Identification of coevolving residues and coevolution potentials emphasizing structure, bond formation and catalytic...

Area of Science:

• Computational Biology
• Protein Science
• Bioinformatics

Background:

• Protein structure and function rely on intricate residue interactions.
• Previous methods like mutual information detect coevolution but have limitations.
• Understanding coevolutionary pressures is key to predicting mutation effects.

Purpose of the Study:

• To develop a refined mutual information method for detecting protein coevolution.
• To introduce 'coevolution potentials' as a novel measure for amino acid pairings.
• To validate the algorithm's accuracy by correlating predictions with known protein features.

Main Methods:

• Applied a refined mutual information algorithm to a large protein alignment database.
• Developed 'coevolution potentials' to quantify amino acid pair propensities.
• Correlated predicted coevolving pairs with physical proximity and functional roles (e.g., catalytic residues).

Main Results:

• The refined method effectively removes bias and accounts for heteroscedasticity in coevolutionary signals.
• Predicted coevolving residue pairs are significantly more likely to be in close physical proximity.
• Ionic, hydrogen, and disulfide bond-forming pairs show high coevolution potentials; catalytic residues are frequently identified as coevolving.

Conclusions:

• The improved algorithm accurately identifies functionally relevant coevolving residue pairs.
• Coevolutionary pressures shape protein mutational landscapes by favoring interactions.
• This work provides a more reliable tool for studying protein evolution and function.