Depicting Gene Co-expression Networks Underlying eQTLs

Abstract

Deciphering the biological mechanisms underlying a list of genes whose expression is under partial genetic control (i.e., having at least one eQTL) may not be as easy as for a list of differential genes. Indeed, no specific phenotype (e.g., health or production phenotype) is linked to the list of transcripts under study. There is a need to find a coherent biological interpretation of a list of genes under (partial) genetic control. We propose a pipeline using appropriate statistical tools to build a co-expression network from the list of genes, then to finely depict the network structure. Graphical models are relevant because they are based on partial correlations, closely linked with causal dependencies. Highly connected genes (hubs) and genes that are important for the global structure of the network (genes with high betweenness) are often biologically meaningful. Extracting modules of genes that are highly connected permits a significant enrichment in one biological function for each module, thus linking statistical results with biological significance. This approach has been previously used on a pig eQTL dataset (Villa-Vialaneix et al. 2013) and was proven to be highly relevant. Throughout the present chapter, we define statistical notions linked with network theory, and apply them on a reduced dataset of genes with eQTL that were found in the pig species to illustrate the basics of network inference and mining.