|
Abstract:
|
Selecting informative features from tens of thousands of candidate genes becomes increasingly important in current genomic research. A promising approach is to perform variable selection under regression models while incorporating the existing biological structural information such as gene network/pathway, generating biologically meaningful results. Most existing methods focus on the local network structure and require heavy computational costs for the large scale problem. In this work, we propose a novel prior model for Bayesian variable selection over large scale networks in the generalized linear model (GLM) framework: the Thresholded Graph Laplacian Gaussian (TGLG) prior, which adopts the graph Laplacian matrix to characterize the conditional dependence between neighboring predictors accounting for the global network structure. Under mild conditions, we show the proposed model enjoys the posterior consistency with a diverging number of edges and nodes in the network. We illustrate the superiorities of the proposed method compared with existing alternatives via extensive simulation studies and an analysis of the melanoma gene expression dataset in the Cancer Genome Atlas.
|
