A hybrid filter/wrapper gene selection method for microarray classification

Selecting a small subset of genes out of the thousands of genes in microarray is a crucial problem for accurate cancer classification. A hybrid gene selection and cancer classification method is proposed, which is consisted of a novel preselection method and special genetic algorithms (GAs). In the new proposed method, a classifier is also constructed. To evaluate the gene selection method and the classifier, the method is applied to three open data sets, and the experiments show very well the performance.