Parallel Computing in Genome-Wide Association Studies

[摘要] Technological advances have greatly reduced the cost and timeneeded to sequence an individual’s genome (i.e., identify the DNAbase pairs that constitute an individual’s genome), leading to dramaticgrowth in the amount of genomic data to analyze. Although thedevelopment of novel sequencing technologies has been prolific,the growth in methods to detect associations between phenotypes(i.e., physical traits of interest) and locations along the genome,more specifically single nucleotide polymorphisms (SNPs), has beencomparatively limited. Hindrances to methodological developmentinclude a lack of available computing power to analyze data in afeasible time frame (i.e., in a time frame useful to researchers). Onetechnique used to speed up computation (i.e. make computationmore efficient) is parallelization, where multiple tasks are performedsimultaneously on multiple cores or threads within a machine [1].Even with computational and methodological limitations, analysis ofgenome-wide association study (GWAS) data has led to the inferenceof connections among several phenotypes and SNPs [2,3].