Bipartite Community Structure of eQTLs

Area of Science:

• Genetics
• Systems Biology
• Computational Biology

Background:

• Genome-Wide Association Studies (GWAS) and expression quantitative trait locus (eQTL) analyses identify genetic associations with human phenotypes.
• Understanding the combined effects of multiple genetic variants with weak effects remains a challenge.
• A hypothesis suggests complex network interactions among single nucleotide polymorphisms (SNPs) influence functional processes leading to complex phenotypes, including diseases.

Purpose of the Study:

• To present CONDOR, a novel computational method for analyzing genetic networks.
• To contextualize SNPs within their functional networks using graph theory.
• To apply CONDOR to eQTL data in chronic obstructive pulmonary disease (COPD) to identify disease-associated genetic patterns.

Main Methods:

• CONDOR represents cis- and trans-acting SNPs and their associated genes as a bipartite graph.
• The method utilizes the modular structure of the graph to place SNPs into a functional context.
• Application to COPD eQTL data involved identifying global and local network hubs and community structures.

Main Results:

• Global network "hub" SNPs lacked direct GWAS associations for COPD.
• The SNP-gene network organized into 52 communities, many enriched for specific functional gene classes.
• Local network hubs ("core SNPs") within communities were enriched for GWAS SNPs associated with COPD and other diseases.

Conclusions:

• Disease SNPs are associated with the perturbation of gene expression networks, rather than single SNP-gene effects.
• Groups of SNPs are linked to the expression of families of functionally related genes.
• The CONDOR method is broadly applicable to diverse disease processes and phenotypic traits beyond COPD.