLP-1 receptors in the brain and pancreas, particularly in areas controlling appetite and reward. And in 2021, the results from a series of five STEP trials (Semaglutide Treatment Effect in People with obesity) demonstrated marked weight loss even in individuals without type 2 diabetes (T2DM): Phase 3, for example, showed an average loss of about 15 percent over 68 weeks, more than double achieved by any prior obesity drug. In 2023, a trial of 17,604 adults (SELECT) with obesity and without T2DM, followed for over three years, showed a 20 percent reduction in major adverse cardiovascular events.

In response, GLP-1 prescriptions rose sharply. According to CDC data, the share of adults with type 2 diabetes using GLP-1 injectables more than doubled — rising 155 percent between 2018 and 2022 — while overall spending on these medications climbed more than 500 percent between 2018 and 2023.

The majority of patients continue to inject these drugs rather than take them orally. But a 2020 survey of 600 patients found that 76.5 percent would prefer a once‑daily oral dose over a once-weekly injection of GLP-1. Significant funding is going into R&D to launch an oral alternative that performs as well as an injectable version. While oral formulations are beginning to enter the market, so far they remain less effective and more expensive to manufacture. Why?

To the Stomach, All Peptides Are Food

Peptide drugs like GLP-1s are made up of a chain of amino acids chemically bonded together. For a GLP-1 molecule taken by mouth to reach its target, it must first pass through the stomach, where hydrochloric acid and peptidase enzymes cleave these amino acid bonds. Most ingested peptides are simply broken down into their constituent amino acids and absorbed as nutrition, indistinguishable from the protein in food.

The small percentage of GLP-1 molecules that might survive this journey and arrive into the intestines face a second gauntlet. Once in the blood capillaries, DPP-4 quickly degrades the molecule, and the drug is excreted by the kidneys.

Biochemical engineering, however, has allowed drugs like semaglutide to survive longer in the blood. First, an amino acid substitution stops DPP-4 from destroying GLP-1. This is achieved by swapping the alanine amino acid at position 8 in the peptide backbone for alpha-aminoisobutyric acid (Aib), a modified amino acid that DPP-4 cannot recognize or cleave. As a result, this modified form of GLP-1 is not degraded so swiftly.

Yet even a DPP-4-resistant peptide is small enough to be filtered by the glomeruli in the kidneys into the urine. Therefore, the chemists behind semaglutide made a second modification: they fused a fatty acid chain to the peptide in a process called lipidation. This causes the drug to bind to human serum albumin, a large transport protein abundant in the blood, with a natural half-life of approximately 20 days. The drug effectively co-opts albumin’s stability and longevity; the lipidated GLP-1 molecules are bound to albumin, which is too large to be renally filtered, and are thus shielded from plasma proteases.

The combined effect of these modifications increase the drug’s half-life from minutes to approximately 168 hours, enabling its once-weekly injection schedule — an enormous improvement compared to the twice daily injection required for exenatide (Byetta) which was the first GLP-1 drug that was launched.1 Semaglutide (under the brand names Ozempic and Wegovy) swept the market. However, these modifications to the molecule could not solve the problem of degradation and poor absorption when taken by mouth.

From Silkworm to Trojan Horse

We might draw parallels for oral delivery of peptide and protein drugs from the case of serratiopeptidase, a little‑known drug from the 1960s with a peculiar history.

Serratiopeptidase is used to treat post‑operative inflammation in a handful of countries, including Japan, India and Sierra Leone. Produced naturally by the bacterium Serratia marcescens, which resides in the gut of the Japanese silkworm, this enzyme is used by the moth to dissolve silk fibers as it emerges from a cocoon. It is pharmacokinetically puzzling as an oral therapy because, as a peptide, it faces the usual hurdles of gastric degradation and poor epithelial uptake. Yet, surprisingly, no injectable formulation of serratiopeptidase is marketed. The reason is that intravenous administration carries a risk of immunogenicity — the immune system recognizes Serratia‑derived molecular signatures on the enzyme as foreign, which can trigger immune reactions. This danger is blunted or even absent, however, when the drug is given orally.

Serratiopeptidase only works because the oral dose is high enough that the small fraction of the drug which survives the gut reaches systemic circulation at concentrations sufficient for therapeutic effect. In this way, it serves as a proof‑of‑concept that even a peptide or protein drug known to have poor bioavailability can still be delivered orally, and without any modification, if we compensate by delivering a sufficiently high dose.

But in general, turning a peptide (including GLP-1 or insulin) into an oral formulation means overcoming three major challenges: protecting the peptide from digestion in the stomach, improving absorption of the peptide via the gut, and slowing down the enzymatic breakdown and excretion of the peptide once it is in the blood. Novo Nordisk’s oral semaglutide, Rybelsus, employs both co-formulation and clever biochemical engineering of the drug molecule itself to move us closer to this goal.

Specifically, Rybelsus works by co-formulating semaglutide with SNAC (sodium N-[8-(2-hydroxybenzoyl) amino] caprylate), a small synthetic molecule that functions as an absorption enhancer in the gastrointestinal tract. SNAC raises the pH of the immediate environment around the drug. The higher pH protects the peptide from pepsin degradation, which requires acidic conditions. SNAC also acts as a transient permeation enhancer, aiding semaglutide’s absorption into the blood, directly through the stomach lining and the small intestine.

While this co-formulation certainly improves oral delivery, it hasn’t fully solved the problem of dosage requirements. One pharmacokinetic study showed a modest bioavailability of only 0.8 percent for oral semaglutide, meaning less than one percent of the drug is absorbed into the blood where it can take effect.

To be effective, then, Rybelsus must be formulated to contain high doses of semaglutide. Whereas the GLP-1 drug Ozempic is injected once weekly at 2 mg, Rybelsus is taken daily and requires a minimum dose of 14 mg to achieve comparable glycemic effects. That is almost a 50-fold difference in total weekly intake for what is, fundamentally, the same drug, simply because most of the substance is degraded before they can be absorbed into the blood.

Additionally, Rybelsus doesn’t perform as well as its injectable counterpart, even at this higher dosage. In clinical trials, Ozempic (2 mg weekly subcutaneous) delivered a 15-20 percent body weight reduction over 68 weeks. Rybelsus at the approved 14 mg daily dose, though, achieved just 4-5 percent in the PIONEER trials. However, at 50 mg, oral semaglutide (OASIS 1) narrowed that gap substantially, achieving 15.1 percent weight loss at 68 weeks, approaching injectable performance.

While it is no secret that the cost of production of the drug molecule itself is only a small part of the retail price, the higher dosage required by oral delivery becomes a non-trivial difference in manufacturing costs.

Semaglutide costs $70,000 to $100,000 per kilogram to manufacture at commercial scale, according to two studies. Producing the drug ingredient contributes only $0.20 to the cost of each weekly 2mg injection. What forms the bulk of the retail price2 of peptides is not the cost of manufacturing, but the capital infrastructure: the regulatory burden, the cold chain logistics, the pen device (which sometimes costs more than the drug itself), and, above all, the return needed to justify $9 billion in the initial R&D of these GLP-1 drugs.

