Module representatives for refining gene co-expression modules.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Gene co-expression modules represent biological mechanisms.
• Weighted Gene Co-expression Network Analysis (WGCNA) is a common method for module detection.
• Eigengenes are currently used as centroids in clustering algorithms to refine module memberships.

Purpose of the Study:

• To present four novel module representatives for refining Weighted Gene Co-expression Network Analysis (WGCNA) module membership.
• To improve the identification and biological interpretation of gene co-expression modules.
• To enhance the classification of modules based on phenotype and biological significance.

Main Methods:

• Introduced four new module representatives: eigengene subspace, flag mean, flag median, and module expression vector.
• Applied these representatives within Linde-Buzo-Gray clustering algorithms.
• Refined Weighted Gene Co-expression Network Analysis (WGCNA) module membership using the new representatives.

Main Results:

• The eigengene subspace, flag mean, and flag median capture greater gene expression variance within modules.
• The module expression vector utilizes module co-expression network structure for weighted centroid calculation.
• Module refinement techniques improved module classification between phenotype and biological significance (Gene Ontology terms).

Conclusions:

• The novel module representatives offer improved variance capture and leverage network structure.
• These refined modules demonstrate enhanced biological significance and phenotypic classification.
• The proposed methods advance the identification and interpretation of gene co-expression modules in transcriptomics data.