Parallel Design And Implementation Of Group Selection

With the development of the neutral theory, molecular population genetics and DNA sequencing technology, DNA molecular data analysis method being set up and improved, the research of population selection, population diversity and population structure is getting more extensive attention.On the basis of theoretical research, this paper further discusses the population selection method of parallel process, in order to improve operating efficiency. In terms of theoretical research, this article from the theory of population selection method, the method of neutral evolution, linkage disequilibrium and haplotype frequency are carried on the thorough research. In terms of practical applications, in order to improve the operation efficiency of population selection method, this article expounds the population selection of parallel computing from three aspects, Tajima’s D algorithm parallel computing in the neutral theory, linkage disequilibrium parallel computing and EM algorithm inference haplotype frequency parallel computing. Through a careful study on the algorithm, parallel granularity partition, load balancing arrangement, scheduling policy implement and thread setting for program are systematically analysed and parallel optimized, it increases speed-up and CPU efficiency.The main calculation of population selection method is based on the sequence of the DNA molecule, Tajima’s D method is adopted Open MP parallel technology to achieve genotype frequency calculation on the single nucleotide polymorphism loci and get neutral theory result. For linkage disequilibrium process, it mainly studies the genotype sequence alignment between different sites of parallel computing, by using genetic distance properties of genetic loci, sliding windows are set and the linkage disequilibrium results are obtained. And then, haplotype frequency of calculation depends on the results of linkage disequilibrium, which mainly studies the method based on EM algorithm to estimate haplotype frequency of parallel computing. In the paper, serial algorithm and parallel algorithm of three methods are compared. The result shows that the parallel algorithm based on Open MP can improve operation efficiency of population selection method. This method is great significance for the work of subsequent population genetic efficient research.