Online Program Home
  My Program

Abstract Details

Activity Number: 488 - Hypothesis Testing When Signals Are Rare and Weak
Type: Invited
Date/Time: Wednesday, August 2, 2017 : 10:30 AM to 12:20 PM
Sponsor: Section on Risk Analysis
Abstract #322426
Title: Testing of Mediation Effects in Genome-Wide Studies: Testing a Large Number of Composite Null Hypotheses
Author(s): Xihong Lin*
Companies: Harvard TH Chan School of Public Health

It is often of scientific interest in assessing the mediation effects of a large number of mediators, e.g., DNA methylations in genome-wide epigenetic studies, that lie in the causal pathway of an exposure on a clinical outcome. Testing for the mediation effect is statistically challenged by the fact that the null hypothesis is composite. We show that the standard mediation analysis tests using the maximum p-value method and the Wald test based on the product method fail and give too conservative tests in genome-wide mediation analysis when one needs to test for a large number of composite null hypotheses. We propose a divide-aggregate test (DAT) for assessing mediation for genome-wide epigenetic studies. We show that this composite testing procedure performs much better than existing methods for genome-wide epigenetic studies where the signals are usually very sparse. Simulation studies were conducted to evaluate the type I error rates and powers under various practical settings. An application to the Normative Aging Study (NAS) identified putative DNA methylation CpG sites as mediators in the causal pathway from smoking behavior to lung functions.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2017 program

Copyright © American Statistical Association