Selecting informative genes from leukemia gene expression data using a hybrid approach for cancer classification

The development of microarray-based highthroughput gene profiling has led to the hope that this technology could provide an efficient and accurate means of diagnosing and classifying cancers. However, the large amount of data generated by microarrays requires effective selection of informative genes for cancer classification. Key issue that needs to be addressed is a selection of small number of informative genes that contribute to a disease from the thousands of genes measured on microarrays. This work deals with finding the small subset of informative genes from gene expression microarray data which maximize the classification accuracy. We introduce an improved version of hybrid of genetic algorithm and support vector machine for genes selection and classification. We show that the classification accuracy of the proposed approach is superior to a number of current state-of-the-art methods of one widely used benchmark dataset. The informative genes from the best subset are validated and verified by comparing them with the biological results produced from biology and computer scientist researchers in order to explore the biological plausibility.