The cost breakdown looks much different for an oral semaglutide. In the OASIS 1 trial, patients took a 50 mg daily dose — meaning a monthly supply contains roughly 1,500 mg of active drug compound. A monthly injectable supply, by comparison, contains only about 8 mg. At $100,000 per kilogram, the drug compound alone for an oral monthly supply could already cost around $150, versus less than $1 for the injectable — before factoring in formulation, packaging, distribution, or profit.

This is the central problem with oral GLP-1s, then; biochemical engineering can technically solve the half-life problem, and co-formulations can improve the absorption problem, but the higher doses required make for formidable raw ingredient costs.3

Turning Food into Drug Delivery Vehicles

A solution to this problem may come from foregoing chemistry and turning, instead, to biology.

Genentech pioneered recombinant manufacturing of human insulin by cloning the human gene into E. coli as far back as 1978. Even today, many GLP-1 peptides are made by cloning the corresponding gene into microbes, culturing them in a fermenter, and then subjecting the resulting broth to successive rounds of filtration and chromatography to isolate the peptide precursor before undergoing chemical modifications. E. coli, though, is not edible, as it is laden with endotoxins that can trigger a septic shock. Therefore, any drug made in a cell like E. coli must be purified extensively, even if delivered orally. This downstream purification accounts for up to 80 percent of the total manufacturing cost.

Suppose, instead, that the microbes used for production were also edible. The drug could then be swallowed inside intact cells, and the massive cost of purification would disappear. In theory, there are millions of microbe species that can be engineered to synthesize peptides. But currently, we have the tools to genetically engineer only a few, and most microbes have never been studied for food safety and toxicology.

One attractive candidate is the edible, single-celled cyanobacterium Arthrospira platensis (spirulina). Cyanobacteria and many microalgae species have been consumed by humans for centuries and are Generally Recognized As Safe (GRAS) by regulatory agencies. Many people already eat spirulina for its antioxidants. Thus, the organism could be quite appealing for direct oral delivery in its intact cellular form, if it could be engineered to synthesize a GLP-1 molecule in sufficient quantities. Given the low cost of production, a 20-gram scoop of dried “algae powder” taken twice daily could, in principle, contain an orally adjusted therapeutic dose of a GLP-1 agonist for around $4/day.4

The cost argument alone is compelling, but using edible algae provides another benefit: keeping the GLP-1 inside the algal cell wall may improve the drug’s survival in the gut. Experimental data have already demonstrated that over 70 percent of the expressed proteins in spirulina remained intact after two hours of incubation under simulated gastric conditions that would fully degrade the same protein in a purified form within minutes. The cell wall acts as a natural enteric capsule, shielding the GLP-1 molecules from pepsin and the acid bath of the stomach, and releasing it progressively as it transitions into the higher-pH environment of the duodenum.

The reason this concept has not yet been tested in the context of GLP-1 is because, until very recently, it was not possible to engineer microalgae and cyanobacteria. Spirulina was, until a few years ago, considered genetically intractable because it has multiple copies of its genome and is resistant to standard transformation methods. In 2022, though, a paper reported successful engineering and high-level expression of therapeutic proteins in spirulina, with the encoded proteins accumulating to up to 15 percent of total dry biomass.

Today, newer organisms like UTEX 3222 (discovered in 2023 and also known as “Chonkus”) can double several times faster than spirulina and be engineered to express recombinant peptides. Some of these strains are quite simple to grow, needing just light and CO₂. Their adoption might not only make oral GLP-1s affordable, but also allow the mass production of customized peptides from photosynthesis — all inside edible cells.

Many cyanobacteria (and microalgae broadly) are safe to eat, though not every strain of cyanobacteria has a GRAS certification.5 Under the GRAS framework, an ingredient has to be demonstrated as safe either “through (1) scientific procedures or (2) in the case of a substance used in food prior to January 1, 1958, through experience based on common use in food.” For cyanobacteria to be used as a platform for oral GLP-1 delivery, either a GRAS organism like spirulina must serve as the host organism or, if choosing a different species, the manufacturer must prove with adequate in vivo trial data that it is safe to eat.

In addition, one must gain FDA approval by clinically demonstrating that the delivery of the microbe-encapsulated GLP-1 through this oral route leaves the drug molecule itself unchanged (with no post-translational modifications), that it has a similar pharmacokinetic profile to the reference molecule, and that it shows no adverse effects. Since this is a new delivery platform of existing drugs rather than a new drug, the regulatory bodies in many jurisdictions typically have alternative, shorter approval pathways available to drug sponsors that do not require the standard Phase I-III clinical trials for a new drug application.

A third argument for adopting edible organisms for GLP-1 delivery would be to increase global accessibility. Bangladesh, for example, ranks among the world’s top ten countries for T2DM patient burden, with a prevalence of 9.2 percent among adults. The disease currently impacts over eight million people, a number projected to grow to 15 million by 2045. Yet GLP-1 drugs are effectively absent from the formularies of Low and Middle Income Countries (LMICs). A 2024 study found that GLP-1 drugs can likely be manufactured more cheaply than their current market costs of, for example, $95 in Brazil, $115 in South Africa, and $353 in the United States.

Médecins Sans Frontières has called for Novo Nordisk and Eli Lilly to license their patents to generic manufacturers, noting that neither company has adopted pricing strategies for low-income markets. The surge in U.S. demand has also triggered global shortages that hinder the drug’s availability for essential diabetes care in LMICs. If we can provide alternative biomanufacturing capacity that slashes cost, turns injectables into safe oral, algae-encapsulated formulations, and eliminates the need for refrigeration, we may be able to ensure that patients in LMICs are no longer last in line for the latest and best therapeutics.

So far, three major problems have kept GLP-1 drugs injectable rather than oral: enzymatic degradation, poor bioavailability, and the sheer cost of manufacturing. Novo Nordisk's Rybelsus showed that transcellular gastric absorption, boosted with SNAC, can partially get around the first two problems, but cannot solve the third. This is why research groups around the world are trying to engineer microbes to act as edible drug factories. The idea is to skip the most expensive part of manufacturing, downstream purification, by keeping the expressed peptide locked inside the protective walls of edible cells.

The science of oral peptides, at this point, is no longer a mystery. We know oral peptide drugs can work. The harder question is whether patients will come to accept the idea of eating their medicine in a spoonful of algae.

Subscribe now

David S. Kim is the founder of CyanoCapture [General Bio], a synthetic biology company that is manufacturing peptides by genetically engineering non-model microbes. He is a medical doctor based in the U.K.

Cite: Kim, D.S. “The Quest for Oral GLP-1s.” Asimov Press (2026). DOI: 10.62211/82ej-91yt

Living organisms are surprisingly creative when it comes to finding their way into hostile environments. Microbes, after all, thrive everywhere from hyperarid deserts and hydrothermal vents to the stratosphere.

