Abstract
Accurate binding free energy prediction is vital for drug design, motivating the assessment of new force field models. Here, we evaluate the ABCG2 charge model with nonequilibrium alchemical free-energy simulations. GAFF2/ABCG2 achieves higher hydration free energy accuracy but does not outperform GAFF2/AM1-BCC for protein-ligand binding free energy. Both charge models exhibit comparable accuracy and compound ranking across targets, indicating that property-specific force field optimization does not guarantee improved related-property performance.