Molecular docking with conformer-dependent charges.

Abstract

Determination of protein-ligand interactions is crucial for structure-based drug design. But, accurate prediction of the binding structures of protein-ligand is still a major challenge for molecular docking methods. Herein, we developed molecular docking with conformer-dependent charges (MDCC), a docking method to combine conformational search with RESP charges. Compared to the conventional Glide SP method (51.9%) and Glide XP method (52.6%), the MDCC method (60%) exhibited a higher docking success rate based on 285 protein-small molecule ligand complexes from the PDBbind core set. And when the ligand met one of the conditions (the total hydrophobic surface area > 438, the number of hydrophobic atoms > 10, and molecular weight > 235), the docking success rate of the MDCC method (>90%) was higher than that of the Glide SP method (∼72%). Furthermore, we also applied the MDCC method to predict the protein-ligand interactions in GPCR Dock 2021 competition, with our prediction models ranking 2nd out of all 202 participating models for APJ and 5th out of all 193 participating models for GPR139, demonstrating the relatively high docking accuracy of the MDCC method. In addition, we also found that the MDCC method combined with molecular dynamics simulations could facilitate the application of AlphaFold 2 in drug discovery. The above results all set the stage for the application of the MDCC method in future practical drug design.