Enhancing Relative Binding Free Energy Calculation with Grand Canonical Monte Carlo, Water-swap Monte Carlo,...

Alchemical free energy calculations, particularly free energy perturbation (FEP), have become routine in drug design. When molecular dynamics (MD) is employed as the sampling method in FEP, accurately sampling water molecules deeply buried in protein binding pockets remains challenging, despite their critical role in protein-ligand interactions. We implemented three enhanced sampling methods, Grand Canonical Monte Carlo (GCMC), water-swap Monte Carlo (water-swap MC), and replica exchange with solute tempering (REST2). GCMC or water-swap MC accelerates the hydration state equilibration in buried pockets and REST2 accelerates the sampling of ligand conformations. Combining REST2 with GCMC or water-swap MC enables accurate calculation of ligand relative binding free energies. In the benchmark on the water set, our package achieved an RMSE of 0.92 kcal/mol, compared to 0.86 kcal/mol from FEP+ and 1.60 kcal/mol from OpenFE. This development provides the scientific community with an open-source sampling engine for robust and accurate FEP calculations.