Abstract 1444: Weighted Gene Co-expression Network Analysis of Adipose and Liver Reveals Gene Modules Related to Plasma HDL Levels and Containing Candidate Genes at Loci Identified in Genome Wide Association Studies

Abstract

The genetic and environmental factors contributing to HDL-cholesterol levels are highly complex. For example, a recent meta-analysis of three genome wide association studies (GWAS), consisting of over 9000 individuals, revealed several loci, but altogether these explained less than 10% of HDL variation. Since HDL has a heritability of about 50%, there clearly must be many as yet unidentified factors. To better address this complexity, we have utilized integrative genomic approaches to relate common DNA variation to gene networks and HDL metabolism. We report a Weighted Gene Co-expression Network Analysis (WGCNA) of genome-wide expression data from a CAST X C57BL6/J F2 intercross. WGCNA is a systems-based gene expression analysis and gene screening method. It utilizes co-expression patterns among genes to identify gene modules (groups of highly co-expressed genes) significantly associated with a clinical trait, in this case plasma HDL levels. Co-expression modules may represent cellular processes and interacting pathways that provide a bridge between individual genes and a systems-level view of the organism. A module-centric analysis effectively alleviates the multiple testing problems inherent in microarray data analysis and can be considered a biologically motivated data-reduction scheme. Using data from liver and adipose tissues, we have identified several modules strongly associated with plasma HDL levels (p-values ranging from below 1e-20 to 1e-5). Gene ontology and functional enrichment analysis indicate that these modules are indeed biologically meaningful. The modules contain variants of several genes under loci that were recently implicated by three GWA studies: liver modules include GCKR, ANGPTL4, ABCA3, APOA1, and APOA4, while the adipose modules include ABCA6, ANGPTL11 and 12, MMAB, MLXIPL, SORT1, PBX4, PLTP, and APOL6. Thus, our study also serves to help identify likely candidates from GWAS. In conclusion, applying WGCNA methods reveals modules that are biologically meaningful, statistically significant, and enriched for genes and pathways related to HDL metabolism and transport.