Research On Epistasis Of Complex Disease Analysis Method Based On Ant Colony Algorithm

Unlike Mendelian disease,the formation and development of complex disease usually involves the interaction of multiple genes or gene-environment interaction which is epistasis.The exploration of susceptibility genes of complex disease in genome wide and the way to influence the disease will contribute to a more comprehensive understanding of the pathogenesis of complex diseases,in order to achieve complex disease prevention,diagnosis and treatment.Although single nucleotide polymorphism(SNP)chips for complex diseases have produced vast amounts of data,the large scale data has its characteristic of high dimension and small sample and so on,so that the analysis of this kind of data is very difficult.The studies of how to reduce dimensionality effectively,and retention of the critical SNPs combination of epistasis,and effectively portray the relationship between epistasis and complex diseases have become important research area in complex disease genome-wide association.Thus,in this paper we proposed an approach based on ant colony algorithm for epistasis analysis and the main work as follows:The epistasis analysis of complex diseases leads to feature combination explosive.At the same time,the low sample characteristics make different SNP loci assessment methods vary greatly.Therefore,learning the strategy of multiple classifier fusion,this paper propose a fusion method with using multiple criteria to comprehensive evaluate each SNP loci,so that it can more reliably exclude unrelated SNP loci to complex diseases,while effectively retain the important single SNP site.After removing a large number of unrelated sites,the search of epistasis in SNPs combination space is still a difficult task.In this phase,we use ant colony algorithm to quickly search the epistasis spatial.In this process,we design the selection probability function according to the characteristics of the SNP data of complex disease,and improve the pheromone update function based on the classification accuracy of the disease state.Thus,our method effectively identifies the epistasis,so that it has satisfactory classification accuracy.Finally,a comprehensive experiment has been carried on simulated data sets of complex diseases.We use complex diseases classification accuracy as well as running time indicator to evaluate the proposed method.Experimental results show that the proposed method of two-stage strategy designed to reduce the time and effectively identify pathogenic epistasis combination has practical significance.