Differential Network Analysis in Genomics

• Activity Number: 289
• Type: Contributed
• Date/Time: Tuesday, August 5, 2014 : 8:30 AM to 10:20 AM
• Sponsor: ENAR
• Author(s): Min Jin Ha*+ and Veera Baladandayuthapani and Kim-Ahn Do
• Companies: and MD Anderson Cancer Center and MD Anderson Cancer Center
• Keywords: co-expression network ; differential network ; baseline network ; covariance model

Abstract

The gene co-expression analysis is a popular approach to detect functionally related genes under a static condition. In differential co-expression analysis, co-expression networks are measured and compared across conditions to identify co-expressions that are differentially present, absent or modified. The relations that are highly altered between conditions can be detected while ubiquitous relations are filtered out in the differential networks. The existing differential co-expression methods are limited to correlation networks, which were independently estimated from separate condition specific data. We introduce a baseline network, which represents the co-expressions among genes when there is no effect of condition. After eliminating those baseline relations, we jointly estimate the condition specific networks by a covariance model. Extensive simulation studies demonstrate the better performance of our method than the separate estimations. In real data analysis, we discuss several hypothesis of interest and the test statistics to evaluate the differential networks.