Functional Module Analysis for Gene Coexpression Networks with Network Integration.

Area of Science:

• Systems Biology
• Bioinformatics
• Computational Biology

Background:

• Biological networks are crucial for understanding molecular interactions.
• Identifying modules in individual networks is limited by data noise and parameter tuning.
• Integrating multiple networks offers a more robust approach to module analysis.

Purpose of the Study:

• To develop an effective method for module identification from multiple biological networks under different conditions.
• To combine module identification within individual networks and module alignment across networks.
• To improve the discovery of disease-specific and conserved biological modules.

Main Methods:

• Formulated module identification as an optimization problem integrating intra-network and inter-network module analysis.
• Developed an approximation algorithm based on eigenvector computation.
• Applied the method to human gene coexpression networks from cancer and morbidly obese patient tissues.

Main Results:

• The proposed method outperforms existing approaches, particularly when modules differ across networks.
• Identified 13 modules (with 3 complete subgraphs) from cancer networks and 11 modules (with 2 complete subgraphs) from obesity networks.
• Validated modules using Gene Ontology and KEGG pathway enrichment analyses, confirming biological relevance.

Conclusions:

• The developed method effectively identifies biologically relevant modules from multiple networks.
• The identified modules provide insights into disease mechanisms and potential therapeutic targets.
• Findings support the integration of multiple biological networks for comprehensive molecular interaction analysis.