The association/dissociation of biomolecular complexes are processes so slow that they can not be studied using straightforward all-atom simulation techniques. Thankfully, over the decades several enhanced sampling algorithms have been proposed that can perform sampling of such rare events in an accelerated but controllable manner. However, a large class of these methods (arguably all?) need an a priori, at least approximate, sense of a low-dimensional reaction coordinate (RC) even before performing the sampling. In order to deal with this cyclic problem where one needs extensive sampling of the rare events to know the RC, but also needs to know the RC in the first place to perform sampling, it is thus extremely desirable to construct methods that learn the RC as they perform the sampling. In this talk we will be describing two such recent methods, namely SGOOP and RAVE that use flavors of statistical mechanics and deep learning to solve this problem of simultaneously identifying the reaction coordinate and obtaining converged free energy surfaces and kinetics. SGOOP is based on a Maximum Caliber framework, while RAVE uses the variational autoencoder framework of deep learning. We will possibly discuss connections between the two methods, and also illustrate their applications to challenging problems such as protein-ligand unbinding in Src kinases and others.

How, when and why do small molecules unbind: Insights from predictive all-atom simulations

2019 Fall National Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.