Interspecies gene function prediction using semantic similarity.

Area of Science:

• Bioinformatics
• Genomics
• Computational Biology

Background:

• Gene Ontology (GO) provides a controlled vocabulary for gene product functions and locations.
• Current gene annotations vary across species due to research limitations and ethics, leading to incomplete data.
• Homologous genes often have different GO terms, potentially due to incomplete annotations rather than functional divergence.

Purpose of the Study:

• To investigate the feasibility of improving gene annotations using semantic similarity between homologous genes from different species.
• To assess the contribution of interspecies gene semantic similarity to replenishing incomplete gene annotations.

Main Methods:

• Utilized three semantic similarity metrics to compute similarity between genes from two species.
• Employed a k-nearest neighbor approach to identify homologous genes.
• Applied terms from k neighbors to replenish annotations of target genes.
• Conducted experiments on archived GO annotations for Human, Mouse, Zebrafish, and Arabidopsis thaliana.

Main Results:

• Interspecies semantic similarity significantly improved annotation accuracy by 53.22% compared to single-species or low-homology approaches.
• GO annotations from homologous species are complementary, enhancing overall annotation completeness.
• Semantic similarity-based interspecies gene function annotation is more effective than traditional intraspecies methods.

Conclusions:

• Semantic similarity-based interspecies gene function annotation offers a powerful approach for improving biological data.
• This study highlights the potential of cross-species analysis for advancing gene function prediction.
• The findings encourage further research into interspecies semantic similarity for functional genomics.