Of all hostile environments, one might think the most unbreachable to be the laboratory, where populations of bacteria are stored inside tubes sealed with tight screw-caps and stacked in freezers at -80°C. Of course, normal evolutionary processes don’t magically stop at the doors of the lab: organisms still mutate, fragile microbial species can still be out-competed by faster-growing contaminants, and parasites can still spread. However, vigilant scientists devote considerable effort to making sure containers hold what is written on their labels and, when in doubt, can sequence any part of a sample’s DNA, discarding or incinerating undesired specimens.

Still, interlopers persist. One organism that has been particularly successful at infiltrating microbiology research facilities is called Φ80. It is a bacteriophage, or virus that infects bacteria. Slowly but surely, Φ80 has been spreading. Researchers carry out experiments, make reproducible observations, and publish — usually without ever realizing that their samples have quietly been contaminated with this phage.

On occasion, hidden contaminations of Φ80 have led to paper retractions, with researchers discovering not only that their results are artifactual, but that they personally amplified the invasion and transmitted it elsewhere. In a 2017 paper, for example, scientists confessed that “beyond the confusion created by our erroneous interpretation of results obtained with Φ80-infected strains, we are worried that we have sent these strains to many research groups.”

To understand how such unwitting contamination can occur despite researchers’ best efforts, and to work towards better containment, it helps to first understand how bacteria are typically cultivated.

Life in a Tube

The most commonly studied bacterium, Escherichia coli (E. coli), has spent most of its recent evolution inside the human gut. Acclimated to that environment, E. coli grows best at 37°C — dividing about every 20 minutes in good conditions. If a researcher inoculates just a few E. coli cells in a rich liquid medium and puts the tube in a shaking incubator, it only takes about a dozen hours to form a population of several billion cells. As these billions of cells continue to divide every 20 minutes, their consumption of nutrients also grows exponentially, eventually exhausting all resources, at which point they enter a kind of dormant state.

E. coli cells are almost completely transparent, but their refractive index differs slightly from the medium in which they are grown. Each cell is only about one micrometer wide, but when millions accumulate in the same test tube, visible light gets scattered, and the liquid becomes opaque.

Compared to animal guts, with their folds, crypts, and periodic floods of bile, a test tube is, from the point of view of a bacterium, a featureless space. Most importantly, it is also the only place on Earth where bacteria don’t have to deal with phages. While we don’t pay much attention to phages, they are everywhere, and Earth would be quite different without them. Every day, phages kill 15 to 40 percent of all bacteria in the ocean, keeping microbial ecosystems from spiraling out of balance.

A thriving culture of bacteria, peacefully growing in a test tube, would surely be an appetizing ground for a bacteria-killing phage to replicate, if it could find its way inside. For this reason, microbiologists are extremely vigilant about phage contamination. Merely mentioning phages at lab meetings will draw nervous looks!

Phages get everywhere, are hard to destroy, and tear through experiments. I’ve seen entire labs shut down for days while whole HVAC systems are sterilized. If there is any detectable sign of unwanted phages, then, researchers stop what they’re doing, autoclave every object possible at the highest possible temperature, and wash the rest with the harshest chemicals available. While many bacteriophages are capable of infecting common laboratory strains, they can usually only spread so far.

The most straightforward strategy by which phages spread is by replicating as fast as they can, infecting as many cells as possible, and killing them in the process. Phages that employ this strategy are called lytic.

One example is T7, a lytic phage that infects E. coli. When T7 enters a bacterium, it uses the cell’s resources to generate copies of itself. This takes about 17 minutes, at which point the cell contains hundreds of phage particles. The phages then burst out of the cell in an event called lysis, searching for new cells to infect. The speed of this process depends on the population density of the bacteria. In ideal conditions, a phage population can expand by a factor of 10⁵–10⁶ every hour.

For bacteria in a test tube, phage infection means annihilation. Billions of cells can get entirely wiped out in two hours. As the bacteria burst and release their contents in the medium, the liquid becomes entirely transparent again.

While the lytic strategy employed by T7 is superb for proliferation, it doesn’t work well for secretly invading a laboratory. If an entire bacteriological culture suddenly collapses for no reason, even the most absent-minded researcher is apt to notice that their cells have been contaminated.

In fact, being too destructive is also a problem in the wild: if a phage eradicates its host too quickly, there might be no bacteria left to infect, meaning death for the phage as well. For that reason, many phages have adopted a more complex life cycle than that of T7, one known as lysogeny.

Take Phage λ, one of the best-studied lysogenic phages. Whenever it infects a new cell, it uses a genetic circuit to choose between two outcomes: either it goes through the usual lytic cycle, filling the cell with virions until it bursts, or it inserts its genome into the bacterium’s DNA. The integrated form is called a prophage.

The host bacterium then carries the phage genome even as it grows and divides. HIV behaves similarly in humans, but the advantage of doing so is much more obvious in bacteria: with a host that divides every 20 minutes, it can be worthwhile for phages to hitchhike for several generations of bacterial division, giving the host time to proliferate on its behalf.

It is even common for lysogenic phages to confer some benefit on their host, helping the infected bacteria outcompete their uninfected sisters and maximizing the prevalence of the prophage in the population.1 There is an interesting puzzle here: if there were an easy way to improve the fitness of the host bacteria, surely the bacteria would already have found it through normal natural selection. How, then, can a parasite grant an advantage that the host wouldn’t have evolved on its own?

One way is to provide the bacteria with genes helpful in the short-term at the expense of long-term sustainability. For example, phage-provided genes may offer costly solutions to temporary problems, like immunity to other pathogens or resistance to specific stresses in the environment.

When the host bacterium is itself a pathogen, phages can benefit from making bacteria more virulent to their host, often against the interests of the bacteria themselves. Some of the world’s most horrific diseases, like diphtheria, cholera, and botulism, are due to bacteria releasing deadly toxins into their host’s body. However, these toxin genes don’t come from the bacteria — they are often poisoned gifts from lysogenic phages.

So when do these dormant lysogenic phages elect to attack? The trigger is usually some indication that the cell is imperiled, whether by DNA damage, oxidative stress, or antibiotics. Some phages even force their hosts to secrete molecules into the environment so they can coordinate their decisions with phages in other infected cells. At the signal, like rats abandoning a sinking ship, the prophage leaves the chromosome, replicates profusely, and bursts into the environment.

In the context of the lab, this often resembles lytic infection, with most cells bursting and becoming transparent. But then, a small subpopulation of bacteria survives, having integrated the phage into their DNA. The prophage then expresses a protein that blocks further infection by the same phage, creating an immune population of cells that resumes growth as if nothing happened.

So, by using lysogeny, a phage can stay hidden in the chromosome of bacteria for extended periods of time. However, in a research lab environment, this tactic can only succeed if the phage manages to integrate itself into bacteria without the researchers noticing. And this is precisely why Φ80 is so cunning.

