Bo Wang, Ph.D., is a Xairanaut.
That is, he’s the latest high-profile scientific leader to join the unicorn Xaira Therapeutics, taking the helm as SVP and Head of Biomedical AI.
Wang has experience as Assistant Professor in the Departments of Computer Science and Laboratory Medicine & Pathobiology at the University of Toronto, Temerty Professor in AI Research and Education in Medicine, Chief AI Scientist at the University Health Network (UHN), and CIFAR AI Chair at the Vector Institute.
Explaining the timing of his move to industry, Wang highlighted a significant shift underway: “And this is also an inflection time where I see the whole research field, particularly AI for biology, shifting towards data-driven large-scale model development,” Wang said in an interview.
From foundational models to a new frontier
Bo Wang, Ph.D.
Wang was also active in genAI research before it was cool. That is, Wang helped develop single-cell GPT (scGPT), a foundation model for single-cell biology featured in Nature’s “Seven Technologies to Watch in 2025″ list. “We actually established the research [for scGPT] before ChatGPT was announced,” Wang noted.
He explained the insight behind the model, which was trained on 33 million diverse human cell profiles: “The motivation was quite clear: words/sentences can be described as a collection of words with the right order, and cells can be described as a collection of genes,” Wang noted. “So with this similarity in formation, we adapted the transformer/foundation model idea into the single-cell domain, and it turned out to be extremely effective.” The impact has been significant, with Wang describing scGPT as “extremely revolutionary for this field,” adding that it is now “widely used in different drug discovery companies and has been set as a benchmark method for single-cell research.”
The scGPT paper published in Nature Methods highlights the model’s ability to “distill critical biological insights concerning genes and cells” and its optimization for “perturbation response prediction and gene network inference.” This capability to “predict unseen perturbation responses could expand the scope of perturbation experiments,” the authors wrote, enabling researchers to “leverage the knowledge gained from cellular responses in known experiments and extrapolate them to predict unknown responses.”
The vision: Building the world’s first virtual cell
This foundational work directly informs Wang’s vision at Xaira. “When I met with the leadership of Xaira, and they asked what I wanted to do, my answer was simple: I want to build the first virtual cell in the world,” Wang revealed in the interview.
“My answer was simple: I want to build the first virtual cell in the world.”
— Bo Wang, Ph.D.
He described the concept as a goal being pursued by the broader research community: “aiming to build a new type of virtual cell enabled by Large Language Models. It’s a data-driven, language model-backed system that can model gene expression, perturbation response through in silico modeling.”
The potential is significant, Wang explained: “Imagine you have control/healthy cells and want to understand the response to interventions like CRISPR or drugs. Instead of doing expensive wet lab experiments, you can prompt the virtual cell, and through its AI capabilities, it can predict the response.” Ultimately, this could allow researchers to model not just healthy responses but also disease states and how they might react to new drug candidates. He believes Xaira provides the unique environment needed for this undertaking: “I think Xaira can provide the necessary resources… building upon existing approaches like scGPT and the sequence prediction work from David Baker’s Lab. I’m confident with the right investment and datasets, we can push out this first virtual cell.”
Bridging the gap: Grounding AI in biology
But building a truly predictive virtual cell hinges on overcoming a fundamental limitation, Wang emphasized: ensuring the AI is firmly grounded in real biology. “The key challenge, I think, is that most current AI models lack clinical and biological grounding. They are trained on certain public datasets… and then they are expected to generalize,” Wang explained, highlighting a critical problem his work aims to solve.
“AI provides prediction, the wet lab provides validation, and this validation further improves the AI predictions.”
— Bo Wang, Ph.D.
To address this fundamental limitation, Xaira is implementing a unique approach. “This is what brings uniqueness to the Xaira approach: the combination of wet lab data with computational or dry lab experiments. I keep mentioning this AI prediction-validation loop: AI provides prediction, the wet lab provides validation, and this validation further improves the AI predictions,” Wang emphasized. “We want to design a wet lab to help not just test hypotheses, but to generate informative data that improves the model performance.”
Beyond the iterative feedback loop, Wang envisions a comprehensive platform spanning the entire drug development pipeline. “What’s unique to Xaira… is really to provide multimodal and generalizable AI models connecting the very upstream (target identification) to the very downstream (clinical trial patient stratification) altogether,” Wang noted.
Xaira’s ‘three pillars’: Talent, compute, and data
Wang emphasized that this concentration of expertise, combined with resources and data generation capabilities, was a core reason for joining. “Right now, it is well received that there are three key pillars for any AI success nowadays: One is talent, two is compute, three is data,” Wang explained. “Xaira is a very unique platform to enable access to these three pillars. Obviously, we have huge investment funding into Xaira that enables us to recruit the top talent… Xaira hits the three buckets altogether.”
Wang arrives at Xaira nearly a year after its launch in April 2024 with over $1 billion in committed capital – an amount described by Robert Nelsen, Managing Director and Co-Founder of lead investor ARCH Venture Partners, as “the largest initial funding commitment in ARCH history.”
He joins a company spearheaded by CEO Marc Tessier-Lavigne, Ph.D., former President of Stanford University and Chief Scientific Officer at Genentech. Xaira was co-founded by Dr. David Baker, Ph.D., who shared the 2024 Nobel Prize in Chemistry with DeepMind’s Demis Hassabis, Ph.D., and John Jumper, Ph.D., for work in AI-driven protein structure prediction and design. Wang’s appointment follows a series of strategic hires over the past year as Xaira built out its leadership, including CSO Debbie Law, D.Phil. (formerly SVP at BMS), CMO Paulo Fontoura, M.D., Ph.D. (formerly Global Head of Neuroscience Clinical Development at Roche), and CTO Hetu Kamisetty (formerly of Meta and Baker’s Institute for Protein Design).
A confluence of expertise: Collaboration and team building
When asked about the opportunity to collaborate with Nobel Prize winner David Baker, Wang was enthusiastic: “Dr. David Baker is a top expert in the protein design space. I’m very excited to work with his team on enhancing some of the AI models for protein design, antibody design,” Wang shared: “We were also discussing a future direction to combine sequence models… with expression models… How do we connect the two spaces is something we are very excited to explore.”
Central to Wang’s vision is assembling the right team for this interdisciplinary challenge. “Definitely recruiting people who think across different disciplines: machine learning scientists who speak the language of biology, experimentalists who understand AI,” Wang said.
This concentration of C-suite talent is mirrored by the Board of Directors, featuring fellow Nobel Laureate Carolyn Bertozzi, Ph.D. (Stanford), former FDA Commissioner Scott Gottlieb, M.D., former Johnson & Johnson CEO Alex Gorsky, ARCH’s Robert Nelsen, and Parker Institute for Cancer Immunotherapy founder Sean Parker. Further reinforcing the blend of expertise is the Scientific Advisory Board, which includes Baker alongside AI researchers like Anima Anandkumar, Ph.D. (Caltech/NVIDIA) and Regina Barzilay, Ph.D. (MIT), and biology experts such as Sarah Teichmann, Ph.D. (Wellcome Sanger Institute).
Filed Under: machine learning and AI
