A Hybrid Algorithm for Disease Association Study

Abstract

Single nucleotide polymorphisms (SNPs), which are the most common form of DNA variations, have great potential as a medical diagn ostic tool. However, compared to the number of SNPs involved, the available training data sets generally have a fairly small sample size, which is a main challenge to traditional data analysis methods. This paper proposed an improved univariate marginal distribution algorithm (UMDA) named multi-population UMDA (MPUMDA) for disease association study. In order to illustrate the effectiveness of our algorithm, we compared it with some current known methods, and the results showed that our method is potentially interesting as an alter-native tool in disease association study.