Subscribe now

The Intruder

A key feature of Φ80 is that it replicates very slowly. While T7 goes through a full lifecycle in less than 20 minutes, Φ80 takes 90 minutes. This is a more dramatic difference than it seems: growth rate compounds with each generation, meaning that T7 can produce 10¹⁰ new particles in two hours, while Φ80 makes barely a thousand in the same interval.

Moreover, while the mechanism Φ80 uses to choose between lytic replication and chromosomal integration has not been fully deciphered, it appears to favor integration far more often than comparable lysogenic phages. As a result, a new infection by Φ80 will not cause the bacterial culture to suddenly collapse and become transparent. Instead, the phage will slowly establish itself inside the population in a way that is invisible to human overseers.

Additionally, it takes very little for Φ80 to switch back into lysis mode. Whereas many prophages use DNA damage as a signal that the bacterium is about to die and it’s time to look for a new host, Φ80 is so sensitive that even the slightest hiccup in DNA polymerization (something that happens routinely when a cell replicates) can make it go lytic. As cells carrying the Φ80 prophage grow, a tiny fraction will constantly pour new phage particles into the medium. As a result, billions of these phages will have accumulated in the tube over a cycle of growth, even though the culture continues to look perfectly healthy to the naked eye.

What’s worse is that this phage is mostly active at temperatures below 37°C. But, 37°C being the standard temperature for E. coli, pretty much every measurement is done at it. A researcher could put the culture on a slide in a microscope heated to 37°C, and everything would look fine. But as they took the culture through the corridor between the shaker and the microscope, the temperature would drop a little bit, and that’s precisely when a fraction of the culture would burst into a barely-noticeable, steady stream of Φ80 particles. In other words, the phage replicates precisely when researchers are not watching.

Even so, escaping detection is only half of its strategy. Φ80 also has an ingenious mechanism to break free from its tube and travel to new samples by hitchhiking on a classic molecular biology technique, thus turning human researchers into unwitting dispersers.

Scientists as Phage Dispersers

The technique in question is called P1-transduction. It is a widely used method for copying and pasting genes from one strain of bacteria to another. To understand how it works, we need to meet yet another phage: P1.2

The key to P1’s utility lies in a quirk of phage biology. During their reproductive cycles, phages use the bacteria’s machinery to produce a massive amount of empty capsids, which then house copies of the phage’s genome. This maintains a little army of fresh phage particles ready to go out and infect the next generation of cells.

Occasionally, though, there’s a mistake in the process: a piece of the host bacteria’s genome gets incorporated into a capsid instead of the phage’s genome. While this normally happens only rarely, P1 makes this mistake unusually often, about 0.1 percent of the time.3

As the phage particles are liberated in the medium and infect other bacteria, the random bit of DNA from the original, infected cell ends up injected into a different one.

This recipient cell notices that there’s a piece of linear, double-stranded DNA floating around in the cytoplasm. (E. coli’s chromosome is circular, so the presence of linear DNA is abnormal — it usually means that the chromosome is broken). In response, the bacterium expresses a suite of enzymes that stitch homologous pieces of DNA together based on sequence similarity, hoping to reconstruct a complete, circular chromosome. Thus, if P1 takes a piece of DNA from the first bacterium, then injects it into a second bacterium, the transplanted piece of DNA will often end up incorporated into the recipient’s chromosome, overwriting the native sequence. This process is called transduction.

Transduction gives scientists a clever way to copy and paste genes from one bacterial strain into another. First, the researcher inserts an antibiotic-resistance gene in the chromosome of one bacterium, called the donor strain, next to the gene they want to move.4 Then, they culture this donor bacterium in liquid broth, grow it up, and add a bit of phage P1. The P1 phages infect the donor cells, filling them up with invisible phage particles, including a few that took on a fragment of the donor strain’s DNA.

Next, these phage particles can be used to infect the recipient. At this point, random fragments from the donor DNA will be injected into the recipient and replace the native sequence by recombination. Finally, a researcher spreads the culture over a Petri dish containing the antibiotic from the first step, along with a chemical that inhibits further infection by P1. At this point, only the cells that have received the right region of the donor’s DNA — the bit containing an antibiotic resistance gene — can survive. And so, overnight, the few cells that received the desired edit will grow to form visible colonies.

When a two milliliter culture of bacteria gets infected by P1, this quickly generates tens of billions of phage particles. 0.1 percent of them will have incorporated a piece of the host DNA by mistake, which means our tube will contain around ten million such mistakes. Each of these pieces of DNA is about 90,000 base pairs long (like the phage’s genome). So, the total amount of donor DNA (mis)incorporated by the phage population will be almost a trillion nucleotides. For comparison, the full genome of E. coli is only 5 million base pairs. Therefore, the phages in our tubes collectively contain the entire genome of the donor bacterium: in fact, hundreds of thousands of copies of it. Add this to a recipient culture with several billion cells, and it’s almost guaranteed that one of them will receive the desired segment and incorporate it in the right place.

This is what makes P1 transduction a quintessential tool for microbiology. Although it relies on a series of unlikely coincidences, the sheer number of particles that fit in a droplet of liquid means it somehow works out in the end. And as long as a genetic element has an antibiotic-resistance marker next to it, it can be easily moved from strain to strain. The most famous application of this is the Keio collection, a holding of thousands of E. coli strains, each with a different gene replaced with an antibiotic resistance marker.

This ease of use made P1-transduction a hugely popular technique, used daily by thousands of biologists around the globe. Unfortunately, its popularity also makes it a perfect conduit for a stealthy parasite like Φ80 to travel from strain to strain. This is because, by a twist of fate, Φ80 has figured out a way to “tag along” with P1 phages.

Consider a routine P1-transduction experiment, aimed at moving a gene from donor to recipient. The donor strain, unknown to the researcher, harbors the Φ80 prophage. When P1 is added to the culture, the population collapses after a few hours; the expected sign of successful phage production. But the arrival of P1 is detected by the dormant Φ80 prophages, which now replicate alongside P1 in massive quantities. The oblivious researcher infects the recipient strain with this mixed phage solution, selects for the antibiotic resistance marker, and sequences the target locus. Everything appears to have worked as expected (the desired gene has been successfully transferred), but the new strain now carries Φ80.

This is, in fact, how Φ80 was discovered in the first place. In the late 2000s, researchers at the University of Illinois in Urbana-Champaign were trying to understand how bacteria repaired their DNA after damage from UV light. They suspected that LigB, a ligase that ties strands of DNA together, might play a role.

To test this hypothesis, the researchers did a classic knock-out experiment: they used P1-transduction to remove the ligB gene, then put cultures of the original bacteria and the modified ones side by side and exposed them to UV light. After exposure, the cultures of the strain lacking LigB mostly collapsed, while the natural strain didn’t. From that, they drew the logical conclusion that LigB must somehow repair UV damage. And yet, their assumptions were all wrong; the LigB ligase has nothing to do with UV damage.

What really happened was that their donor strain was infected with Φ80, which took advantage of the P1-transduction to jump to the new strain. When this strain was exposed to UV light, the DNA damage induced Φ80’s lytic cycle, which looked exactly as if the cells were dying from the damage itself.

