Multilevel Gaussian Graphical Model for Gene and Pathway Networks

• Activity Number: 254
• Type: Contributed
• Date/Time: Monday, August 5, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistical Learning and Data Mining
• Author(s): Lulu Cheng*+ and Inyoung Kim
• Companies: Virginia Tech and Virginia Tech
• Keywords: Pathway and Gene Network ; Multilevel Gaussian Graphical Model ; Graphical LASSO

Abstract

Most recent studies were focused on constructing networks among genes using Gaussian graphical model. However, none of the studies were applicable to construct the networks among gene sets (pathways), as well as genes. Therefore, in this talk, we proposed a multilevel Gaussian graphical model (MGGM), in which one level describes the networks for genes and the other for pathways. We developed a multilevel L1 penalized likelihood approach to achieve the sparseness on both levels. And consequently, we developed an iterative weighted graphical LASSO algorithm for MGGM. Some asymptotic properties of the estimator were illustrated. Our simulation results supported the advantages of our approach; our method estimated the network more accurate on the pathway level, and sparser on the gene level. We also demonstrated usefulness of our approach using a canine genes-pathways data set.