The development and validation of a recently developed workflow to setup and run physics based protein-ligand binding affinity calculations will be described. The workflow is applicable to small molecule lead optimization where efficient calculation of binding affinity for congeneric series of lead-like compounds is desired. The first step in the workflow implements a fast end-point method that uses the co-crystal structure of the protein-lead complex as template to reliably generate initial interaction geometries of a congeneric series and approximately score them. The method is used to shortlist promising compounds and generates geometries for initiating FEP calculations. The workflow subsequently plans a minimal set of relative FEP calculations that spans the set of compounds, runs the calculations on grid computing infrastructure, and aggregates the simulation results to compute binding affinities. The workflow was retrospectively validated using 6 congeneric ligand sets for diverse target proteins using transferred force-field parameters. Various aspects of the results will be described including identifying unreliability, directional accuracy, and the effect of altering sampling time. The retrospective validation results suggest promising utility of the automated workflow with minimal manual input for efficient prioritization of compounds for synthesis in lead optimization. Finally, ongoing efforts toward improving the accuracy and scalability of free energy calculations will be discussed.<br/>

Development and validation of a relative Free Energy Perturbation (FEP) workflow

2019 Fall National Meeting

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