Using computational predictions to improve literature-based Gene Ontology annotations: a feasibility study.

Area of Science:

• Bioinformatics
• Genomics
• Computational Biology

Background:

• Gene Ontology (GO) terms provide a standardized way to describe gene product functions across species.
• Accurate GO annotations are crucial for propagating functional information between related proteins.
• Literature-based annotations are vital but require rigorous validation.

Purpose of the Study:

• To develop and evaluate a strategy for identifying and prioritizing gene annotations that require review.
• To assess the utility of comparing literature-based GO annotations with computational predictions.
• To improve the quality and completeness of GO annotations in biological databases.

Main Methods:

• A comparative strategy was employed, contrasting literature-based GO annotations with computational predictions (e.g., InterPro-based).
• Manual 'unknown' annotations in the Saccharomyces Genome Database (SGD) were compared with computational predictions.
• A survey of GO annotations and predictions from the Gene Ontology Annotation (GOA) project (EBI) across multiple databases was conducted.

Main Results:

• The comparison method successfully identified gene annotations in SGD that needed updating.
• The strategy is applicable for maintaining and enhancing GO annotations in various organisms beyond yeast.
• Many genomes have genes lacking computational predictions but possessing manual annotations, highlighting the need for curation.

Conclusions:

• Combining manual, literature-based curation with computational methods is essential for comprehensive and accurate genome annotation.
• The developed strategy offers a robust approach to quality control for GO annotations.
• Continuous comparison and validation are necessary to ensure the reliability of functional genomics data.