Marc Tessier-Lavigne and the Long Search for a Better Map

Somewhere between New York and Oxford, on a ship full of Rhodes scholars, Marc Tessier-Lavigne changed his mind. He had left Montreal with a physics degree and a plan to pursue a doctorate in the subject. By the time the Queen Elizabeth 2 reached England in 1980, he had chosen a different pairing: philosophy and physiology. The first taught him to examine an argument; the second introduced him to the nervous system. A voyage with an itinerary had become a career with an unexpected destination.

That turn is unusually neat as an origin story, though his later career has resisted neatness. Tessier-Lavigne would study how growing neurons navigate the body, direct research at Genentech, run two major universities and then lead a company trying to make biology more predictable with artificial intelligence. Each move changed his tools and his colleagues. The recurring question was what gives living systems their direction, and whether understanding that direction can be useful.

A childhood in motion, a question about direction

He was born in Trenton, Ontario, in 1959, into a Canadian military family. His childhood crossed borders: Canada, Great Britain and Belgium. Mathematics was his favorite school subject, and he read about the history of science. He was the first in his family to attend university. At McGill, he studied physics, a discipline that promises order in equations. At Oxford, the nervous system offered a more unruly kind of order.

The attraction sharpened during his doctoral work at University College London. Under David Attwell, he studied how circuits in the retina function. Yet the question that stayed with him was earlier in the sequence. Before a circuit can do anything, its parts have to find one another. During a course in developmental neurobiology at Cold Spring Harbor Laboratory, he decided to focus on the formation of those connections. He later recalled the decision plainly: “I determined then and there to pivot and work on brain development.”

In postdoctoral work at Columbia with Thomas Jessell and Jane Dodd, he examined developing spinal cord neurons. Their fibers, called axons, can extend over long distances and still end up at remarkably specific destinations. The team showed that tissue along the spinal cord’s midline released a chemical attractant. It was a persuasive answer to the question of how a growing fiber knew where to go, and it set up an even harder question: what was the attractant?

At the University of California, San Francisco, his lab pursued the answer through years of experiments involving more than 25,000 embryonic chick brains. In 1994, Tessier-Lavigne and colleagues reported the molecules they named netrins. The name comes from a Sanskrit word for “one who guides.” It is a lovely piece of naming: a small word for a chemical signal with the practical job of pointing a cell in the right direction.

The discovery helped establish that developing neurons respond to both attractions and repulsions. His group and others went on to map more of that molecular traffic system, including receptors and additional cues. The work mattered because the nervous system’s elaborate wiring could be approached as a sequence of testable interactions. It also made him a prominent figure in neuroscience; the Gruber Foundation recognized his contributions with its 2020 Neuroscience Prize.

“The philosophy taught me how to think, and the physiology gave me my vocation as a neuroscientist.”Marc Tessier-Lavigne, recalling Oxford

From finding a signal to leading a research engine

A lab can identify a promising mechanism and still be a long way from a medicine. Tessier-Lavigne moved to Genentech in 2003 because he wanted to learn that journey from inside a drug company. He eventually became executive vice president for research and chief scientific officer, overseeing an organization of roughly 1,400 researchers. The change in scale was substantial. The questions were no longer only whether a molecule did something interesting, but whether an idea could survive the many stages between an experiment and a treatment.

He has described the experience as an education in both drug development and leadership. At Genentech, different specialists had to produce one decision from many kinds of evidence. That may sound like an administrative detail. In biotechnology it is the daily work: a compelling biological theory, a workable molecule and a sound development plan rarely arrive together by themselves. Tessier-Lavigne came to see helping other scientists do their best work as a scientific contribution in its own right.

He returned to academia in 2011 as president of The Rockefeller University in New York, a smaller institution built around intensive research. His tenure included a long strategic plan and a major campus expansion. In 2016 he took over as Stanford University’s eleventh president. The move placed a neuroscientist with industry experience at the head of a university that has long treated the border between laboratory and company as a busy crossing.

