Moment invariants as shape recognition technique for comparing protein binding sites

Area of Science:

• Biochemistry
• Structural Biology
• Bioinformatics

Background:

• Protein-protein interactions are fundamental to biological processes.
• Identifying similar binding sites is crucial for understanding protein function and drug discovery.
• Existing methods may lack the precision for detailed geometric comparisons.

Purpose of the Study:

• To develop and validate a computational approach for identifying similarities in protein-protein binding site geometry.
• To quantify the effectiveness of the proposed similarity measure.

Main Methods:

• Describing binding site geometry using a feature vector derived from moment invariants.
• Comparing feature vectors to identify similar binding sites.
• Validating the approach on a dataset of 2819 protein-protein binding sites from the SCOPPI database.

Main Results:

• Achieved a median area under the ROC curve of 0.98, indicating high accuracy.
• Identified similar binding sites within the top two of 2819 entries for half of the queries.
• Demonstrated the method's utility through hierarchical clustering of the nitrogenase iron protein-like SCOP family.

Conclusions:

• The moment invariant-based feature vector approach effectively identifies geometrically similar protein-protein binding sites.
• The method offers high accuracy and efficiency in searching large binding site datasets.
• This technique has potential applications in structural biology and drug design.