This talk provids an overview of new free energy and enhanced sampling methods used for prediction of relative binding free energies in AMBER.  Methods include new pairwise smoothstep softcore potentials, Hamiltonian replica exchange molecular dynamics, and alchemical enhanced sampling.  Network-wide free energy analysis for libraries of compounds are achieved using a fast MBARnet and BARnet methods that are based on a constrained variational approach that can also include experimental or high-precision reference constraints, if known, to improve predictions.  Free energy values computing using traditional molecular mechanical (MM) force fields can be enhanced by using "book-ending" methods that rigorously correct the results to new fast, and accurate quantum mechanical force fields (QMFFs) with machine learning corrections achieved using new range-corrected deep learning potentials implementd into DeePMDkit and inerfaced with AMBER.

New free energy methods and machine learning force fields for drug discovery in AMBER

ACS Fall 2021

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