Genome-wide Gene-gene Interaction Analysis for Complex Diseases Using CUDA Platform
|School||Shanghai Jiaotong University|
|Keywords||Gene - gene interactions Bipolar disorder GPU Programming CUDA|
Complex disease genome-wide gene - gene scan can be effectively found that the contribution of the interaction of multiple genes on disease susceptibility . However the amount of calculation for this scanning is usually very large, limits its practical application . Modern graphics processor (GPU, Graphics ) has a very powerful parallel computing ability , while the price is low . Therefore, in this study , we use GPU-based algorithm , namely SHEsisEpi . It is only associated with a plurality of loci in the susceptibility of the entire genome from a single locus associated analysis results ( i.e. marginal function) . We analyzed a the WTCCC database provides bidirectional affective disorder (bipolar disorder, BPD) data , only took 27 hours to complete a total of 500K SNP genome-wide scan , nearly 300 times faster than traditional CPU program . Then by the results of the scan filter ( p = 5.37 × 10-12 ) , the smallest p-value and meet the conditions of the SNP combination , repeat the experiment with independent sample ( 475 BPD cases and 480 control group , the Han population ) prove that the combination of two pairs of genes with BPD susceptibility ( a = 0.05) . Executable files and source code from SHEsis Home the download ( http://analysis.bio-x.cn ) .