Camoco: A Computational Framework for Inter-Relating GWAS Loci and Unraveling Gene Function Using Co-Expression Networks

Co-expression networks have been shown to be a powerful tool for inferring a gene’s function when little is known about it. With the advent of next generation sequencing (NGS) technologies, the construction and analysis of co-expression networks is now possible in many non-model species. Here, we present a computational framework called Camoco, that performs co-expression based analyses using NGS data. Camoco has been used to rapidly build and analyze RNA-Seq based co-expression networks in both plant and animal species. Ontologies such as GO or MapMan are used to vet networks in a supervised manner by assessing their ability to capture known biological information. Built-in network clustering allows for rapid unsupervised analyses when gene annotations are poor or incomplete. Network comparison methods distinguish which biological clusters are shared or distinct to each network.

Camoco also integrates other functional data such as those derived from genome wide association studies (GWAS). GWAS have become the de facto approach to identify candidate genes in complex traits in non-model species. However, complex traits are often associated with potentially hundreds of statistically significant SNPs, many of which are located outside known gene models. Camoco analyzes the overlap between candidate loci generated from GWAS and the co-expression interactions that occur between them. In empirical studies, Camoco was effectively able to reduce the number of candidate genes by over two orders of magnitude and prioritize genes for follow up analysis. Camoco is implemented in Python 3.4 and is freely available at http://github.com/schae234/Camoco.