Predicting Protein Metal Binding Sites with RBF Networks based on PSSM Profiles and Additional Properties

Background: Metal atoms are involved in many biological mechanisms, such as protein structure stability, apoptosis and aging. Therefore, identifying metal-binding sites in proteins is an important issue in helping biologists better understand the workings of these mechanisms. Methods: We propose a method based on Position Specific Scoring Matrix (PSSM) profiles and additional information (conservation score and solvent accessible surface area (ASA)) to identify the metal-binding residues in proteins. Results: We have selected a non-redundant set of 262 metal-binding proteins and 617 disulfide proteins as the independent test set. The proposed method can predict metal-binding sites at 51.0% recall and 73.4% precision. Comparing with the previous work of A. Passerini et al., the proposed method can improve over 7% of precision with the same level of recall on the independent dataset. Conclusions: We have developed a novel approach based on PSSM profiles and additional properties for identifying metal-binding sites from proteins. The proposed approach achieved a significant improvement with newly discovered metal-binding proteins and disulfide proteins.