Unraveling Gene Function Using Gene Co-Expression Networks in the Domestic Horse

Despite their success, the causal genes and functional mechanism underlying Genome Wide Association (GWA) studies, in many cases, remain unknown. GWA studies for complex traits are limited by poor resolution in SNP boundaries as well as potentially dozens of different effective loci being associated with a trait of interest. To mitigate these issues, we have systematically integrated several sources of high-dimensional genomic data in order to better describe and inter-relate putative loci associated with traits of interest. Here, we utilize these genomic tools to functionally annotate loci previously linked to Equine Metabolic Syndrome (EMS).

We built gene co-expression networks using tissue specific RNA-Seq measured from 11 different organs or tissues. Networks encompass patterns of gene co-expression providing a high level, functional view of the transcriptional landscape of the horse. Differential expression analysis between extreme phenotype scores quantifying EMS identifies 529 genes among muscle and adipose tissues indicating these tissues are biologically active for our trait. In muscle/adipose networks, there were 6.5X and 8.3X enrichment for co-expression among genes within AgriGO terms indicating networks capture biologically relevant information. Functional analysis for EMS was performed by directly extracting co-expression interactions among genes near significant SNPs. We characterized 183 loci associated with EMS in Welsh Ponies and Morgans finding 21 significantly co-expressed genes in networks of which 3 had previous AgriGO annotations related to muscular function. Future work will focus on building disease specific, differential co-expression networks targeting specific changes in transcription between EMS and healthy individuals.