He is building the world's first digital organism - a living roundworm reconstructed neuron by neuron inside a computer. The bet: if you can simulate the simplest brain on Earth, you take the first real step toward understanding our own.
There is a roundworm called Caenorhabditis elegans. It is about a millimeter long, transparent, and almost embarrassingly simple - 959 cells, 302 neurons, roughly 10,000 connections between them. It is also the most completely mapped animal in the history of biology. Stephen Larson looked at that map and asked a question most scientists file under "someday": what if you rebuilt the whole thing, alive, inside a computer?
That question became OpenWorm, the open-science project Larson co-founded in 2011 and still directs. The goal is not a cartoon of a worm. It is a working digital organism whose every cell and signal follows the same rules the real animal does - a creature that exists as code because the code captures how life actually behaves. More than 100 scientists, programmers, and designers from around the world have contributed, making it one of the most ambitious volunteer science efforts on the internet.
By day, Larson runs MetaCell, the life-science software company he co-founded and leads as CEO. MetaCell builds the tools that let researchers at pharmaceutical giants and elite universities visualize, analyze, and collaborate on biological data that has grown too big and too tangled for spreadsheets. The two pursuits - the open worm and the for-profit software - feed each other. Both come down to the same conviction: biology is an information problem, and we finally have the machines to treat it like one.
What makes Larson unusual is that he speaks two languages fluently. He trained at MIT in computer science and artificial intelligence, then went to UC San Diego for a PhD in neuroscience. Most people pick a side. He sat down in the gap between them and started building.
His MIT master's thesis was about self-organizing knowledge representations - how a machine might arrange what it knows without a human laying out every shelf. Read it next to OpenWorm and the through-line is obvious. He has always been chasing the same thing: systems that organize themselves into something more than the sum of their parts. A worm is just the most honest test case nature offers, because the worm was never going to cheat.
Scientists have spent half a century taking C. elegans apart. It is the only animal whose wiring diagram - its connectome - is fully known. That makes it the rare case where a complete simulation is even thinkable.
Knowing where every neuron sits is not the same as knowing what the animal will do. A wiring diagram is a still photograph; behavior is the movie. Larson's team spent years translating the worm's biology into differential equations - the language that lets a computer turn structure into motion.
They built a simulation platform named Geppetto - after the woodcarver who wished his puppet into a real boy - to stitch those equations together. The name is a small joke with a serious edge. The whole point is to make the model move on its own.
“We can observe biology by isolating... but how neurons come together to create the whole behavior? We can't see the forest for the trees.”
In 2015 the team did something that sounds like science fiction and reads, in the lab notes, like a Tuesday. They took the worm's connectome model and loaded it into a small LEGO robot. No one wrote code telling the robot to avoid walls. No one scripted a single turn.
The robot rolled forward, met a wall, backed up, and changed direction - behaving like a worm because, in the only sense that matters here, it was running a worm. The behavior emerged from the wiring, not from instructions. For a project built on the belief that life is a set of rules you can re-run, it was about as close to proof-of-concept as a millimeter of roundworm can get.
Larson did not start in a lab. His first job out of MIT was writing software for Morgan Stanley's equity research desk in New York - building systems for an industry that turns information into decisions at speed. The instinct never left him; he just pointed it at cells instead of stocks.
The pivot was almost theatrical. On August 3, 2010, he pitched the idea of simulating C. elegans at Ignite San Diego - a format that gives you five minutes and not a second more. Early the next year he put out an open call for volunteers, teamed up with Giovanni Idili and Matteo Cantarelli, and proposed a name for the thing they were about to attempt: OpenWorm.
OpenWorm could have been a guarded lab project, a patent waiting to happen. Larson made the opposite choice. The code is public, the data is public, and the contributors are scattered across continents, mostly volunteers, mostly people who showed up because the problem was beautiful. That decision is not sentimental. It is strategic. A single lab cannot build a whole organism. A crowd that can see every line of code might.
It also reflects a habit. Before OpenWorm, Larson helped create NeuroLex and the Whole Brain Catalog - open neuroscience resources from an era when "open data" was a fringe idea rather than a funding requirement. He served as CIO at One Mind for Research. The pattern is consistent: build the shared infrastructure first, then let other people do things with it that you would never have thought of.
His work has been published in journals including Frontiers and Nature, and covered by the New York Times, Wired, Discover, and MSNBC. He has presented at more than two dozen forums and holds a patent. But the metric he seems to care about most is whether the worm, eventually, comes alive on the screen - and whether anyone, anywhere, can download it and check his work.
MetaCell calls itself a world leader in software for life science, and its client roster makes the claim hard to argue with. The company builds data-visualization, cloud, and analysis tools - calcium imaging analysis, microscopy data management, cloud workspaces - for groups drowning in the kind of data that modern biology produces by the terabyte.
Larson co-founded it with Idili and Cantarelli, the same partners from the worm. The mission statement is plain: accelerate breakthroughs in research and treatment, and make a meaningful difference in people's lives. The open-source worm proves the idea; the company puts it to work.
The simulation engine is named Geppetto - the woodcarver who wished his puppet alive. The whole point is to make it move on its own.
He is bilingual in a rare way: MIT for code and AI, UCSD for neuroscience. Most people pick one. He works in the seam.
Before worms, he wrote software for Morgan Stanley's equity research desk on Wall Street.
He also helped build NeuroLex and the Whole Brain Catalog - open neuroscience resources from before "open data" was a slogan.