Consensus of Sample-Balanced Classifiers for Identifying Ligand-Binding Residue by Co-evolutionary Physicochemical Characteristics of Amino Acids

Abstract

Protein-ligand binding is an important mechanism for some proteins to perform their functions, and those binding sites are the residues of proteins that physically bind to ligands. So far, the state-of-the-art methods search for similar, known structures of the query and predict the binding sites based on the solved structures. However, such structural information is not commonly available. In this paper, we propose a sequence-based approach to identify protein-ligand binding residues. Due to the highly imbalanced samples between the ligand-binding sites and non ligand-binding sites, we constructed several balanced data sets, for each of which a random forest (RF)-based classifier was trained. The ensemble of these RF classifiers formed a sequence-based protein-ligand binding site predictor. Experimental results on CASP9 targets demonstrated that our method compared favorably with the state-of-the-art.