Gene Selection in Microarray Data Using an Improved Approach of CLONALG

Abstract

Expression of a gene at a molecular level can lead to the diagnosis and prediction of diseases especially cancer. In microarray datasets, large amount of genes exists and this makes the classification process complex. Since the selected genes must contain enough number of features for classification, selecting suitable features for classification is crucial. Due to the complexity of the problem, it is accepted as a NP-hard problem and an evolutionary approach; Clonal Selection Algorithm (CLONALG) is chosen to produce solution for this problem. In this paper, an Improved Clonal Selection Algorithm (ICSAT) with K-nearest neighbor (K-NN) method are used together to select the relevant features of genes for an accurate gene classification. The proposed method ICSAT-KNN is tested on three gene datasets and compared with two existing algorithms. The obtained classification accuracy values are quite competitive or even better than the values of the compared algorithms. Experimental results show that ICSAT-KNN method can serve as a reliable tool for feature selection and accurate data classification.