Integrated Methods for the Prediction of Binding Sites

The conditio sine qua non for structure-based drug design is the identification and functional annotation of the relevant binding site(s) in a target protein. A number of methods, closely related to the characteristics of binding sites and the restraints imposed on the formation of functional structural units, are discussed in Section 4.2. The most commonly used methods can be classified into geometry-based methods for cavity detection, methods for identifying specific patterns, and evolutionary methods. 

Recently, some more advanced methods have been reported that integrate the disparate features used for the characterization of functional binding sites. It can be expected that the cooperative effect of using all the information available will greatly enhance the reliability of binding site prediction and detection tools. 

A related approach integrating sequence information (conservation), geometric information (cleft detection), and data on local stability calculated by Poisson-Boltz-mann methods was reported by Ota et al. [54], The method was used for predicting catalytic residues (polar atoms only) in enzymes. A number of putative active sites for a series of hypothetical proteins were found and are discussed in the study. 

---