But why does P1 activate Φ80? Is it an ancestral relationship evolved over the course of millennia, or is it a new adaptation evolved specifically within the research lab environment? While no one has studied the question in detail, we can find some clues by digging through the scientific archive.

Φ80 was first isolated in 1961 from a collection of clinical samples from Osaka University’s Department of Clinical Evaluation. The version of Φ80 studied by the Urbana-Champaign team is genetically identical to the original, Japanese variety, so any special adaptations must have occurred before 1961. While the phenomenon of P1 transduction was already known at the time, it was not yet used as a method for genome engineering, so it’s unlikely that the “hitchhiking” of molecular biology techniques was a new trait actively selected for.

All in all, it seems likely that Φ80’s ability to infiltrate biology labs is an exaptation: a trait that was originally selected for unrelated reasons, but was later repurposed in a different way after a change of environment. The genetic diversity of phages, after all, is so immense that it is unsurprising that one of them would turn out to be perfectly suited to life in the laboratory.

What Now?

Addressing the Φ80 crisis requires overcoming three obstacles: One is technological, one is social, and one is related to the very nature of biology as a science.

The technological challenge is the most tractable. For a long time, it was only practical to sequence short regions of the genome (using the Sanger method), so researchers would only PCR-amplify and sequence the particular locus they intended to modify. This can grant them great confidence that a strain carries the expected genotype, but it’s basically useless against a phage that inserts itself in a different region of the genome.

While whole-genome sequencing has become much more accessible, it doesn’t really solve the problem. First, it is more expensive than local Sanger sequencing. Second, short-read sequencing (e.g., Illumina) only generates short segments of DNA that must then be assembled on a computer. It is possible to use these short segments to stitch together the entire genome, much like a puzzle, but it’s slightly faster to align it to a reference genome, with the drawback that only sequences that map to the reference are considered. Sadly, the latter method will only detect a hidden phage if a researcher provides the phage genome as one of those references.

In other words, it would be feasible to find Φ80 through whole-genome sequencing, but it would require making two sub-optimal choices: using a slightly more expensive sequencing run and processing the results with a slightly slower algorithm. This is why, in practice, nobody does it. Fortunately, a third generation of sequencing can help: it’s becoming increasingly popular to sequence entire genomes using long-read (e.g., Nanopore) sequencing. This method reads DNA in long segments of thousands of nucleotides, making Φ80’s 46,000 nucleotides of foreign DNA much easier to spot. Hopefully, in the near future, this technique will help us identify Φ80 invasions more readily.

In the meantime, however, it remains difficult to identify Φ80 unless actively looking for it. This brings us to the social obstacles. For a microbiologist to check their strains, they need to be aware of the phage’s existence and dedicate time to checking for it.

Imagine being a fourth-year PhD student who finally managed to measure an interesting, statistically-significant, reproducible difference between two bacterial strains. Their hypothesis makes sense. They did everything right. As far as publication goes, their findings as they are would merit publication. The painful truth is that double-checking for Φ80 contamination is guaranteed to hurt their career. At best, they spend a few hours performing an additional experiment and find nothing; at worst, they discover that their results are artifactual and no longer publishable. There is no reward for discovering that your results are false positives.5

Finally, while Φ80 contamination may be unfortunate, there is something inevitable about this sort of setback given the inherent messiness and complexity of living organisms. So while Occam’s razor is a convenient heuristic, it does not actually work that well for biology; there is always a long tail of insanely complicated speculative possibilities that, every once in a while, turn out to materialize. A stealthy phage spreading from culture to culture? It may not be the most plausible explanation a priori, but it’s plausible enough. And if it’s not Φ80, it could still be any number of improbable but real bottlenecks: autoclave-resistant RNases, traces of phosphorus in the water, novel organelles, or parasites from an entirely unknown domain of life.

This is simply part of life as a biologist. In the 1990s, cognitive scientist Kevin Dunbar spent a year conducting ethnographic observations in a molecular biology lab. According to his notes, 40 to 60 percent of experimental results go against expectation, often for totally unknown reasons. As scientists push the frontier of knowledge, phenomena that were once considered mysterious become settled science, opening the way for new experimental techniques that themselves unearth new mysteries.

This actually makes me fairly hopeful about the future of biology. When people started experimenting with E. coli in the 1940s, they didn’t know a single thing about it. Since then, there has been considerable progress, and we now have at least some information about 3803 of its 4747 genes. The more we understand it, the less room there is left for unexpected artifacts creating misleading results and false interpretations, and the easier it becomes to make progress. Many outlandish oddities, higher-order emergent interactions, and bizarre edge cases loom ahead, but the fog is lifting.

In the end, we should embrace experimental artifacts and contamination not as setbacks to our research agenda, but rather as essential parts of the systems we study. Sometimes, as in the case of Φ80, the artifacts may even have a life of their own.

Antoine Vigouroux is a systems biologist. He writes about science and progress at Telescopic Turnip.

Thanks to Jay and Jasmine for feedback.

Cite: Vigouroux A. “How Φ80 Infiltrates Research Labs.” Asimov Press (2026). DOI: 10.62211/75eh-27rw

By Brian Naughton

Over the past few months, AI-based tools have emerged that enable scientists to design original antibodies on the computer for the first time. A year ago, none could reliably do this computationally. But now companies like Nabla Bio, Chai Discovery, Latent Labs, Manifold Bio and, most recently, DeepMind-spinoff Isomorphic Labs have allowed high success rates. There are even open source tools, such as BoltzGen and Germinal, that deliver similar performance.

The rapid progress in antibody design matters because these molecules are among the most versatile tools in biology. Many medicines — including Humira and Adalimumab — are antibodies, and cheap diagnostics, including $1 COVID tests, rely on them as well. These Y-shaped proteins make excellent binders, as the two arms can latch onto proteins or other molecules and block their activity.

Before these AI tools existed, scientists searching for a useful antibody would first need to screen billions of candidates in laboratory assays to identify just a handful with high affinity for a target. BindCraft, released in 2024, changed this. For many targets, a suitable binder can now be found after just tens of attempts rather than billions. BindCraft uses the AlphaFold 2 model, but inverts it: the model creates a protein structure expected to fit onto a chosen target, then converts that “shape” back into an amino acid sequence that can be synthesized and tested in the laboratory.

Antibodies are not the only type of protein binder, and binders are generally catalogued according to their size. A “peptide” is any protein smaller than 30 amino acids, a “mini-binder” is between 50 and 250 amino acids in size, and an “antibody” covers anything from a 120-amino-acid antibody fragment to a multi-chain, 1,300 amino acid molecule.

Computational drug developers spend most of their time designing a particular type of antibody fragment, called the VHH, or “nanobody.” These are small variants of antibodies, naturally produced by llamas and alpacas, consisting of a single heavy chain. VHHs are much easier to design than a full antibody because they are about one-tenth the size and more compact. They can also be cloned and expressed in bacteria, unlike full antibodies, which require glycosylation, a chemical modification that only yeasts and mammalian cells can perform.

