A particle swarm optimization based gene identification technique for classification of cancer subgroups

Abstract

Microarray gene expression data generally consist of huge number of genes compared to very less number of samples available. Therefore it is a stimulating task to identify a small subset of relevant genes from microarray gene expression data where the identified genes can solely be used for accurately classifying the cancer subgroups. Therefore, in this paper a computationally efficient but accurate gene identification technique has been proposed. At the onset the t-test method has been utilized to reduce the dimension of the dataset and then the proposed particle swarm optimization based approach has been employed to find useful genes. The proposed method has been applied on the small round blue cell tumor (SRBCT) data to classify the four subgroups specifically neuroblastoma, non-Hodgkin lymphoma, rhabdomyosarcoma and Ewing sarcoma. The results demonstrate that the proposed technique could identify only fourteen genes that can be efficiently exploited for the diagnostic prediction task with high accuracy.