Bo Wang is a renowned AI scientist at the University of Toronto. He’s bringing his open-source culture and computational biology to Xaira Therapeutics in June.
Every time you want to scale up an AI model, it costs money. Big money, according to University of Toronto professor Bo Wang, a machine learning and computational biology specialist. At some point, academic labs like his can’t keep up. That’s why industry has been behind some of the more exciting advances in AI of late, Wang said.
Luckily for Wang, industry came calling just as he was granted tenure with two precious years of sabbatical. He will step into one of the buzziest AI biotechs in the game come June as the head of biomedical AI at Xaira Therapeutics, seeking to unite predictive AI and validation technologies to find new drug candidates.
“This field became messy . . . in the sense that people publish their model without comprehensive validation,” Wang said. “Some companies just share lots of predictions from the model, nobody knows how to use it. Do you trust them?”
That challenge gets Wang up every day, and Xaira, which broke cover last year with its $1 billion mega-raise but has effectively been operating effectively in stealth, will be his next home to work on the problem.
Back to Business
The new role will mark a return to industry of sorts for Wang, who did an internship at Genentech and a stint at Illumina after earning his PhD from Stanford University. In addition to his position at the University of Toronto (U of T), Wang is the chief AI scientist at the University Health Network, Canada’s largest health research organization, and the AI chair for the Canadian Institute for Advanced Research at the Vector Institute—a group with a mandate to “convene extraordinary minds to address the most important questions facing science and humanity.”
“Since chatGPT was released, the AI landscape is totally different. AI is developing so fast, and I do see lots of limitations as an academic lab,” Wang said. “Returning to industry will give me access to a large amount of resources.”
So how did he get wrapped up in a biotech unicorn? Wang says he heard about them just as everyone else did. But the company piqued his interest because it seemed to be doing research similar to his own, just at a much bigger scale. He often travels to San Francisco, he says, and was invited by one of the company’s vice presidents to give a presentation.
Wang presented, among other things, scGPT, which is a foundation model for single-cell multi-omics that he developed with others at U of T, UHN and the Vector Institute. He learned that Xaira was doing something similar but had found the work “very tricky.” Wang and the executives at Xaira quickly realized they needed each other.
Wang was particularly impressed by Xaira’s abilities in analyzing large scale perturbation data, which is sensitive data such as health records that has been altered slightly for enhanced data privacy. “This is exactly what I needed to extend my research such as scGPT to the next level, so that got us talking a lot,” he said.
When he arrives at Xaira in June, Wang will be in charge of the biotech’s AI strategy across the R&D pipeline, including developing a multimodal foundation model for biology, building connections between the wet lab and clinical team to ensure meaningful experiments are conducted and, perhaps most importantly, hiring a team of people.
“One of Xaira’s superpowers, in my view, is its integration. We have our wet lab, we have our computational modeling, we have a clinical team working hand in hand,” Wang said. “We are scaling very fast, so I’m helping build a team that is as comfortable debugging a neural network as they are collaborating with a bench scientist. So this kind of hybrid talent is key to making our vision real.”
Xaira hasn’t had too much to say since it burst onto the scene last April, other than the occasional hire like today’s announcement of Wang. And Wang said Xaira is still not ready to reveal its pipeline. Butthat day will come, he added—and not just if the company has a success story to share.
“Many biotech are very secretive about their findings, and they only make announcements when things are quite successful, which I think is very much understandable,” Wang said. At Xaira, he continued, there are already plans to make some of the biotech’s perturbation data open source, just like the famed ChatGPT that has taken over aspects of everyday life.
One of Xaira’s founders, David Baker, professor of biochemistry and director of the Institute for Protein Design at the University of Washington School of Medicine, is also an advocate for open-source data. Baker has opened up access to his lab’s protein structure prediction models and other key work in the name of advancing collective science.
This, Wang said, is the legacy of the academics like him who are streaming into industry to drive forward the next phase of AI as they bump against those academic limitations.
“I hope to bring that culture into Xaira so that we are more open to sharing our models and data,” Wang said. “Certainly, they have to be aligned with the company’s strategy, but I do see this shift in terms of culture.”
Xaira also stands out among a sea of AI biotech companies because it connects predictive technology with the ability to run great experiments. According to Wang, most companies focus on one or the other. “The future lies in unifying the two.”
Without aligning prediction and validation —and being able to show your work—AI drug discovery companies will be plagued by too many candidates with the same challenge as traditional drugs, Wang said: you still have to test them in humans.
For now, Xaira is staying mostly mum. Wang said they just need time to show the world what’s behind the curtain. “We’re backed with sufficient funding. So give us time, horizon and ambition to build something really enduring.”
