Adapting Community Detection Algorithms for Disease Module Identification in Heterogeneous Biological Networks

Area of Science:

• Systems Biology
• Bioinformatics
• Computational Biology

Background:

• Biological networks represent complex molecular interactions within cells, often exhibiting modular structures linked to specific biological functions.
• Dysfunction in these protein modules is frequently implicated in human diseases, highlighting the need for effective module identification methods.
• Existing community detection algorithms struggle with the dense, interconnected nature of biological networks and the small size of relevant disease modules.

Purpose of the Study:

• To comprehensively assess classic community detection algorithms for biological networks and propose heuristics for identifying small, well-defined 'core' disease modules.
• To analyze the structural characteristics of disease-enriched modules and understand limitations of standard algorithms in detecting them.
• To develop a methodology for extracting high-quality disease modules by incorporating neighboring genes and advocate for overlapping community detection algorithms.

Main Methods:

• Evaluated multiple community detection algorithms on biological networks, developing heuristics to identify compact, core modules.
• Analyzed interaction patterns of known disease genes to understand module structure and cluster quality metrics (modularity, conductance).
• Developed a novel methodology to form disease modules by including neighboring genes of known disease genes, and assessed overlapping community detection algorithms.

Main Results:

• The proposed approach identified 50% more disease-relevant modules compared to traditional methods, emphasizing the importance of compact module identification.
• Standard algorithms detected fewer disease modules due to the specific structure of disease-enriched networks; incorporating neighboring genes improved cluster quality.
• Demonstrated that overlapping community detection algorithms are superior for disease module identification due to the multi-functional nature of genes.

Conclusions:

• Identifying compact and overlapping disease modules significantly enhances the discovery of disease-relevant biological modules.
• The developed methodology provides a robust framework for extracting high-quality disease modules, improving our understanding of disease etiology.
• Overlapping community detection algorithms are essential for accurately capturing the complex, multi-functional roles of genes in human diseases.