Prediction of water positions in the binding sites of proteins based on collections of multi-source heterogeneous atoms.

Abstract

Water molecules play an important role in mediating the interactions between proteins and ligands. However, it is difficult to distinguish the key water molecules directly because they are widely and irregularly distributed. Based on the results of statistical analysis, a composite tetrahedral model is proposed to predict the potential hydration sites in the binding sites of crystal structures. By analyzing the different protein atoms and ligand atoms that interact with water molecules,the unified representation and measurement of these multi-source heterogeneous atoms in the multi-dimensional feature space were adopted. The potential hydration sites could be predicted based on the results of the preference analysis and the shape-matching method. A test set was used to evaluate the model performance and extensive comparison with the tetrahedral-water-cluster model and Dowser++ revealed that the composite tetrahedral model can not only predict the potential sites of multiple key water molecules in the binding sites but also has a better prediction accuracy.