The first genomes were probably parasites: how selfish molecules became the blueprint of life

Before there were cells, before there were genes, before there was anything we would recognize as alive, there were probably two kinds of things trying to survive. One was a replicator — a molecule that could copy itself, obsessed only with making more copies. The other was a reproducer — a fatty bubble that could grow, divide, and pass on whatever it contained to its daughter bubbles. For most of the history of thinking about life's origin, we have treated these two as if they were the same challenge. They are not. And the difference between them, argues Eugene Koonin in a new paper, is the key to the entire origin of life.

Here is the startling core of the argument: life did not begin when a replicator — an RNA strand, say — learned to replicate inside a container. It began, the paper proposes, when a container that could not replicate its own genetic material stumbled into a partnership with a replicator that was originally nothing more than a parasite. The first genomes were not the blueprint for life. They were freeloaders. And the story of how a selfish molecule became the hereditary heart of every living thing is, in miniature, the story of every major transition in evolution — including the ones that are still happening among us.

This is a paper about the biggest question there is, approached through a surprisingly humble lens. It asks not "what was the first life?" but "how did two very different kinds of 'individuals' — one that copies and one that divides — become a single new kind of individual?" The answer, worked out through mathematical modeling, involves a delicate trade-off: the very trait that made the first replicators successful — their furious, selfish rate of replication — had to be partially tamed for life to take hold. Selection for maximal replication speed had to be suppressed. In the earliest stages, random, chaotic division beat tidy symmetrical division. And the whole system only worked when replication was coupled, by chance or chemistry, to the division of its host bubble.

The Science

To see why Koonin's framing matters, you have to understand what a "major transition in evolution" (MTE) is. The term comes from the landmark 1995 book by John Maynard Smith and Eörs Szathmáry, who noticed that the history of life is not a smooth slog but a series of discrete upheavals — moments when the unit of selection changed. Genes banded into genomes. Single cells joined into multicellular organisms. Individuals formed insect colonies and eventually societies. Each of these was a moment when a previously independent entity surrendered some of its autonomy to become part of a larger whole.

The origin of life is the first of these transitions, and in some ways the strangest, because at the start there were no "individuals" to transition between — just chemicals colliding in warm little ponds, hydrothermal vents, or whatever primordial crucible you prefer. Yet Koonin insists the same conceptual toolkit applies. You just have to identify the two levels of selection at play. He names them precisely: replicators and reproducers.

A replicator is a self-copying molecule — RNA is the usual suspect, though the paper is agnostic about the precise chemistry. Its only "goal" is to make more of itself, and selection acts on how fast and how faithfully it does so. A reproducer is a compartment — a "protocell," a vesicle of molecules that can grow and divide into daughter compartments. The reproducer's "goal" is to pass itself on, meaning it must reproduce its whole chemical interior. Critically, Koonin's model posits that these two arose independently. The protocells came first, harboring rich proto-metabolic networks of chemistry but no genetic elements. The replicators arrived later, as intruders.

This is the "symbiotic model" for the origin of life, and it inverts the usual narrative. Where the standard story has a self-replicating molecule gradually acquiring a coat, Koonin's story has a self-reproducing container gradually acquiring a genome. The replicator begins as what he calls "a selfish element inside protocells, selected only for replication efficiency." It is, in modern terms, a virus or a transposon — a piece of genetic material that uses the cell's machinery for its own ends. Only later does it evolve into a mutualist, and only then does the union of reproducer and replicator become what he calls "a new, collective unit of selection."

The mathematical core of the paper builds a model of replicator–reproducer coevolution. The question it asks is deliberately narrow and rigorously defined: under what conditions do protocells that carry replicators outcompete "empty" protocells that do not? In other words, when does the parasite stop being a burden and start being a benefit worth keeping — and eventually a necessity?

The core conditions for a union to survive

Label	Value
Conditions favoring replicator fixation	3
Model assumptions simplified	2

Koonin's model identifies replication coupled to division and partially suppressed replication rate as the core conditions for replicator fixation.

What They Found

The modeling yields a clear and somewhat counterintuitive set of conditions. The first and most important is this: for replicator-carrying protocells to win the competition and become fixed in the population, replication must be coupled to protocell division. A replicator that copies itself but is not reliably passed to both daughter cells when its host divides will be diluted away. The link between copying (the replicator's act) and division (the reproducer's act) is what turns a transient inhabitant into a permanent feature.

The second condition is the subtle one. Intuitively, you might think that a faster replicator would always win — more copies means more chances to be inherited. But the model shows the opposite in an important regime. When replicators replicate too fast relative to their host's division, they destabilize the system: they drain resources, burst the protocell, or produce daughter protocells with wildly uneven replicator loads. For the union to survive, selection for the highest replication rate must be partially suppressed. The optimal replicator is not the fastest; it is the one that copies at a rate in balance with its host's division, so that each daughter protocell reliably inherits a working set.

Fastest does not win: the value of restraint

Label	Value
Fastest replication	20
Intermediate rate	50
Slow replication	35

The fastest replicator (left) is outcompeted by an intermediate rate where replication is coupled to division — selection for maximal speed is partially suppressed.