The main goal in binder design is to find a molecule with a high affinity to some target, meaning the binder latches on tightly and won’t let go. Such affinity is quantified by a dissociation constant, or Kd, which describes the concentration of free binder that must be present for a target molecule to have a 50 percent chance of being bound. A lower Kd means tighter binding. Picomolar (pM) and nanomolar (nM) Kds are typical for drugs, while micromolar (µM) is considered quite weak.

Semaglutide, for example, binds its target with sub-nanomolar affinity, while many natural signaling proteins, like T-cell receptors, have affinities in the micromolar range; a thousand-fold weaker. In the binder-design literature, a sub-micromolar affinity is often used as the threshold for calling a design “successful,” though this affinity would likely not be strong enough to be therapeutically useful.

To learn antibody design, most people start with BoltzGen, a tool developed by Hannes Stärk and the team at Boltz. (This is the same team behind Boltz-2, one of the leading AlphaFold 3-like structure prediction models.) BoltzGen is, arguably, the leading open-source approach for computational antibody design, and unlike many competitors, it uses the permissive MIT license, meaning anyone can use it, even commercially.

Working with multiple academic labs, the Boltz team showed that BoltzGen achieves sub-micromolar binders in a majority of cases, on both well-studied proteins like insulin and on difficult targets with no known similar structures. That said, BoltzGen is not the only option, and the landscape is shifting quickly. RFantibody can design full antibodies rather than just fragments, and Mosaic allows users to incorporate custom scoring functions.

However, none of these tools are easy to use. Designing an antibody on the computer requires navigating a huge tangle of jargon, a general understanding of the pros and cons of available software tools, and enough familiarity with wet-lab biology to know what to do after the design process has been completed. This guide walks readers through the full process of designing an antibody from home using BoltzGen.

Tool            Commercial  Full Ab  VHH  Mini-binder  Peptide
-----------------------------------------------------------------
BoltzGen            ✅        ✅      ✅       ✅           ✅
Germinal            ❌        ✅      ✅       ❌           ❌
mBER                ✅        ❌      ✅       ❌           ❌
RFantibody          ✅        ✅      ✅       ❌           ❌
Mosaic              ✅        ✅      ✅       ✅           ✅
BindCraft*          ❌        ❌      ❌       ✅           ✅

* BindCraft is not an antibody design tool, and is only included for reference. The FreeBindCraft fork allows for commercial use.

Antibodies from Scratch

The computational process involves five steps: choosing a target, preparing its structure, running a design campaign, filtering candidates, and experimentally validating the results. The guide follows a single hypothetical candidate and describes each step using the best open-source tools currently available.

1. Choosing a Target

This guide will use Nipah virus Glycoprotein G — “Nipah G” — as its protein target. Nipah G sits on the surface of the virus, a dangerous pathogen with a mortality rate of between 40 and 75 percent, and is essential for binding human cells. A binder that blocked Nipah G would presumably help prevent infection. (One antibody, m102.4, has already completed Phase I clinical trials on exactly this basis.)

Nipah G was also the subject of a recent binder design competition hosted by Adaptyv Bio, a cloud lab for protein designers. These competitions matter because they pit design tools against one another, chasing the same target under identical conditions. The Nipah G competition attracted a few hundred entrants, including the BindCraft developer, Martin Pacesa, and the Mosaic developer, Nick Boyd. Adaptyv screened more than 10,000 designs in total, making this arguably the richest publicly available dataset for comparing binder design tools to date.

This guide’s main tool, BoltzGen, did not fare well. In fact, only one percent of designs passed the experimental binding threshold. The reason for this is unclear, but every protein target behaves differently, and no tool excels across all of them. While Mosaic had the best results of any tool in the competition, with 8 out of 9 designs successfully binding the target, it requires more hands-on coding and is better suited for those already comfortable with protein design pipelines.

2. Preparing the Target Structure

Choosing a target protein is only the first step; you also need to choose a specific crystal structure to work with. A single protein can have hundreds of structures in the Protein Data Bank (PDB), each produced under different experimental conditions, with small but consequential differences in atomic positions.

Proteins also shift shape depending on their state. A protein bound to another molecule looks different from one floating freely in solution. These two forms, called the holo and apo forms, respectively, can have entirely different geometries.

The PDB is the main repository for protein structures, but its search engine is imprecise. Querying “Nipah virus Glycoprotein G” returns 32 results, many unrelated to Nipah, and the interface is hard to navigate unless you are a crystallographer.

A better repository is UniProt. If you search for your target protein there, you will find that the UniProt team has already done the hard work of linking every relevant PDB entry to each protein it contains. The entries, however, vary as to what resolution they have, which protein chains are included, and even how much of the protein is physically present in the structure. It is not uncommon for a crystal structure to cover only part of the full protein. In the case of Nipah G, the available structures cover positions 176-602. This excludes the transmembrane domain of the protein, since that region is notoriously difficult to crystallize.

Looking at the UniProt list, 2VSM seems like a reasonable choice of target. It has the highest resolution of the available options, at 1.8 Å, slightly longer than a carbon-carbon bond. (Anything below 2 Å is considered high resolution for X-ray crystal structures.) The 2VSM structure also includes the virus’s natural receptor, Ephrin-B2, which shows exactly where on the surface a binder might attach.

Once you have a possible target, see what prior research exists. In this case, there is a 2025 paper that identifies binding sites (or “hotspots”) on Nipah G suited to binder design. The first hotspot is located at residues Q559, E579, I580, Y581, and I588, exactly where the virus contacts Ephrin-B2 and where the clinical antibody m102.4 binds. The second site — at residues V235, S236, Y237, R555, and S586 — sits on another face of the protein. Antibodies that bind there do not block receptor attachment directly but prevent the virus from undergoing the conformational changes needed to enter a cell. The third site, formed by W504, F458, and L305, helps to stabilize the receptor-binding domain.

Even though 2VSM seems like a promising structure, it is still worth testing several options before committing. Small differences in atomic positions can affect the designs you get, and nearly identical structures sometimes yield distinct results for reasons that are not obvious in advance.

Predicting a Structure

When preparing the target structure, it’s important to remember that some proteins have no crystal structure in the PDB. UniProt often links to predicted structures directly, but if nothing suitable turns up, you will need to predict one yourself. For non-commercial projects, the best tool for this is AlphaFold 3.

AlphaFold 3 is straightforward to use. Log into alphafoldserver.com, enter your protein sequence, and wait a few minutes. You do not need to set any parameters. The output is a 3D view of the predicted structure and a downloadable .cif file, which is a modern version of the older .pdb format.

Protein structure prediction tools output a few different scores, or “confidence metrics,” though two come up most often. The first is pLDDT, which measures overall confidence in the predicted structure on a scale of 0 to 100. The second is ipTM, which measures confidence in the relative positions of two proteins in a complex, scored from 0 to 1. The Nipah G competition used a third metric, ipSAE, which is derived from ipTM and highly correlated with it, but tuned to better predict whether two proteins actually bind.

