FINDSITE: a combined evolution/structure-based approach to protein function prediction.

Area of Science:

• Bioinformatics
• Computational Biology
• Structural Biology

Background:

• The post-genomic era faces the challenge of assigning functions to numerous uncharacterized molecules.
• Approximately 50% of open reading frames (ORFs) in proteomes lack assigned functions, hindering biological understanding.
• Sequence and structure-based methods are vital for inferring protein functions and screening ligands.

Purpose of the Study:

• To review current sequence and structure-based approaches for protein function inference and ligand screening.
• To introduce FINDSITE, a novel algorithm for predicting ligand binding sites and molecular functions.
• To evaluate FINDSITE's performance using both experimental structures and predicted protein models.

Main Methods:

• Review of existing sequence and structure-based computational methods.
• Description of the FINDSITE algorithm, utilizing binding site conservation across evolutionarily distant proteins identified by threading.
• Application of FINDSITE to predict ligand binding sites, screen for ligands, and infer molecular functions.

Main Results:

• FINDSITE demonstrates effectiveness in predicting ligand binding sites and molecular functions.
• The algorithm achieves comparable results whether using high-resolution experimental protein structures or predicted protein models.
• FINDSITE provides functional insights for a substantial portion of proteins with unassigned functions.

Conclusions:

• FINDSITE is a valuable tool for accelerating the functional annotation of proteins in the post-genomic era.
• The algorithm's ability to work with both experimental and predicted structures enhances its applicability.
• FINDSITE contributes to addressing the challenge of unassigned ORFs by providing functional predictions.