Director’s guest: Raphaël Guérois, I2BC Paris-Saclay – “Decoding Protein Interaction Networks with 3D and AI-Driven Evolutionary Insights”

Abstract

The rapid advancements in artificial intelligence have, within a few years, enabled the prediction of the three-dimensional structure of most proteins at atomic resolution. These predictions critically depend on the evolutionary information available in the sequences of homologous proteins. Beyond modeling the structures of individual proteins, there is now a fundamental challenge to establish the most comprehensive mapping possible of the structures of multiprotein assemblies, which are central to cellular function. This goal faces several constraints, such as the combinatorial complexity of the interaction space and the type of evolutionary information needed to identify coevolution signals, which are much weaker at complex interfaces than in the structures of individual proteins. To achieve this goal, we are developing computational methods with the aim of targeting interactions mediated by disordered regions and pushing further the current boundaries to predict the structure of assemblies at large scale.