Combining physicochemical and evolutionary information for protein contact prediction.

Area of Science:

• Computational Biology
• Structural Bioinformatics
• Protein Science

Background:

• Accurate prediction of native protein contacts is crucial for determining protein structure.
• Existing methods often struggle with limited evolutionary information.

Purpose of the Study:

• To develop a novel contact prediction method combining evolutionary and physicochemical information.
• To improve the accuracy of *ab initio* protein structure prediction.

Main Methods:

• Utilizing evolutionary information from multiple-sequence alignments.
• Extracting physicochemical information from predicted *ab initio* protein structures.
• Employing a graph-based representation of contact structural context.
• Training a support vector machine classifier to identify native contacts.

Main Results:

• Achieved high prediction accuracy for native contacts.
• Maintained prediction accuracy even with sparse sequence homologs.
• Demonstrated improved *ab initio* protein structure prediction using predicted contacts.

Conclusions:

• The combined evolutionary and physicochemical approach offers a robust method for contact prediction.
• This method enhances the capabilities of *ab initio* protein structure prediction.
• A web service is available for utilizing the EPC-MAP tool.