Predicting the accuracy of protein-ligand docking on homology models

Ligand-protein docking is a computational approach that is increasingly used in Drug Discovery. The initial limitations imposed by a reduced availability of experimental structures of target proteins have been overcome by theoretical modeling with bioinformatics methods as homology modelling.1 While this greatly extended the use of docking simulations, it also introduced the need for general criteria to estimate the reliability of docking results given the model quality.2 To this end, we performed a large-scale experiment on a diverse set including experimental structures and homology models for a group of representative ligandprotein complexes.3 A wide spectrum of model quality was sampled using templates at different evolutionary distances and different strategies for target-template alignment and modelling. The obtained models were evaluated with a selection of model quality indices.4 The binding geometries were generated using AutoDock,5 a widely used docking program. The results demonstrated that general relationships do exist between the accuracy of the binding geometries obtained by docking and the quality of the homology models. In particular, it was highlighted that successful docking requires an accurate modelling of the binding site and that, conversely, the development of docking approaches that allow flexibility in the binding site may have a great impact on the effective use of docking into homology models. More importantly, state-of-the-art indices for model quality assessment are an effective tool for a priori prediction of the accuracy of docking experiments in the context of groups of proteins with conserved structural characteristics. On these bases, we presented a strategy to exploit information on homologous proteins to predict the accuracy of docking results on theoretical models.