Identification of direct residue contacts in protein-protein interaction by message passing

Identification of direct residue contacts in protein-protein interaction by message passing

| Martin Weigt, Robert A. White, Hendrik Szurmant, James A. Hoch, Terence Hwa
The paper presents a novel computational method to identify direct residue contacts in protein-protein interactions by combining co-variance analysis with global inference using message passing. The method is applied to a set of over 2500 bacterial two-component signal transduction system proteins, specifically focusing on the interaction between sensor kinase (SK) and response regulator (RR) proteins, as well as homo-interactions between RR proteins. The co-variance analysis identifies correlated amino acid positions, while the global inference approach distinguishes between direct and indirect correlations. The method successfully identifies residue pairs that are proximal in space, both for hetero-interactions and homo-interactions, without the need for ad hoc tuning parameters. The results are validated using structural information from co-crystal structures, demonstrating that the method can accurately predict direct interactions based solely on sequence data. The authors expect this method to be applicable to a broader range of proteins as the number of sequenced genomes increases, potentially expanding the potential targets for therapeutic intervention and enhancing our understanding of protein-protein interaction mechanisms.The paper presents a novel computational method to identify direct residue contacts in protein-protein interactions by combining co-variance analysis with global inference using message passing. The method is applied to a set of over 2500 bacterial two-component signal transduction system proteins, specifically focusing on the interaction between sensor kinase (SK) and response regulator (RR) proteins, as well as homo-interactions between RR proteins. The co-variance analysis identifies correlated amino acid positions, while the global inference approach distinguishes between direct and indirect correlations. The method successfully identifies residue pairs that are proximal in space, both for hetero-interactions and homo-interactions, without the need for ad hoc tuning parameters. The results are validated using structural information from co-crystal structures, demonstrating that the method can accurately predict direct interactions based solely on sequence data. The authors expect this method to be applicable to a broader range of proteins as the number of sequenced genomes increases, potentially expanding the potential targets for therapeutic intervention and enhancing our understanding of protein-protein interaction mechanisms.
Reach us at info@study.space