An ab initio information-theoretic approach to proteins and protein-ligand interactions
Differing conformational structure of a protein, associated with two distinct signaling states or between ligand-free and ligand-bound states, leads to differing inter-residue interactions and consequently different biological function. We propose a fresh first-principles information-theoretic approach for studying such proteins and their interactions. A de novo measure called protein residue information (PRI), which incorporates details of interactions between all pairs of atoms within and across all residues of the protein, is introduced herein. We formulate a method to calculate the intrastate and inter-state entropy of every residue, needed to determine PRI across any two states of a protein. The intra-state entropy can be determined for every state of a protein possessing one or more states. The inter-state entropy can be calculated pairwise for proteins possessing more than one state. We analyze twenty eight distinct pairs of protein structures from ten different classes. PRI successfully identifies important residues displaying significant conformational changes bearing influence with respect to itself and all other residues. Furthermore, it also successfully identifies important residues displaying rather subtle conformational changes. The identified residues exhibit influential roles in diverse performative features of proteins like stability, allostery, signaling, etc. PRI successfully recovers known experimental results from literature and predicts important roles for many hitherto unstudied residues.