Bayesian Method for Testing Differential Directed Acyclic Graphs

• Activity Number: 665
• Type: Contributed
• Date/Time: Thursday, August 4, 2016 : 8:30 AM to 10:20 AM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Hongmei Zhang* and Xianzheng Huang and Shengtong Han and Wilfried Karmaus
• Companies: University of Memphis and University of South Carolina and The University of Chicago and University of Memphis
• Keywords: Bayesian network ; Variable selection ; Differential networks

Abstract

Graphical models are essential to describe networks among different factors. Various methods to construct graphs have been proposed. In practice, it is critical to assess the agreement between networks constructed under different treats, e.g., a network formed by a large number of epigenetic factors (e.g., DNA methylation) among subjects who smoke versus that among non-smokers. There is rather limited effort in this area. We propose a Bayesian method to build directed acyclic graphs (DAGs) based on a given order of nodes and simultaneously test the agreement between DAGs. The network construction and differential graphs testing are built upon the concept of variable selection. Simulations demonstrate the applicability of the method and a real data application to an epigenetic data is implemented to illustrate the method.