Earlier Tuesday, Endpoints News reported that former Stanford president Marc Tessier-Lavigne will lead one of biotech’s biggest-ever startup launches: Xaira Therapeutics, which has secured over $1 billion to transform drug discovery and development with AI.
The move is sure to raise eyebrows. Tessier-Lavigne resigned from his post at Stanford last summer after an intense investigation found evidence of falsified data in some of his old research papers. Although Tessier-Lavigne was cleared of wrongdoing himself, the panel critiqued his lab oversight and said he was too slow to correct or retract the bad data.
A respected neuroscientist and drug developer, Tessier-Lavigne has hopped between the top ranks of academia and industry. He ran labs at UCSF and Stanford before joining Genentech to lead its research efforts, thinking he’d stay there forever. Instead, he went to Rockefeller University after it offered him the role of president in 2011, then to Stanford’s top post five years later.
After keeping a low profile in the nine months since the announcement of his resignation, Tessier-Lavigne is now back with the support from some of biotech’s most powerful investors, including Bob Nelsen, managing director of ARCH Venture Partners, one of the two main VC firms behind Xaira.
Endpoints’ Andrew Dunn and Ryan Cross talked to Tessier-Lavigne about his path to Xaira, why he believes the time is right for AI, his lessons learned from the Stanford controversy, and more. This interview has been substantially edited for length and clarity.
Andrew Dunn: We’d love to hear more about your relationship with Bob Nelsen.
Marc Tessier-Lavigne: I met Bob when I was brought onto the Agios board. It was a fantastic board. Bob stood out, however, because he just always keeps the big picture in mind. He wants to build significant companies for the long term to make a difference. And he stands by people, through the inevitable ups and downs that there are in companies.
Dunn: And he recruited you to Xaira through a fateful text?
Tessier-Lavigne: It was a text along the lines of “need to talk, big AI co, now.” It was that kind of urgency, out of the blue.
Ryan Cross: This is your first time as CEO. What made you want to not just jump back into drug development, but to lead a new startup?
Tessier-Lavigne: I’ve been approached at various times about leading a company, but the time wasn’t right. At this point, it’s an opportunity to leverage my experience in R&D in the industry and in running large organizations.
But it’s also a transformative moment. I do believe AI is going to change the whole drug discovery and development enterprise. And it’s going to change more rapidly than we think. This is an inflection point.
Dunn: You don’t have a typical “AI guy” background. What are your unique strengths and weaknesses running an AI-focused company, since you haven’t built foundational models or done the more in-the-weeds things?
Tessier-Lavigne: When I was running research at Genentech, I didn’t have deep domain expertise in all of the therapeutic areas that I was overseeing. But you learn how to build the right kinds of teams with the right kinds of people, and to ask the right kinds of questions. I’ve done it not just in drug discovery, but also in the academic sector, where I’m dealing with people with even more diverse backgrounds.
Cross: You’ve been around the research world long enough to have seen waves of excitement for AI come and go. What makes you think this moment is any different?
Tessier-Lavigne: This is a period where all the different pieces of the puzzle are going to be transformed. And they’re going to happen in a protracted way. Some will come rapidly; some will come more slowly.
The advances from the Baker lab, in terms of design capabilities, create a near-term opportunity that justifies the founding of a company like this. We have something to get started with immediately, which is the generation of large molecule therapeutics against targets that have been of interest but have been refractory to development.
Cross: What was the “a-ha moment” for you?
Tessier-Lavigne: Well, like everybody, I was blown away by RFdiffusion. (Editor’s note: RFdiffusion is the generative AI model from David Baker’s lab at the University of Washington for designing proteins from scratch. Baker is a co-founder of Xaira.)
But like most drug developers or people with drug development experience, I understood that the binders you get from that are going to be fantastic tool compounds, but not yet drugs. It’s the fact that they were able to go the next step and essentially constrain the program to generate antibodies or antibody-like molecules that was a really big step forward.
Dunn: On your time at Stanford, the investigation found no evidence of data manipulation or fraud on your part. But it did criticize your laboratory oversight and management, and you acknowledged there were things you should have done better. What specifically do you plan to do differently at Xaira in terms of supervising science? And why should you be trusted in that type of position?
Tessier-Lavigne: Well, first, I’m obviously here to talk about Xaira. But maybe I can share a few thoughts.
First, I stand by the quality and rigor of my body of research as a whole. I’ve run a research lab for over three decades with 74 papers which I’m a principal author, and a handful of them had issues, as you pointed out. The panel concluded that I didn’t engage in any scientific misconduct, and I could not have reasonably been expected to have known about it.
And in terms of the lessons I take from this, I have run my lab by putting trust in my team. I’ve had exceptional lab members. I trusted that the data they were presenting to me were real and accurate. In general, that trust was entirely merited. But that created a vulnerability that allowed for these issues to arise in a few papers. What I take from that is the importance of also having an appropriate level of skepticism about data that are brought.
Dunn: Are there specific changes you plan to incorporate in overseeing and leading Xaira?
Tessier-Lavigne: Certainly. What this has driven home for me is the importance of having controls in place to ensure that your data that come forward are real and accurate, and not to take that at face value. And it’s certainly a principle that I’ll be applying here.
Cross: Is there an extra impetus to have those controls in place in the AI field, where it’s very hard for any one person to understand all the aspects, both the computational and biological sides of things?
Tessier-Lavigne: I think the answer is yes. In a company like Xaira, there will, of course, be feedback loops here in the kind of work that we’ll be doing with AI. Predictions will be made; they will be tested in the laboratory and the information fed back. So there’s almost intrinsically a verification step for a lot of the things that we’re going to be doing. That’s different, perhaps, from some of the other applications of AI that we see, say, in the consumer space.
Cross: We’ve seen some richly funded startups like Altos Labs and Arena BioWorks lure top scientists from academia to create these hybrid research institutes and biotech companies. As someone who’s been at the top of both of these worlds, how do you view these changing relationships between academia and industry? And how will that change the way that science gets done?
Tessier-Lavigne: It’s an exciting time where a lot of different models are being tried and tested. Many academics still prefer to be able to focus on pursuing their own ideas. I think it’s a positive development that there are these other research models that make it possible to tackle problems, sometimes at a scale that can’t be achieved with the traditional academic structures.
I’m a big believer in having a diversity of approaches to tackling scientific questions. And so I love the ferment. We have to make sure that we also always support the individual investigator in universities, because at the end of the day, they will be training the graduate students. They will be pursuing new ideas that will lead to the transformative advances of tomorrow.
Cross: Thank you for your time, we really appreciate it.