A Framework for Accessible Cluster-Enabled Epistatic Analysis.

Complex diseases are typically caused by joint effects of multiple genetic variations, rather than a single variant. Multiple single nucleotide polymorphism (SNP) interactions, epistatic interactions, potentially provide information about the causes of complex diseases, building on studies that focus on the association between single SNPs and phenotypes. However, execution of epistatic methods on desktop computers is not practical, owing to the huge number of interactions that have to be calculated. These models have tended to be command line based, presenting a barrier for users such as biologists that are not comfortable with this environment. To overcome this, we present a framework with a front-end GUI deployed on a cluster that allows users to analyse genotype/phenotype correlations using computationally accelerated epistatic models. The parallel processing of the data results in a typical epistatic analysis taking a few days, presenting a feasible approach for the analysis of genetic variants associated with disease.