Hybrid protein language models for functional protein design
Speaker
Pascal Notin
Harvard Medical School
Time
Venue
Online
Meeting Info
Abstract
The ability to accurately model the fitness landscape of protein sequences is critical to a wide range of applications, from quantifying the effects of human variants on disease likelihood, to predicting immune-escape mutations in viruses and designing novel biotherapeutic proteins. Deep generative models trained on large quantities of evolutionary sequences have demonstrated state-of-the-art performance on several of these tasks. However, engineering novel proteins to operate in non-natural conditions, or redesigning their function is currently best addressed via supervision on functional labels. This talk will introduce various hybrid sequence-only and sequence-label model architectures, and discuss their relative strengths across diverse functional protein design tasks.
Bio
Pascal Notin is a Scientific Lead in the Marks lab at Harvard Medical School, where he leads the subgroup focusing on protein engineering. His research lies at the intersection of Generative AI, Computational Biology, and Chemistry. He is passionate about the use of Machine Learning models to design novel biomolecules to address challenges in healthcare and sustainability.
Pascal completed his PhD in the Oxford Applied and Theoretical Machine Learning Group under the supervision of Yarin Gal. During his time there, he developed large-scale protein language models for fitness prediction and design, such as Tranception, TranceptEVE, RITA, and ProteinNPT. He also worked on deep generative models to predict the impact of genetic mutations in humans (EVE) and identify viral mutations likely to escape immunity (EVEscape). Alongside colleagues from the Marks lab, he has led the development of the ProteinGym benchmarks, which have been widely recognized as a useful resource for evaluating protein models.
Before returning to academia, Notin was a Senior Engagement Manager at McKinsey & Company, where he developed expertise in Digital & Analytics strategy and led cross-disciplinary teams on high-impact analytics projects, primarily in the healthcare and pharmaceutical sectors. He obtained an M.S. in Operations Research from Columbia University and a B.S. & M.S. in Applied Mathematics and Physics from Ecole Polytechnique.
Sponsor
Institute of Natural Sciences, Shanghai Jiao Tong University
Shanghai National Center for Applied Mathematics (SJTU Center)
