Fireside Chat, Interview, Conference Presentation
How AI is Driving Drug Discovery: Xaira Therapeutics' Marc Tessier Lavigne
- The company anticipates transforming drug discovery from an empirical search to a design process, aiming to reduce the current ~10% clinic entry success rate and the associated $2–4 billion development costs per drug.
- Strategic plans involve leveraging a billion-dollar funding round to build a high-throughput biology system supporting three areas: antibody drug design, biology foundation models, and patient representation models.
- The company will address target identification, drug design, and patient selection simultaneously, initially applying models to easily screened targets before focusing on "undruggable" targets difficult for current methods to address.
- Predicted outcomes include fewer experimental iterations, higher success rates, and the ability to choose patients and targets with greater precision, potentially reducing timelines from the historical 5–50 years to 2–5 years.
- The company expects to make significant improvements in drug design effectiveness and patient selection success within three to five years by combining frontier AI, massive-scale biology, and drug discovery experience.
- A philosophy of continuous advancement will be adopted where models are updated constantly based on current technology without waiting for perfection, balancing immediate pipeline progress with the pursuit of the best available technology.
- The approach intends to overcome structural industry issues by generating massive data synergies to enable the design of therapeutics for targets previously considered impossible to treat.
- The company acknowledges that while AI technology evolves rapidly, biology and regulatory environments move slower, requiring a staged approach that accepts current technology capabilities while preparing for future advancements.
- The company predicts that the integration of AI will enable the precise activation or deactivation of specific cellular mechanisms, shifting drug discovery from an artisanal process to a systematic, AI-driven one.
- Future capabilities will allow for the identification of patient populations most likely to respond before development begins, thereby preventing failures caused by targeting incorrect demographics.