Title: GeneRank-based partly adaptive group-penalised multinomial regression for microarray classification
Authors: Juntao Li; Wenpeng Dong
Addresses: College of Computer and Information Engineering, Henan University of Economics and Law, Zhengzhou 450002, China ' Henan Engineering Laboratory for Big Data Statistical Analysis and Optimal Control, School of Mathematics and Information Science, Henan Normal University, Xinxiang 453007, China
Abstract: This paper proposes a partly adaptive group-penalised multinomial regression for gene selection. Weights with biological significance are constructed by combing the gene expression information with gene ontology network via GeneRank. By introducing the weights into group lasso penalty, the partly adaptive group-penalised multinomial regression is proposed. Two algorithms for fitting the proposed model are presented on the base of blockwise descent. Experimental results on gene expression data of yeast diauxic shift demonstrate that the proposed method can select the stable genes and achieve the better classification performance.
Keywords: multinomial regression; microarray classification; GeneRank; gene selection; group lasso penalty; gene ontology network; bioinformatics; gene expression data; yeast; diauxic shift.
DOI: 10.1504/IJDMB.2016.080674
International Journal of Data Mining and Bioinformatics, 2016 Vol.16 No.3, pp.252 - 268
Received: 08 Aug 2015
Accepted: 02 Oct 2016
Published online: 01 Dec 2016 *