Mutual information in protein multiple sequence alignments reveals two classes of coevolving positions

Area of Science:

• Bioinformatics
• Computational Biology
• Structural Biology

Background:

• Protein sequence alignments contain conserved and nonconserved positions.
• Coevolution analysis can reveal functional and structural relationships between amino acid positions.
• Understanding coevolutionary patterns is key to deciphering protein function.

Purpose of the Study:

• To apply information theory to identify nonconserved, coevolving positions in protein families.
• To categorize coevolving positions based on their interaction patterns.
• To assess the functional significance of coevolving positions.

Main Methods:

• Utilized information theory to analyze multiple sequence alignments across diverse protein families.
• Identified and categorized coevolving positions based on the number of interacting partners.
• Correlated coevolutionary patterns with known functional regions (e.g., active sites).

Main Results:

• Discovered two main categories of coevolving positions: those interacting with few partners (often direct side-chain interactions) and those interacting with many partners (often in functional regions).
• Found that coevolving positions are more sensitive to functional changes upon mutation compared to less coevolving positions.
• Demonstrated that information theory can effectively identify functional sites in uncharacterized proteins via coevolution analysis.

Conclusions:

• Nonconserved coevolving positions are critical indicators of functional sites within protein families.
• These coevolving positions may be as vital to protein structure and function as highly conserved positions.
• Information theory provides a powerful tool for discovering functional insights from protein sequence data.