Protein-protein binding affinity prediction from amino acid sequence

Motivation: Protein-protein interactions play crucial roles in many biological processes and are responsible for smooth functioning of the machinery in living organisms. Predicting the binding affinity of protein-protein complexes provides deep insights to understand the recognition mechanism and identify the strong binding partners in protein-protein interaction networks. Results: In this work, we have collected the experimental binding affinity data for a set of 135 protein-protein complexes and analyzed the relationship between binding affinity and 642 properties obtained from amino acid sequence. We noticed that the overall correlation is poor, and the factors influencing affinity depends on the type of the complex based on their function, molecular weight and binding site residues. Based on the results, we have developed a novel methodology for predicting the binding affinity of protein-protein complexes using sequence-based features by classifying the complexes with respect to their function and predicted percentage of binding site residues. We have developed regression models for the complexes belonging to different classes with three to five properties, which showed a correlation in the range of 0.739-0.992 using jack-knife test. We suggest that our approach adds a new aspect of biological significance in terms of classifying the protein-protein complexes for affinity prediction.