High-Dimensional Hypothesis Screening Using P-Value Perturbation

• Activity Number: 589
• Type: Invited
• Date/Time: Thursday, August 2, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Ali Shojaie*+ and Moulinath Banerjee and George Michailidis
• Companies: University of Washington and University of Michigan and University of Michigan
• Keywords: Hypothesis screening ; high dimensional inference ; p-value perturbation ; pathway enrichment analysis ; large p small n

Abstract

Assessing the enrichment of biological pathways is a critical problem in systems biology, which has received much attention in recent years. However, available methods either do not provide an asymptotic analysis of the properties of test statistics, or are based on strong simplifying assumptions that do not hold in real applications. We propose a novel method, based on a perturbed version of regular p-values, for screening significance (or active) genes in high dimensional pathways under arbitrary correlation structures. The new methodology results in simultaneous reduction in false positive and false negative rates, and the resulting method for enrichment analysis of pathways offers improved inferential properties over existing methods. We establish the consistency of the proposed hypothesis screening procedure in high dimensional settings, and evaluate its performance on variety of simulated and real data examples.