Support Vector Machines with Disease-Gene-Centric Network Penalty for High-Dimensional Microarray Data

• Activity Number: 379
• Type: Topic Contributed
• Date/Time: Tuesday, August 4, 2009 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Yanni Zhu*+ and Wei Pan and Xiaotong Shen
• Companies: The University of Minnesota and The University of Minnesota and The University of Minnesota
• Keywords: support vector machine ; penalization method ; microarray gene expression

Abstract

We propose disease-gene-centric support vector machine (DGC-SVM) that explicitly incorporates gene network to build classifiers. By considering network as being centered on certain disease gene(s), we convert an undirected network into a directed acyclic graph that imposes a hierarchy on the network and thus facilitates us to define groups of genes according to two ways of grouping - pathway or partial tree. The hinge loss penalized by the sum of the L-infinity-norm being applied to each group leads to DGC-SVM that encourages gene selection along pathways. The simulation studies show that DGC-SVM not only detects more disease genes along pathways than standard-SVM and L1-SVM but also captures disease genes that affect the outcome weakly. Two real data applications demonstrate that DGC-SVM improves gene selection with predictive accuracy comparable to standard-SVM and L1-SVM.