Kennedy Lectures with Mark Tuckerman NYU at 11:00am
Machine learning has become an integral tool in the theoretical and computational molecular sciences. Uses of machine learning in this area include prediction of molecule and materials properties from large databases of descriptors, design of new molecules with desired characteristics, design of chemical reactions and processes, representation of high-dimensional potential energy and free energy surfaces, creation of new enhanced sampling strategies, and bypassing of costly quantum chemical calculations, to highlight just a few. This lecture will focus on the use of machine learning and rare-event sampling strategies for finding reaction coordinates that characterize transitions between different basins on high-dimensional free energy surfaces and generating pathways between these basins. We will first review collective-variable based enhanced sampling techniques as implemented in a branch of the OpenMM package, which we are calling. We will then show how machine learning techniques can be leveraged to regress high-dimensional free energy surfaces via an active learning approach for the computation of observables and to perform dimensional reduction to learn reaction coordinates using SHAP analysis. Specific examples from biomolecular conformational sampling and transition mechanisms will be presented.
Mark E. Tuckerman 1,2,3,4,5
1Department of Chemistry, New York University, New York, NY 10003 USA
2Department of Physics, New York University, New York, NY 10003 USA
3Courant Institute of Mathematical Sciences, New York University, New York, NY 10012 USA
4NYU-ECNU Center for Computational Chemistry at NYU Shanghai, 3663 Zhongshan Rd. N. Shanghai 200062
5Simons Center for Computational Physical Chemistry at New York University, New York, NY 10003