NoGOA: predicting noisy GO annotations using evidences and sparse representation.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Gene Ontology (GO) annotations link gene products to functional terms.
• Electronically inferred annotations often contain noise, impacting research.
• Accurate identification of noisy GO annotations is an open challenge.

Purpose of the Study:

• To develop a novel method for predicting noisy Gene Ontology annotations.
• To improve the reliability of gene function prediction by addressing annotation noise.

Main Methods:

• Utilized sparse representation on the gene-term association matrix.
• Incorporated semantic similarity measures between genes.
• Developed a voting system based on semantic neighborhoods to predict noisy annotations.
• Integrated evidence code ratios and GO hierarchy weighting for robust prediction.

Main Results:

• The NoGOA approach significantly outperforms existing methods in predicting noisy annotations.
• Experiments were conducted on six model species, demonstrating broad applicability.
• Removing identified noisy annotations enhanced gene function prediction performance.

Conclusions:

• The study validates the effectiveness of combining evidence codes with sparse representation for noisy GO annotation prediction.
• The developed NoGOA method offers a valuable tool for improving GO annotation quality.
• Associated code and datasets are publicly available for community use.