8th January 2020, 14h00 - 15h00
(Conference room 13 rue Moreau)

Martin Weigt (UMR 7238 - Computational and Quantitative Biology, Sorbonne Université)

Invited by: Deniz Dalkara, Ulisse Ferrari

Proteins belong to the most fascinating complex systems in nature. They are simultaneously robust and fragile: Proteins may conserve their 3D structure and function throughout evolution, in many cases even from bacteria to humans, while substituting more than 80% of their amino acids; however, very few random mutations may destabilize a protein or interrupt its function. Thanks to the sequencing revolution in biology, data-driven approaches modeling this sequence variability based on inspiration from statistical physics and inference are becoming increasingly popular. In my talk, I will review their astonishing capacity to extract information about 3D protein structure, about protein-protein interactions or about the effects of mutations from publicly available sequence data. In particular this last example leads to the concept of protein sequence landscapes (closely related to fitness landscapes), which have promising applications from interpreting evolutionary data to de-novo protein design.