The AlphaFold 3 prediction for Nipah G bound to Ephrin-B2 looks good. Most of the structure shows a pLDDT above 90, the threshold for high confidence, and the ipTM is 0.9, meaning the model is confident about how Ephrin-B2 sits against Nipah G. The N and C termini score lower, but that is normal — those regions are disordered by nature. As a rule of thumb, a pLDDT above 90 and an ipTM above 0.8 are both good signs. This structure meets both of those cutoffs.

Trimming

Once you decide on a structure, it is often worth trimming it — that entails removing the parts of the protein irrelevant to the binding site. The cost of a design campaign scales roughly linearly with the combined length of the binder and target, so trimming irrelevant regions can save thousands of dollars.

The standard tool for this is PyMOL, a program for visualizing and manipulating protein structures. The open-source version can be installed via Conda, while the commercial version is available at pymol.org. Here are the specific steps you can take, using PyMol, to trim your target protein:

1. First, load the 2VSM.cif file into PyMOL by typing fetch 2VSM or clicking File / Open and selecting the downloaded file. This loads the complete complex, including Ephrin-B2.

2. To remove everything except Nipah G (chain A), type remove not chain A. To also show the protein sequence, type set seq_view, on. The red dots on the screen are mostly water atoms, which can be removed with the command remove hetatm.

3. Select and color the first hotspot with select hotspot1, chain A and resi 559+579+580+581+588; color red, hotspot1 so you can see where it is. Then, manually select residues (pink dots) by highlighting the sequence (here, I have selected positions 188 to 207.) Every amino acid after position 207 is quite close to the hotspot residues, so they cannot be safely removed.

4. Finally, remove the sequence with the command remove resi 188-207 and save the edited structure with save 2VSM_trimmed.cif and save 2VSM_trimmed.pdb. (It is often useful to have both .cif and .pdb files.)

In this particular case, removing 20 amino acids from a 400 amino acid structure is hardly worth the effort, since the Nipah G hotspot sits close to both ends of the protein. Residues can also be removed from the middle of a structure, creating gaps, though this requires more testing. Gaps can cause problems with structure prediction and change the numbering of hotspot residues in BoltzGen. For large antibody design campaigns, however, trimming as aggressively as possible is usually worth the effort.

3. Running a Design Campaign on Ariax

BoltzGen is not easy to run locally. Your options are cloning the GitHub repository and running it yourself, using a cloud service like Modal, or working through platforms like Tamarind Bio, Litefold, or Neurosnap. The BoltzGen team also recently launched their own platform, Boltz Lab, currently in beta.

This guide uses a web-based tool called Ariax, arguably the easiest way to run BoltzGen and BindCraft. Ariax charges directly for GPU time, without the need for a subscription, and the team behind it are protein-design experts whose explicit goal is to make binder design easier.

Start small and run short campaigns with 100 or fewer designs for each of your candidate hotspots first. This will give you a sense of which one looks most promising. Once you have identified the best hotspot, scale up to a full campaign — roughly 50,000 designs. (The right number depends on the difficulty of your target, your budget, and how many high-scoring designs you need. More designs means a higher chance of finding something that works.)

Setting up a run on Ariax takes only a few minutes. The Ariax team have written a helpful tutorial with additional tips, and the defaults are sensible enough for both beginners and experienced users, though not every BoltzGen parameter is exposed. If you wanted to use a custom VHH framework, for example, you would need to generate the appropriate configuration files and run BoltzGen yourself. The BoltzGen documentation covers this, though getting GPU-dependent tools like BoltzGen running locally can be genuinely challenging.

To run a Nipah G campaign on Ariax, follow these steps:

1. Select VHH as the design scaffold, then upload the 2VSM_trimmed.pdb file created earlier. Although .cif files have largely replaced .pdb as the standard format, .pdb is still the safer choice here as .cif files exported from PyMOL are not always compatible with BoltzGen. If you need to convert between the two formats, use maxit, a conversion tool maintained by the PDB team.

2. Under Binding Rules, set “binding” to positions 349, 369, 370, 371, and 378. Note that these are not the original residue numbers, and this is an awkward but persistent problem with both .pdb and .cif files. After trimming, the 2VSM structure starts at position 211, but BoltzGen always counts from position 1. Ariax helpfully notes in its log that it has applied an index shift of -210, so to match BoltzGen’s numbering to the original, subtract 210 from each residue number. Visually check the hotspot positions in Ariax, which highlights them in yellow, to confirm they land on the expected residues.

3. You can optionally set “not_binding” positions to steer the design away from parts of the protein you want to avoid — glycosylation sites, for example, or regions far from the binding interface. This option is most useful for larger proteins with multiple domains; for Nipah G, the binders tend to land close to the target site without much guidance.

4. Set the # Designs to 50 and the Budget to 50. This just means you want to generate 50 designs, and return information for all of them. At this stage, it’s important to see the full distribution of scores across all designs, rather than just the top hits. But for the full campaign, when you scale up # Designs to 50,000, you’ll set the Budget at 100 or below to get back just the top scoring designs.

5. For GPU Selection, Cost or Performance mode both work fine for a small run. The only obstacle that can arise is if the protein is too large to fit in the GPU’s memory — for proteins above 400 amino acids, only the latest GPUs, such as the B200 and H200, may be able to handle it. For the full campaign, select Turbo Mode to run multiple GPUs in parallel (otherwise, the run can take days.) There is no real cost penalty for Turbo Mode, and the faster premium GPUs are not necessarily more expensive overall since they finish sooner.

6. Finally, click Validate Settings to check for errors, then click Start BoltzGen.

Before the campaign starts, you will also need to fund your Ariax account. Generating the 100 Nipah G test designs costs around $10. A full 50,000-design run would cost roughly $3,000-$6,000 for a protein of this size, since Nipah G is around 400 amino acids in length. The main way to lower costs is to trim the structure further. (Remember, though, that any time you remove part of the protein, you risk subtly altering its shape.) You also need to keep track of how the residue numbering shifts as a result.

Scoring

Scoring and ranking designs is one of the most involved — and most confusing — parts of binder design. BoltzGen outputs many metrics, including ipTM, predicted hydrogen bonds, and surface accessibility, combining them into a single “Quality Score” using a heuristic formula. The team behind PXDesign did a comprehensive review of scoring across BindCraft and other tools, and landed on thresholds of ipTM > 0.85, pTM > 0.8, and complex RMSD < 2.5Å as a reasonable way to separate binders from non-binders. These thresholds work fairly well, though as far as anyone can tell, no single metric or combination of metrics reliably predicts affinity.

There is no consensus on the best way to score designs. ipTM is a far-from-perfect predictor of binding, but it is a robust enough metric for comparing designs across hotspots. Previous Adaptyv Bio competitions ranked designs by ipTM, and although the latest competition used ipSAE instead, the two scores are highly correlated. For our Nipah G example, the differences between hotspots are large enough to be genuinely informative; the first hotspot — the Ephrin-B2 contact site — seems to be best, with an average ipTM of 0.68, compared to 0.56 for the third hotspot and 0.4 for the second.

