CvManGO, a method for leveraging computational predictions to improve literature-based Gene Ontology annotations

Area of Science:

• * Genomics
• * Bioinformatics
• * Molecular Biology

Background:

• * The Saccharomyces Genome Database (SGD) utilizes Gene Ontology (GO) terms to classify S. cerevisiae gene product functions.
• * GO annotations are crucial for functional comparisons across organisms and predicting functions of related genes.
• * Accurate and high-quality GO annotations are a priority for SGD, sourced from manual curation of literature or automated computational predictions.

Purpose of the Study:

• * To discuss the relationship between literature-based and computationally predicted GO annotations at SGD.
• * To introduce and validate a strategy for identifying gene annotations that require review by comparing manual and computational data.
• * To explore factors influencing the effectiveness of this strategy in finding under-annotated genes.

Main Methods:

• * Developed CvManGO (Computational versus Manual GO annotations) method.
• * Paired literature-based GO annotations with computational GO predictions.
• * Evaluated the relationship between annotation types within the GO hierarchy to detect discrepancies.

Main Results:

• * The CvManGO method successfully identifies genes requiring annotation updates, aiding in prioritizing literature review.
• * Explored factors influencing the method's effectiveness but found no immediate criteria to enrich for under-annotated genes.
• * Demonstrated the utility of comparing manual and computational GO annotations for quality control.

Conclusions:

• * The CvManGO strategy is effective in highlighting discrepancies between manual and computational GO annotations, signaling a need for review.
• * Further improvements to the strategy and its applicability to other GO curation projects are discussed.
• * This approach contributes to maintaining the accuracy and quality of functional annotations in biological databases.