Gene selection for classification of microarray data based on the Bayes error

In this study, we propose a new method, Based Bayes error Filter (BBF), to select relevant genes and remove redundant genes in classification analyses of microarray data. The effectiveness and accuracy of this method is demonstrated through analyses of five publicly available microarray datasets. The results show that our gene selection method is capable of achieving better accuracies than previous studies, while being able to effectively select relevant genes, remove redundant genes and obtain efficient and small gene sets for sample classification purposes.The proposed method can effectively identify a compact set of genes with high classification accuracy. This study also indicates that application of the Bayes error is a feasible and effective wayfor removing redundant genes in gene selection.One of the major applications of DNA microarray technology is to perform sample classification analyses between different disease phenotypes, for diagnostic and prognostic purposes [1-3]. The classification analyses involve a wide range of algorithms such as differential gene expression analyses, clustering analyses and supervised machine learning [4-6], etc. In classification analyses of microarray data, gene selection is one of the critical aspects [5,7-12]. Efficient gene selection can drastically ease computational burden of the subsequent classification task, and can yield a much smaller and more compact gene set without the loss of classification accuracy [13-17]. In addition, a smaller number of selected genes can be more conveniently and economically used for diagnostic purposes in clinical settings [18].In the presence of thousands of genes in microarray experiments, it is common that a large number of genes are not informative for classification because they are either irrelevant or redundant [19]. Based on a review of the definitions of relevance [20,21], the genes can be classified into three disjoint categories, namely, strongly relevant, weakly relevant, and i