Paper detail

Contact and first layer residues prediction in protein dimers using the Gaussian Network model with adjustable number of fast modes

Interactions between proteins are hard to decipher. Protein-protein interactions are difficult problem to address because they are not based on differences in charge type like protein-DNA or protein-lipid interactions. In this manuscript we present a few methods aimed at recognizing contact and their neighboring first layer residues. The methods are based on the simplified analysis of normal modes via the Gaussian Network Model (GNM). The methods adjust the number of modes used in analysis on the basis of the comparison of predicted and expected number of targets (contact and first layer residues). The number of modes used in weighted sum calculation is increased or decreased if the number of predictions falls out of range for a given protein sequence length. The methods are able to recognize more than 50% of targets on average. At the same time they generate slightly more than 40% of false positives. They are also fairly successful in recognizing near native decoys of the Vakser decoy sets. The results depicted here show that kinetically active residues are important in protein-protein interactions. The GNM analysis depicted also shows that the two interacting chains usually exhibit opposite behavior. While the binding surface of one of the partners is rigid and stable, the interfacial area of the other partner is more flexible.