Philip Kim @ UToronto
[ Ссылка ]
Meet: [ Ссылка ]
Slides and Summary: [ Ссылка ]
Join group to receive calendar invite: [ Ссылка ]
Summary:
I will cover machine learning methods developed in my lab that cover protein sequence design, de novo protein design and antibody design, making use of graph neural networks, and diffusion models. We also have developed methods to model the conformational dynamics of peptides using PepFlow, a sequence-conditioned boltzmann-generator model. We show that PepFlow achieves state-of-the-art performance for peptide structure prediction and matches experimental conformational ensembles.
Bio:
Philip M. Kim is a professor at the University of Toronto at the Donnelly Centre and the Departments of Computer Science and Molecular Genetics. In his academic research, he has been developing novel machine learning methods for protein and peptide engineering and authored over 90 publications, 7 invention disclosures and 5 patent applications. He has co-founded several biotechnology companies and serves as consultant and member of the scientific advisory board for others. Before setting up his lab in 2009, he was a postdoctoral fellow at Yale University and an associate with McKinsey & Co. He holds a Ph.D. from the Artificial Intelligence Laboratory and Department of Chemistry at the Massachusetts Institute of Technology and a B.S. in Physics and Biochemistry from the University of Tuebingen.