His Stanford years produced visible institutional work, including the creation of the Doerr School of Sustainability and a long-range vision announced in 2019. They also ended under scrutiny. In 2023, after allegations concerning papers on which he was an author, Stanford’s Board of Trustees released an independent scientific panel’s report and accepted his resignation as president.

The findings require care. The panel did not find evidence that Tessier-Lavigne personally engaged in research misconduct, or that he knew before publication about misconduct by others. It did find inappropriate data manipulation or deficient scientific practices by members of labs he oversaw, significant flaws in several papers, and instances in which he took insufficient steps to correct the record after problems emerged. He accepted the report’s conclusions and said he should have been more diligent in seeking corrections. He remained a Stanford biology professor on leave. This chapter is part of his career, as much as the discoveries and honors are.

The company with three languages

In 2024, he became co-founder, chairman and chief executive of Xaira Therapeutics. The South San Francisco company launched with more than $1 billion in committed financing, an unusual sum for a new venture. It was assembled around work in computational protein design from co-founder David Baker and his University of Washington team, alongside plans for large-scale biological data generation and drug development. Baker received the 2024 Nobel Prize in Chemistry for his work on computational protein design.

Xaira’s ambition rests on bringing together people who often use different vocabularies: AI researchers, experimental biologists and drug developers. Tessier-Lavigne has said that making those groups productive together is his chief challenge as CEO. The claim is less glamorous than a computer drawing a new molecule, but it is where the company’s thesis lives. An algorithm needs a question worth answering. An experiment needs a result that can refine the model. A drug program needs a candidate that survives testing beyond a screen.

One branch of the effort designs antibody-like molecules for targets that conventional methods have struggled to reach. The other tries to make models of cell biology useful for choosing the right targets in the first place. A model trained only on snapshots can learn correlations. Xaira is interested in what happens when a gene is deliberately changed and the rest of the cell responds. That difference, between observing a state and provoking a change, is central to its pursuit of causal predictions.

In June 2025, Xaira released a large public Perturb-seq dataset, X-Atlas/Orion. In March 2026 it introduced X-Cell, a virtual-cell model trained on its X-Atlas/Pisces genome-wide perturbation dataset. The company says the model can make predictions beyond its training data. Such claims are invitations to further experiments, not an endpoint. The usefulness of a prediction depends on whether it holds in cells, yields a sensible drug target and eventually helps a real development program.

“What’s really limiting in biology is data.”Marc Tessier-Lavigne, on building AI models for drug discovery

He makes that case without pretending the computer has replaced the bench. In a 2026 interview, he described verification as a constant requirement: AI-designed candidates must face the same checks of binding, specificity, biological behavior and development potential as any other candidate. The model may improve with each round of feedback. It still needs the feedback. Perhaps this is the most revealing continuity in his career. The netrin experiments did not begin with a finished map; they produced one through repeated tests of a difficult idea.

A map is only useful if someone follows it

There is a temptation to read each new platform as a clean break from the previous one. Tessier-Lavigne’s route suggests something messier and more human. The student who changed subjects on a ship became a scientist who spent years isolating a faint biological signal. The scientist became an executive who had to learn how many people and decisions stand between insight and application. The executive became a university president, then left that office amid findings that made the obligations of scientific oversight impossible to ignore.

At Xaira, he is again working at a meeting point of disciplines, with larger datasets and different machinery. The open question is whether its models can make discovery more reliable, especially where established approaches have failed. An ample funding round and a striking model release cannot answer that alone. The answer will come from experiments, programs and, in time, results that hold up outside the company.

Tessier-Lavigne once pursued the molecules that tell a young nerve fiber where to grow. Now he leads a company asking which signals inside a cell are worth changing and which molecules might do the job. It is a long way from that crossing to Oxford, but the instinct is familiar: when the path is uncertain, look closely for what provides direction.