Epistatic analysis of NSAIDs hypersensitivity using high performance computing

Abstract

Genome wide association studies (GWAS) have tended to focus on the association between single nucleotide polymorphisms (SNPs) and human diseases by looking at individual variants. However, complex diseases are more likely due to certain combinations of genetic variants rather than single variants acting independently in an additive fashion. As such, the interactions between multiple SNPs, epistatic interactions, have the potential to provide insights about their underlying causes and mechanisms. Using epistatic analysis methods, it is possible to identify SNP pairs associated with disease; these can then be mapped to genes allowing the inference of a gene interaction network. This network can be analysed using graph theory metrics to identify important hub genes, which although often insignificant on their own can be important in combination with other variants. Here, we analyse SNP and gene interaction networks obtained from a nonsteroidal anti-inflammatory drugs (NSAIDs) hypersensitivity GWAS dataset, obtained through the application of epistatic analysis methods. We identify several combinations of SNPs and genes potentially involved in and predictive of NSAIDs hypersensitivity, that may not be identified through a single SNP association approach.