Prediction of protein binding sites and hot spots

Protein–protein interactions are involved in the majority of cell processes, and their detailed structural and functional characterization has become one of the most important challenges in current structural biology. The first ideal goal is to determine the structure of the specific complex formed upon interaction of two or more given proteins. However, since this is not always technically possible, the practical approach is often to locate and characterize the protein residues that are involved in the interaction. This can be achieved by experimental means at expense of time and cost, so a growing number of computer tools are becoming available to complement experimental efforts. Reported methods for interface prediction are based on sequence information or on structural data, and make use of a variety of evolutionary, geometrical, and physicochemical parameters. As we show here, computer predictions can achieve a high degree of success, and they are of practical use to guide mutational experiments as well as to explain functional and mechanistic aspects of the interaction. Interestingly, it has been found that typically only a few of the interacting residues contribute significantly to the binding energy. The identification of such hot‐spot residues is important for understanding basic aspects of protein association. In addition, these residues have received recent attention as possible targets for drug design, so several computer methods have been developed to predict them. We will review here existing computer approaches for the prediction of protein binding sites and hot‐spot residues, with a discussion on their applicability and limitations. © 2011 John Wiley & Sons, Ltd. WIREs Comput Mol Sci 2011 1 680–698 DOI: 10.1002/wcms.45 This article is categorized under: Structure and Mechanism > Computational Biochemistry and Biophysics