Accelerating processing speed in pathway research based on GPU

Genome-wide association study (GWAS) has become an effective and successful method to identify disease loci by considering SNPs independently. However, it may be invalid for uncovering the disease loci that not reaching a stringent genome-wide significance threshold. As a result, multi-SNP GWAS is developing rapidly as a complement to traditional GWAS. However, the high computational cost becomes a major limitation for it. The graphical processing unit (GPU) is a programmable graphics processor which has powerful parallel computing ability. And with the development, GPUs have been feasible for many scientific studies. Hence, we are motivated to use GPUs for pathway-based GWAS to improve computational efficiency. The experiment results attained showed the speed-up ratio can reach up to more than 160.