Localization of binding sites in protein structures by optimization of a composite scoring function

Area of Science:

• Structural Biology
• Computational Biology
• Bioinformatics

Background:

• The increasing number of uncharacterized protein structures necessitates efficient functional annotation methods.
• Structure-based approaches are crucial for understanding protein function.

Purpose of the Study:

• To develop and validate an automated method for predicting the location of specific binding sites on protein structures.
• To enhance the process of functional annotation for proteins with known structures.

Main Methods:

• A scoring function was developed using z-scores of protein properties (conservation, compactness, protrusion, etc.).
• Monte Carlo optimization was employed to identify optimal patches on the protein surface.
• Weights for the scoring function were derived from known binding site instances.

Main Results:

• The method correctly identified binding sites for nonsugar ligands (e.g., nucleotides) in 55%-73% of tested cases.
• The approach demonstrated high accuracy and applicability by incorporating diverse information types.
• The method is fully automated and scalable for large-scale structural genomics projects.

Conclusions:

• The developed method provides an effective and automated solution for predicting protein binding site locations.
• This tool can significantly aid in the functional annotation of proteins within structural genomics initiatives.
• The approach's flexibility allows for adaptation to various binding site types and information sources.