Inferring Biological Networks Using the Bayesian Graphical Lasso with Informative Priors

• Activity Number: 181
• Type: Contributed
• Date/Time: Monday, July 30, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Christine Peterson*+ and Marina Vannucci
• Companies: Rice University and Rice University
• Keywords: Graphical models ; Bayesian graphical lasso ; informative priors ; metabolic networks

Abstract

To infer the structure of biological networks, we apply the Bayesian graphical lasso with informative priors that incorporate known relationships between covariates. To encourage sparsity, the Bayesian graphical lasso places double exponential priors on the entries of the precision matrix. An informative prior can be constructed by allowing each double exponential prior to have a unique shrinkage parameter. By formulating the gamma hyperprior on these shrinkage parameters so that probability mass is shifted towards zero for nodes that are close together in a reference network, edges are encouraged between covariates with known relationships. These informative priors can improve the reliability of inference given small sample sizes. Applications include inference of metabolic networks from NMR spectra and inference of gene-gene interaction networks from gene expression data.