•   When: Friday, May 25, 2018 from 01:00 PM to 02:00 PM
•   Speakers: Didier Devaurs
•   Location: ENGR 4801
•   Export to iCal

Abstract

Proteins are the main effectors of genomic information and are involved in a wide range of physiological and pathological processes. In this context, a protein’s function is known to be influenced or mediated by its interactions with other molecules, through the formation of molecular complexes. These interactions are often associated with changes in the protein's conformation (i.e., its three-dimensional structure). Studying this structure-function relationship requires gathering information about a protein's conformational space, i.e., the space of all possible states of the protein. While experimental techniques such as X-ray crystallography have enabled the description of numerous molecular structures, computational methods such as molecular dynamics are required to exhaustively explore the conformational space of proteins and molecular complexes. However, because of the curse of dimensionality, conformational space exploration remains a critical challenge for computational structural biology. In this talk, I will present three strategies (that can be combined) to mitigate the curse of dimensionality when computationally exploring the conformational space of a protein or a molecular complex. The first strategy consists of using coarse-grained conformational sampling methods, such as those derived from the field of robotics, instead of more accurate but more computationally-expensive all-atom simulation methods. The second strategy consists of guiding the conformational exploration with experimental data, such as the low-resolution data obtained through hydrogen-exchange monitoring. The third strategy consists of adopting a purely geometrical abstraction of the conformational exploration problem to enhance the scalability of existing computational methods, e.g., for the molecular docking of large ligands to proteins.

Short Bio

Didier Devaurs currently works in the field of biomedical computing with Dr. Lydia Kavraki. His research focuses on the development of new computational methods for the conformational modeling and analysis of peptides and proteins. Dr. Devaurs holds a B.S. in computer science and a B.S. in mathematics from the University of Clermont-Ferrand, France. He earned an M.S. in computer science from the University of Lyon, France, with major in knowledge and reasoning. He received his Ph.D. in computer science from the INPT, University of Toulouse, France, with majors in artificial intelligence and robotics.