Genome-wide Gene-gene Interaction Analysis For Complex Diseases Using CUDA Platform

Genome-wide gene-gene interaction scan is a promising method to study susceptibility epistasis effects of complex diseases, but its huge computational burden is the biggest challenge for practice. Graphic Processing Unit (GPU) is highlighted to have extremely high parallel computing ability and low cost. In this study, we developed a GPU-based analytical method, named as SHEsisEpi, which purely focuses on risk epistasis in genome-wide association study (GWAS) of complex traits, excluding the contamination of marginal effects caused by single locus association. In the practice of analyzing WTCCC's GWAS data of bipolar disorder with 500K SNPs, our algorithm only used 27 hours to finish the exhaustive scan and was more than 300 times faster than CPU-based analysis on our system. We demonstrated that SHEsisEpi was very efficient, low cost and applicable.Furthermore, by genotyping the top finding meeting out criterias (p=5.37×10-12) and its nearby SNP pairs in another independent 475 bipolar disorder patients and 480 normal controls from Chinese Han population, we validated these findings, related with 2 gene pairs, conferring risk (a=0.05) to bipolar disorder. Binary files and source codes of our program can be downloaded athttp://analysis.bio-x.cn from the main menu of SHEsis.