Large Scale Prediction of Computational Hot Spots in Protein Interfaces

Abstract

Hot spots are residues comprising only a small fraction of interfaces yet accounting for the majority of the binding energy. These residues are critical in understanding the principles of protein interactions. Experimental studies like alanine scanning mutagenesis require significant effort; therefore, there is a need for computational methods to predict hot spots in protein interfaces. We present a new efficient method to determine computational hot spots based on sequence conservation and solvent accessibility of the interface residues (Tuncbag et al.; Guney et al.). The predicted hot spots are observed to correlate with the experimental hot spots with an accuracy of 71% and a positive predictive value of 79%. Several machine learning methods (SVM, Decision Trees and Decision Lists) are also applied to predict hot spots and compared to our method. The results reveal that our empirical approach performs better. We observe that both the change in accessible surface area upon complexation and residue accessibility in the complex forms improve detection of hot spots. Predicted computational hot spots for all protein interfaces (49512 interfaces as of 2006) are available at HotSprint database. HotSprint (a database of computational hot spots in protein interfaces) can be accessed at http://prism.ccbb.ku.edu.tr/hotsprint.Guney E, Tuncbag N, Keskin O, Gursoy A: HotSprint: database of computational hot spots in protein interfaces Nucleic Acids Res 2008, 36(Database issue):D662-666.Tuncbag N, Gursoy A, Guney E, Nussinov R, Keskin O: Architectures and Functional Coverage of Protein-Protein Interfaces J Mol Biol 2008.