Wade, Rebecca

Rebecca Wade leads the Molecular and Cellular Modeling group at Heidelberg Institute for Theoretical Studies (HITS) and is Professor of Computational Structural Biology at the Center for Molecular Biology at Heidelberg University (ZMBH).  Rebecca Wade’s research is focused on the development and application of computer-aided methods to model and simulate biomolecular interactions. Her research group has developed novel protein structure-based methods for drug discovery and protein engineering, most recently for studying drug binding kinetics, as well as multiresolution computational approaches to investigate macromolecular association and the effects of macromolecular crowding.  URL: mcm.h-its.org.

Abstract:

Combining molecular simulation and machine learning approaches for structure-based drug design

Topic:

Structure-based drug design approaches increasingly require the handling of very large amounts of data, such as large compound libraries for screening or many protein conformations generated by molecular dynamics simulations. There is also the need to integrate diverse types of experimental and computational data into the design process. I will describe examples of how we are addressing these issues by combining molecular simulation and machine learning approaches [1-5]. We focus in particular on the challenges and opportunities for drug design provided by protein binding pocket dynamics. I will present the development and recent applications of a machine learning approach to identify pocket conformations with high druggability in TRAPP, a toolbox of computational approaches to identify TRAnsient Pockets in Proteins for ligand design (https://trapp.h-its.org/).  Protein binding site flexibility is one of the factors that can affect drug-target binding kinetics. Growing evidence that the efficacy of a drug can be correlated to target binding kinetics has led to the development of many new methods aimed at computing rate constants for receptor-ligand binding processes, see: kbbox.h-its.org. I will introduce the t-random acceleration molecular dynamics simulation (tRAMD) method to compute relative residence times and discuss how interaction fingerprint (MD-IFP) and machine learning analysis of tRAMD trajectories can be used to decipher the determinants of drug-target residence times.