Using gene ontology to enhance effectiveness of similarity measures for microarray data

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Feature selection methods aim to reduce data redundancy, often relying on gene expression correlations.
• Existing methods may not fully capture biological relationships, potentially missing optimal feature subsets.

Purpose of the Study:

• To develop a novel gene similarity measure incorporating Gene Ontology (GO) semantics and expression levels.
• To reduce redundancy in gene expression data more effectively than traditional methods.

Main Methods:

• Proposed a new similarity measure combining semantic similarity from Gene Ontology with expression level similarity.
• Evaluated the method on public microarray datasets using classification tasks.

Main Results:

• The proposed method achieved higher or equivalent classification accuracy compared to expression value-only models.
• The approach successfully identified smaller, yet effective, gene feature sets, indicating reduced redundancy.

Conclusions:

• Integrating Gene Ontology semantics provides a powerful way to measure gene redundancy.
• This novel approach enhances feature selection by improving classification performance and reducing feature dimensionality.