Large-scale, automated prediction of protein-ligand binding structures

Molecular docking is a useful and important approach for the prediction of protein-ligand structures and for structure-based drug design. The growing number of protein-ligand complex structures, particularly the structures of proteins co-bound with different ligands, in the Protein Data Bank helps us tackle two major challenges in molecular docking studies: the protein flexibility and the energy scoring function. Here, we introduced a systematic strategy by using the information embedded in the known protein-ligand complex structures to improve binding mode predictions. We also developed and integrated several methods for large-scale binding mode prediction, and systematically tested these methods on the weekly Continuous Evaluation of Ligand Pose Prediction (CELPP) competition, an automated workflow to process and evaluate the challenge of ligand pose prediction. Up to October 26, 2018, the quantitative analysis of our docking results for over 3000 targets released by CELPP revealed that our methods improve the success rates of ligand pose prediction.