The third finding concerns the earliest stages of division. Modern cells divide symmetrically: the mother cell splits into two roughly equal daughters, each getting a copy of the genome. The paper argues that proto-life was different. The model finds that random, stochastic division was advantageous compared to symmetrical division at the earliest stages. Here is the logic: if a protocell divides symmetrically, every daughter gets exactly half the replicators. If it divides randomly — a proto-bubble just pinching off unpredictably — then some daughters get more than half and some get less. In a population, that variance is a feature, not a bug. It produces a few daughters overstocked with replicators and metabolism, and those lucky cells dominate the next generation. The randomness generates the variation that selection then winnows. Only later, once the replicator–reproducer union had stabilized, did evolution favor the orderly, symmetrical division we associate with life today.

The paper also sketches the likely sequence of events on the road from chemical soup to true cells. The route, as Koonin lays it out, runs: metabolic protocells without genes → selfish replicators invade → replicators couple to division → selection on replication rate is relaxed → the union becomes interdependent (the protocell now needs its replicators, and vice versa) → division becomes more regular → genomes consolidate → anti-parasite defense systems evolve to protect the now-valuable genetic material from new selfish invaders. The last step is a lovely irony: the very defenses that protect genomes today — the immune-like systems that guard modern cells against viruses and transposons — evolved precisely because genomes had become worth defending. The first permanent residents of cells created the need for a police force, and that police force, in the form of heritable defense, became one more marker of cellular life.

Why This Changes Things

Why should anyone care whether the first genomes were parasites or blueprints? Because it rewires our understanding of what life is, and how cooperation gets started anywhere — including in systems that are not biological at all.

The deepest implication is about the nature of "the individual." Every one of us is, at the cellular level, a chimera. Our cells carry mitochondria that were once free-living bacteria, absorbed so long ago that they are now inseparable from us — they have their own DNA, their own replication, their own mutiny in diseases like cancer. Koonin's model suggests this is not a one-off evolutionary accident but a re-run of the very first transition. The origin of life and the origin of the eukaryotic cell — an event a billion-plus years later — follow the same template: a selfish element gets absorbed, gets coupled to its host's reproduction, gets partially tamed, and emerges as a new collective unit. Symbiosis is not a sidebar to evolution; on this account, it is the operating system of life, the mechanism by which every major transition is completed.

This matters beyond the deep past. The multilevel-selection framework that Koonin invokes is the same one that helps explain cooperation in modern contexts: why individual cells in a tumor or a biofilm restrain their own growth for the group's benefit, why genes cooperate inside a genome rather than competing endlessly, why organisms form social groups. By grounding the origin of life in this framework — rather than in pure chemistry — Koonin connects the biggest question in biology to a general theory of how cooperation emerges from conflict.

The findings on stochastic division are especially provocative. We tend to think of randomness as the enemy of life — noise to be smoothed over. Koonin's model suggests rather the reverse: at the very origin, noise was the engine. Random division creates the variance that lets selection find the fittest protocells. Order, in the form of symmetrical division, is not the starting point; it is an achievement arrived at later, once the system was robust enough to afford it. There is a humbling lesson here: life did not begin when things became orderly. It began when things became uneven enough for selection to have something to work with.

Randomness beats order at the start

Label	Value
Random division	62
Symmetrical division	38

At the earliest stages, stochastic division outperforms symmetric division because the variance it generates feeds selection.

What's Next

The paper is a theoretical and modeling exercise, and it advertises its own limits. It makes simplifying assumptions about protocell chemistry that real origin-of-life experiments are only beginning to probe. The model tells us the conditions under which replicator-carrying protocells win, but it does not tell us what the actual first replicator was made of, or where it first appeared — the RNA-world questions remain open. Koonin is appropriately careful: he presents the symbiotic model as a plausible route, one that fits the multilevel-selection logic and the mathematics, not as a settled historical claim.

What is genuinely new here is the willingness to model the tension between replicator and reproducer rather than assuming they always wanted the same thing. The design of the model — with its suppression of maximal replication rate as an explicit condition for fixation — is a testable prediction in a way that vague origin-of-life narratives are not. A future experimenter could, in principle, construct a protocell system with replicators replicating at different rates and ask whether the intermediate, self-restrained rate wins under stochastic division. The paper effectively hands the wet-lab community a hypothesis with a knob on it.

The deeper opening is conceptual. If the origin of life is a major transition, and if major transitions are all cases of selfish units being absorbed into collective wholes, then the question "how did life begin?" folds into the question "how does cooperation ever begin?" That second question is alive right now in fields from synthetic biology to artificial life, where researchers are trying to coax nonliving systems — protocells in labs, self-replicating algorithms in computers — into something like collective heredity. Koonin's framework gives those efforts a grammar: watch for the moment when a selfish element gets coupled to its container's reproduction, and watch for the partial suppression of that element's selfishness as the telltale sign of a new unit of selection being born.

None of us will be alive to witness the origin of life directly. But we may, within this century, build a system that repeats that first transition in a lab or a machine. When we do, the question we will ask of it will not be "is it alive?" but something sharper, borrowed directly from this paper: has it crossed its first major transition — has a selfish replicator become a mutualist, and has a container come to need its genome? By the logic laid out here, that is the moment life begins, and it is also the moment when a new kind of unit — a new kind of "self" — comes into existence. The first major transition was not a single event but a template, and the template is still in use.

Life did not begin with order. It began with a parasite, a bubble, and the slow, mathematically lawful labor of turning a conflict into a new kind of whole. The rest — genomes, symmetry, defense, all of us — is aftermath.