The accurate prediction of protein-small molecule affinities is a central goal of computational chemistry. With growing computer power, relative binding free energy (RBFE) calculations based on atomistic molecular simulations1 have emerged as more accurate alternatives to less detailed methods, such as docking. The RBFE approach is particularly useful during the lead optimization phase of drug discovery, as it estimates the difference in affinity between two chemically similar compounds. It is not so well suited for virtual screening (VS) of a diverse compound catalog, simply because the compounds are so varied. In contrast, absolute binding free energy (ABFE) calculations via e.g., double decoupling (DD)2, attach-pull-release (APR)3, and simultaneous decoupling and recoupling (SDR)4, are applicable to VS, given adequate computer time. In particular, the SDR method, which simultaneously decouples the ligand from the binding site and recouples it with the solvent outside the protein, avoids simulations in vacuo and maintain the net charge of the simulated system, even for charged ligands. To enable the use of ABFE calculations in VS, we have created BAT.py (https://github.com/GHeinzelmann/BAT.py, v. 2)5, which automates DD and SDR ABFE calculations. Early studies using docking to generate poses and ABFE calculations to estimate the affinity for each pose showed promise5. Here we will present a more systematic study the use of docking plus ABFE calculations for virtual compound screening.

Absolute binding free energy calculations for virtual compound screening

ACS Spring 2022

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