A New gene selection approach based on Minimum Redundancy-Maximum Relevance (MRMR) and Genetic Algorithm (GA)

Gene expression data usually contains a large number of genes (several thousand or more) but a small number of samples (usually <100). Among all the genes, many are irrelevant, insignificant or redundant to the discriminant problem under investigation. Hence the identification of informative genes, which have the greatest power for classification, is of fundamental and practical importance to the investigation of specific discriminant problems, such as cancer versus non-cancer or different tumor types. In this paper, we propose a two-stage selection algorithm by combining minimum redundancy maximum relevance (MRMR) and genetic algorithm (GA). In the first stage, MRMR is used to filter noisy and redundant genes in high dimensional microarray data. In the second stage, GA and SVM work together for selecting the highly discriminating genes. The proposed method is tested on five open datasets: NCI, Lymphoma, Lung, Leukemia and Colon. In the experiments, we compare our proposed MRMR-GA algorithm with GA-SVM wrapper and MRMR filter. The experimental results show that the proposed method has excellent selection and classification performances.