Alchemical free energy methods provide a rigorous framework for the calculation of protein-ligand binding affinities. These methods have issues surrounding poor convergence, speed and scalability. Recent trends have ameliorated the issues of convergence and speed, but scalability still remains. We have addressed the issues of speed and scalability, in our newly developed GPU version of multisite λ-dynamics (MSλD) using the CHARMM molecular dynamics package. Using the GPU version of MSλD, we are able to compute the binding affinities of 512 inhibitors of HIV-RT in a few days, with accuracy. Our GPU-based method is capable of accelerating the calculation of protein-ligand binding affinities by a significant amount, where accurate calculations of tens to hundreds of thousands of protein-ligand complexes can be computed within a short period of time. This is the largest computational exploration for the calculation of binding affinities known, and has the potential for involvement in large-scale applications for drug discovery campaigns, specifically with co-optimization of the ligand binding affinity against multiple drug targets.

Predicting binding-free energies in a combinatorial chemical space using a GPU implementation of multisite lambda dynamics

251st National Meeting (2016)

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