Multiple Testing of Subnetworks with Application to Identification of Pathway Crosstalk

• Activity Number: 111
• Type: Invited
• Date/Time: Monday, August 10, 2015 : 8:30 AM to 10:20 AM
• Sponsor: IMS
• Author(s): Yin Xia* and Tony Cai and Tianxi Cai
• Companies: The University of North Carolina at Chapel Hill and University of Pennsylvania and Harvard University
• Keywords: Conditional dependence ; false discovery rate ; Gaussian graphical model ; multiple testing ; pathway cross-talk ; subnetwork

Abstract

Motivated by applications on detection of gene pathway crosstalk in genomics, we consider in this talk simultaneous detection of interactions among subsets of variables that account for joint effects from all other variables of interest. Under a Gaussian graphical model framework, the problem is translated into the statistical problem of simultaneous testing of sub-matrices of a high-dimensional precision matrix. A test statistic is constructed for testing the conditional independence of a given pair of gene sets and it has asymptotically the same limiting null distribution as a mixture of chi-square random variables with certain correlation structures which needs to be incorporated appropriately. In addition, a multiple testing procedure with false discovery rate control is developed to infer the specific structures of the subnetworks. Theoretical justification is provided to ensure the validity of the proposed tests. A simulation study demonstrates that the proposed tests maintain the desired error rates under the null and have good power under the alternative. The methods are applied to a breast cancer gene expression study to identify interactions between gene pathways.