You can also run BoltzGen without specifying a hotspot at all, letting the model find its own binding site. This approach worked remarkably well for Mosaic in the Nipah G competition, and Nick Boyd discusses it in more detail on the Escalante blog.

For the full campaign, it is worth starting with around 1,000 designs to check the cost per design and confirm that the designed binders look sensible. If the top designs from that first batch are not engaging with the intended hotspot, something has gone wrong; adding “not_binding” residues can help steer the model away from wherever it is mistakenly landing. Once everything looks reasonable, scaling up is easy. On Ariax, just click “Clone & Reuse,” adjust the number of designs to 50,000, and start the run.

4. Filtering and Selecting Candidates

For this example, the campaign ran to 1,000 designs. But the process for evaluating the top designs is the same regardless of how many you generate.

The best binder from this run, ranked by BoltzGen’s heuristic score, has an ipTM of 0.78. ipTM is not an intrinsic property of the binder itself, but a metric of Boltz-2’s confidence in the predicted binding pose. A different model, generating exactly the same binder design, might give a different score for the same metric.

AlphaFold 3 is generally thought to be the best structure prediction model overall and especially strong with antibodies and nanobodies. For greater confidence, you can run the same binder design through alphafoldserver.com as an independent check. A “good” design will typically get good ipTMs from both Boltz-2 and AlphaFold 3.

In PyMOL, load both .cif files, align them, and compare the BoltzGen pose to the AlphaFold 3 pose against the Nipah G structure. Since Nipah G is in the PDB, both structures should align closely. In this case, BoltzGen gives the design an ipTM of 0.78, while AlphaFold 3 gives its slightly different pose an ipTM of 0.73.

Several things about this design are worth noting. First, the Boltz and AlphaFold 3 poses are very similar. (If the two models had predicted different binding poses, the design should probably be rejected.) Second, the binding pose is next to the hotspot, and binding appears to be driven by the VHH’s hypervariable regions (called “CDRs”). It is not uncommon to have designs where, instead, the VHH binds “side-on.” This is not necessarily a bad thing, but it reduces the probability that it is a specific binder, only binding to the Nipah G target.

Third, the ipTM scores seem reasonable. It would be better if both were above 0.8 or even 0.9, but 0.73 is a decent ipTM from AlphaFold 3. And finally, the amino acid sequence for the designed binder looks plausible. Sometimes, binder design tools will output suspiciously low complexity sequences, like long runs of glycines or glutamates.

Computationally-designed proteins may also contain stretches of cysteines, which are likely to cause aggregation or clumping. A more thorough check would involve running the sequence through an antibody language model like AbLang, which scores how “antibody-like” a sequence looks.

On balance, this design does not score especially highly, but it may still be worth testing. Further computational tests could be done, though the returns diminish quickly. The only way to know whether a design actually binds is to test it in the lab.

5. Experimentally Validating the Results

Once you have selected your most promising designs, the next step is to test them in the lab. Almost everyone working in protein design uses Adaptyv Bio for this — the same group that ran the Nipah G competition. Adaptyv is a modern contract research organization, which in practice means you can go online, submit a list of protein sequences, and pay by credit card without a sales call. Adaptyv makes a small quantity of each design in a cell-free system and returns binding affinity data against your target in a few weeks. It is the closest thing available today to a true cloud lab.

Each design costs $119-215 to test, depending on how many you submit. Assuming a full campaign of 50,000 designs and testing the top 50, the total comes to roughly $4,000 for compute and $12,000 for testing — plus a few hundred dollars for Adaptyv to acquire the target protein. Careful filtering before submission can reduce these figures.

Published results from groups like Nabla Bio, Chai Discovery, Latent Labs, BoltzGen, Germinal, and mBER suggest binder design success rates (defined as finding at least one sub-micromolar binder from ten tested designs) of up to 66 percent. But the messy reality, based on wet-lab testing, is that the true number is probably far lower.

The most comprehensive dataset available for a single target binder is the Nipah G competition, where the success rate was under 10 percent. Mini-binders appear to be easier to design computationally than VHHs and, anecdotally, the success rate for tools like BindCraft and PXDesign across a range of proteins is probably a bit above 25 percent. Generating high-affinity binders remains genuinely difficult, though, and success is highly target-dependent.

The lowest-cost version of this process that could plausibly produce a binder would be around $1,000 of compute — enough for 10,000 or more BoltzGen designs — followed by testing the top ten at Adaptyv, for a total of roughly $4,000. That is the floor, and even then there’s no guarantee a “true” binder will be found in those ten molecules.

The Last 90 Percent

With some iteration, this process should yield a VHH that binds your target at below one micromolar — a real result. That said, a working binder is closer to the beginning of the story than the end.

If the binding affinity is not high enough, improving it computationally is not straightforward. The most reliable approach is to brute-force a solution, perhaps using saturation mutagenesis or a similar method. This lets you exhaustively test sequences close to your binder in the hope that one or more show improvement.

Specificity is the next hard question. Does your binder attach only to its intended target, or does it stick to other proteins too? This is difficult to measure, partly because you cannot know in advance which off-target proteins to test. Testing even ten candidates at Adaptyv would be expensive, because the platform is set up to test one target against many binders, not the other way around. Each additional target protein would need to be sourced separately, from suppliers like Acro Biosystems, at a cost of $300-1,000 per protein.

Beyond affinity and specificity, there is also a long list of other properties to check, depending on how the antibody will be used. For therapeutics, it will likely need to be non-immunogenic, meaning it doesn’t fire up the immune system, and have a reasonable half-life, so it circulates in the body for a long time. For in vitro applications, like making a biosensor to detect botulinum toxin, there are far fewer requirements.

For drug development, most of the major costs come long after the binder design stage, emerging during clinical trials, for example. The opportunity exists, therefore, to optimize binding alongside other properties, like immunogenicity, thermostability, and half-life, during the computational design process. Theoretically, if you can optimize binding in tandem with these other properties, drug development costs could come down.

The real gains from these new computational tools probably lie, not in shaving weeks off existing timelines, but rather in making things possible that weren’t before.

David Baker’s group, for example, recently demonstrated “facilitated dissociation” — a binder that releases its target when a second ligand is added. Looking ahead just a year or two, it is not hard to imagine routinely designing binders that engage multiple independent targets, contain no immunogenic fragments by design, or change their structure and properties in response to pH or other environmental signals. That kind of complexity is only achievable through AI-guided design, and we are just at the beginning of it. Guides like this should facilitate the process.

Subscribe now

Brian Naughton is a computational biologist working in protein design and AI. He is co-founder and CTO of the personalized cancer company, Decade. Previously, he was co-founder and CTO at Hexagon Bio, and Founding Scientist at 23andMe. Brian has a PhD in Biomedical Informatics from Stanford University.

Cite: Naughton, B. “How to Design Antibodies.” Asimov Press (2026). DOI: 10.62211/58wh-12qp