PROJECT TITLE :

Parallelizing Epistasis Detection in GWAS on FPGA and GPU-Accelerated Computing Systems

ABSTRACT:

High-throughput genotyping technologies (like SNP-arrays) allow the rapid assortment of up to a couple million genetic markers of a private. Detecting epistasis (based on 2-SNP interactions) in Genome-Wide Association Studies is a crucial however time consuming operation since statistical computations should be performed for every combine of measured markers. Computational ways to detect epistasis thus suffer from prohibitively long runtimes; e.g., processing a moderately-sized dataset consisting of concerning five hundred,000 SNPs and 5,00zero samples requires many days using state-of-the-art tools on a commonplace 3 GHz CPU. During this paper, we have a tendency to demonstrate how this task will be accelerated using a combination of fine-grained and coarse-grained parallelism on 2 different computing systems. The first design relies on reconfigurable hardware (FPGAs) while the second architecture uses multiple GPUs connected to the identical host. We have a tendency to show that each systems will achieve speedups of around four orders-of-magnitude compared to the sequential implementation. This significantly reduces the runtimes for detecting epistasis to solely a jiffy for moderatelysized datasets and to some hours for large-scale datasets.