Term Matrix: a novel Gene Ontology annotation quality control system based on ontology term co-annotation patterns

Area of Science:

• Bioinformatics
• Molecular Biology
• Genomics

Background:

• Gene products often participate in multiple biological processes, leading to multiple Gene Ontology (GO) term annotations.
• Co-annotation of gene products to functionally, temporally, or spatially distant processes may indicate annotation errors.
• Existing annotation pipelines and literature curation can introduce inaccuracies in GO term assignments.

Purpose of the Study:

• To develop and validate a novel annotation quality control workflow for detecting errors in Gene Ontology (GO) annotations.
• To identify and correct erroneous co-annotations of gene products to mutually exclusive biological processes.
• To enhance the reliability of GO annotations for biological research.

Main Methods:

• Developed a quality control workflow utilizing rules based on mutually exclusive biological processes.
• Validated the workflow through case studies involving fission yeast, human cohesin complex subunits, and GO biological process terms in multiple species.
• Reviewed GO annotations, identified unlikely co-annotations, and traced them to their sources.

Main Results:

• Generated 107 quality control rules to detect annotation errors.
• Corrected 289 manual GO annotations in eukaryotes.
• Identified and corrected over 52,700 automatically propagated annotations across all taxa.

Conclusions:

• The developed workflow effectively identifies and corrects errors in Gene Ontology annotations.
• This quality control method improves the accuracy of biological process assignments for gene products.
• The findings contribute to more reliable genomic data for biological research.