Title: A novel filter feature selection method for paired microarray expression data analysis

Authors: Zhongbo Cao; Yan Wang; Ying Sun; Wei Du; Yanchun Liang

Addresses: Key Laboratory of Symbolic Computation and Knowledge Engineering of the Ministry of Education, College of Computer Science and Technology, Jilin University, Changchun 130012, China ' Key Laboratory of Symbolic Computation and Knowledge Engineering of the Ministry of Education, College of Computer Science and Technology, Jilin University, Changchun 130012, China ' Key Laboratory of Symbolic Computation and Knowledge Engineering of the Ministry of Education, College of Computer Science and Technology, Jilin University, Changchun 130012, China ' Key Laboratory of Symbolic Computation and Knowledge Engineering of the Ministry of Education, College of Computer Science and Technology, Jilin University, Changchun 130012, China; College of Chemistry, Jilin University, Changchun 130012, China ' Key Laboratory of Symbolic Computation and Knowledge Engineering of the Ministry of Education, College of Computer Science and Technology, Jilin University, Changchun 130012, China

Abstract: In recent years, a large amount of microarray data sets are produced with tens of thousands of genes. Feature selection has become a very sharp tool to select the informative genes. However, few feature selection methods consider the effect of paired samples, which are much more considered in the experiments of these years. Here, we propose a new feature selection method for paired microarray data sets analysis. It uses the fold change instead of the subtraction in the original approach, measures the statistical significant using the q-value of False Discovery Rate (FDR) and also decreases the influence of redundant genes. We compare the proposed method with another six existing methods in predict performance, stability of gene lists, functional stability and functional enrichment analysis using six kinds of paired cancer data sets. Comparison results show that our proposed method achieves better effectiveness, stability and consistency when it is applied to paired data sets.

Keywords: feature selection; filter method; paired microarray data; gene correlation; paired t-test; statistical significant; redundant genes; differentially expressed genes; data mining; bioinformatics; gene expression data; paired samples; paired cancer datasets.

DOI: 10.1504/IJDMB.2015.070071

International Journal of Data Mining and Bioinformatics, 2015 Vol.12 No.4, pp.363 - 386

Received: 02 Aug 2014
Accepted: 08 Sep 2014
Published online: 26 Jun 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article