XDock: A General Docking Method for Modeling Protein–Ligand\nand Nucleic Acid–Ligand Interactions

Molecular docking is an essential\ncomputational tool in structure-based\ndrug discovery and the investigation of the molecular mechanisms underlying\nbiological processes. Despite the development of many molecular docking\nprograms for various systems, a universal tool that can accurately\ndock ligands across multiple system types remains elusive. Meeting\nthe need, we developed XDock, a versatile docking framework built\nfor both protein–ligand and nucleic acid–ligand interactions.\nXDock efficiently accounts for ligand flexibility by docking multiple\nconformations of a ligand and flexibly refining the final binding\nposes. It utilizes a distance geometric method for ligand sampling\nand leverages our knowledge-based scoring functions for assessing\nprotein–ligand and nucleic acid–ligand interactions.\nXDock has undergone extensive validations on diverse benchmarks of\nprotein–ligand and nucleic acid–ligand complexes and\nwas compared with six other docking methods, including DOCK 6, AutoDock\nVina, PLANTS, LeDock, rDock, and RLDock. In addition, XDock is also\ncomputationally efficient and on average can dock a ligand within\n1 min.