A Novel Hybrid Method for Gene Selection of Microarray Data

Nowadays, using microarray technologies, we can analysis thousands of gene expression values at one time. So we can have an overall understanding of the cell. However, gene expression profiles commonly have thousands of gene expression values but only limited tissue samples, for some reasons. In this paper, we want to identify the most significant genes that demonstrate the highest capabilities of discrimination between the classes of samples. We first use filter method to rank the genes in terms of their expression difference, and then a clustering method based on k-NN principle is used in clustering gene expression data. Afterwards a maximal cliques search method is applied at each gene clustering. Besides significant candidate genes determined by our clustering method, we also select genes with the highest "information exponential" in maximal cliques at each gene clustering and then further reduce the number of genes. A support vector machine is applied to validate the classification performance of candidate genes. The experimental results demonstrate the effectiveness of our method in addressing the problem.