An unsupervised approach to predict functional relations between genes based on expression data.

Area of Science:

• Computational Biology
• Systems Biology
• Bioinformatics

Background:

• Genes exhibit various functional relationships, including regulatory interactions, participation in protein complexes, and involvement in metabolic or signaling pathways.
• Gene expression data contains implicit information about these functional relationships.

Purpose of the Study:

• To develop and validate a novel computational approach for predicting functional gene relations using gene expression data.
• To identify gene pairs with significant functional associations and construct gene networks.

Main Methods:

• Digitization of log-ratio gene expression data for S. cerevisiae into a matrix of 1, 0, and -1.
• Construction of probability density mass function tables for gene pairs to assess linear and probabilistic relationships.
• Gene pair selection based on calculated probability masses and False Discovery Rate (FDR) analysis.
• Network construction and module detection through clustering.

Main Results:

• Selected gene pairs frequently share Gene Ontology terms, indicating functional similarity.
• Network clustering successfully generated modules enriched with genes of similar functions.
• Analysis of gene promoters within modules revealed enrichment of known transcription factor binding sites, suggesting prediction of regulatory relationships.

Conclusions:

• The proposed method effectively predicts functional and regulatory relationships between genes using gene expression data.
• The constructed gene networks and identified modules provide insights into gene function and biological pathways.
• The approach demonstrates potential for uncovering transcription factor-mediated gene regulation.