Event

Abstract: Recent advances in computational protein design promise to accelerate the development of therapeutic molecules, however methods that are published in scientific literature do not always translate smoothly to applications at large pharmaceutical companies. This talk features anecdotal examples from the Protein Design and Informatics group at GSK that demonstrate both the impact of predictive modeling on current medicines and its limitations. I'll cover problems ranging from “Can we predict which human proteins will recombinantly express in E. coli?” to “Can we design efficient libraries for enzyme and antibody evolution?”, which we address with methods that span the gamut from first principles physics to deep learning protein language models.

Google Scholar
Twitter/X: @lqtza