Article: Molecular pathway identification using a new L1/2 solver and biological network-constrained model Journal: International Journal of Data Mining and Bioinformatics (IJDMB) 2017 Vol.17 No.3 pp.189 - 205 Abstract: Molecular research is moving towards big data epoch. There are various large-scale popular databases abstracted from different biological processes. Integrating such valuable information with the statistical model may shed light on how human cells work from a system-level perspective. In this paper, we propose a novel penalised network-constrained regression model with a new L1/2 solver for integrating gene regulatory networks into an analysis of gene-expression data, where the network is graph Laplacian regularised. Extensive simulation studies showed that our proposed approach outperforms L1 regularisation and old L1/2 regularisation regarding prediction accuracy and predictive stability. We also apply our method to three kinds of cancer data sets. Particularly, our method achieves comparable or higher predictive accuracy than the old solver L1/2 and L1 regularisation approaches, while fewer but informative genes and pathways are selected. Inderscience Publishers - linking academia, business and industry through research

Title: Molecular pathway identification using a new L_1/2 solver and biological network-constrained model

Authors: Hai-Hui Huang; Xiao-Ying Liu; Hui-Min Li; Yong Liang

Addresses: School of Information Science and Engineering, Shaoguan University, Shaoguan, China ' Faculty of Information Technology, State Key Laboratory of Quality Research in Chinese Medicines, Macau University of Science and Technology, Avenida Wai Long, Taipa, Macau 999078, China ' Faculty of Information Technology, State Key Laboratory of Quality Research in Chinese Medicines, Macau University of Science and Technology, Avenida Wai Long, Taipa, Macau 999078, China ' Faculty of Information Technology, State Key Laboratory of Quality Research in Chinese Medicines, Macau University of Science and Technology, Avenida Wai Long, Taipa, Macau 999078, China

Abstract: Molecular research is moving towards big data epoch. There are various large-scale popular databases abstracted from different biological processes. Integrating such valuable information with the statistical model may shed light on how human cells work from a system-level perspective. In this paper, we propose a novel penalised network-constrained regression model with a new L_1/2 solver for integrating gene regulatory networks into an analysis of gene-expression data, where the network is graph Laplacian regularised. Extensive simulation studies showed that our proposed approach outperforms L₁ regularisation and old L_1/2 regularisation regarding prediction accuracy and predictive stability. We also apply our method to three kinds of cancer data sets. Particularly, our method achieves comparable or higher predictive accuracy than the old solver L_1/2 and L₁ regularisation approaches, while fewer but informative genes and pathways are selected.

Keywords: big data; network analysis; variable selection; regularisation; L_1/2 penalty.

DOI: 10.1504/IJDMB.2017.085277

International Journal of Data Mining and Bioinformatics, 2017 Vol.17 No.3, pp.189 - 205

Received: 13 Nov 2015
Accepted: 08 Mar 2017
Published online: 19 Jul 2017 *

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

Title: Molecular pathway identification using a new L1/2 solver and biological network-constrained model

Keep up-to-date

Title: Molecular pathway identification using a new L_1/2 solver and biological network-constrained model