MM-GB/SA rescoring of docking poses in structure-based lead optimization.

The critical issues in docking include the prediction of the correct binding pose and the accurate estimation of the corresponding binding affinity. Different docking methodologies have all been successful in reproducing the crystallographic binding modes but struggle when predicting the corresponding binding affinities. The aim of this work is to evaluate the performance of the MM-GB/SA rescoring of docking poses in structure-based lead optimization. To accomplish that, a diverse set of pharmaceutically relevant targets, including CDK2, FactorXa, Thrombin, and HIV-RT were selected. The correlation between the MM-GB/SA results and experimental data in all cases is remarkable. It even qualifies this approach as a more attractive alternative for rank-ordering than the Free Energy Perturbation and Thermodynamic Integration methodologies because, while as accurate, it can handle more structurally dissimilar ligands and provides results at a fraction of the computational cost. On the technical side, the benefit of performing a conformational analysis and having an ensemble of conformers to represent each ligand in the unbound state during the MM-GB/SA rescoring procedure was investigated. In addition, the estimation of conformational entropy penalties for the ligands upon binding, computed from the Boltzmann distribution in water, was evaluated and compared to a commonly used approach employed by many docking scoring functions.