From workstations to workbenches: Towards predicting physicochemically viable protein–protein interactions across a host and a pathogen

Abstract

The understanding of protein–protein interactions is indispensable in comprehending most of the biological processes in a cell. Small-scale experiments as well as large-scale high-throughput techniques over the past few decades have facilitated identification and analysis of protein–protein interactions which form the basis of much of our knowledge on functional and regulatory aspects of proteins. However, such rich catalog of interaction data should be used with caution when establishing protein–protein interactions in silico, as the high-throughput datasets are prone to false positives. Numerous computational means developed to pursue genome-wide studies on protein–protein interactions at times overlook the mechanistic and molecular details, thus questioning the reliability of predicted protein–protein interactions. We review the development, advantages, and shortcomings of varied approaches and demonstrate that by providing a structural viewpoint in terms of shape complementarity and interaction energies at protein–protein interfaces coupled with information on expression and localization of proteins homologous to an interacting pair, it is possible to assess the credibility of predicted interactions in biological context. With a focus on human pathogen Mycobacterium tuberculosis H37Rv, we show that such scrupulous use of details at the molecular level can predict physicochemically viable protein–protein interactions across host and pathogen. Such predicted interactions have the potential to provide molecular basis of probable mechanisms of pathogenesis and hence open up ways to explore their usefulness as targets in the light of drug discovery.