USVM: Selection of SNPs in Diseases Association Study Using UMDA and SVM

With the rapid development of high-throughput genotyping technologies, more and more attentions are paid to the disease association study identifying DNA variations that are highly associated with a specific disease. One main challenge for this study is to find the optimal subsets of Single Nucleotide Polymorphisms (SNPs) which are most tightly associated with diseases. Feature selection has become a necessity in many bioinformatics applications. In this paper, we propose a wrapper algorithm named USVM which combines Univariate Marginal Distribution Algorithm (UMDA) and Support Vector Machine (SVM) for disease association study. USVM not only eliminates the redundancy of feature, but also solves the problem of SVM's parameters selection. We use USVM to analyze the Crohn's disease (CD) dataset including 387 samples and each one has 103 SNPs. The experimental results show that our algorithm outperforms the current algorithms including DNF, CSP, ORF and so on.