Online Program Home
My Program

Abstract Details

Activity Number: 232
Type: Topic Contributed
Date/Time: Monday, August 1, 2016 : 2:00 PM to 3:50 PM
Sponsor: Biometrics Section
Abstract #320060
Title: Region-Based Hypothesis Test of Mediation Effect with Epigenomic Data Using Kernel Machine Regression
Author(s): Jincheng Shen*
Companies: Harvard T.H. Chan School of Public Health
Keywords: Mediation Analysis ; Multiple Mediators ; Kernel Machine

With development of genetic and genomic technologies, it is of increasing interest in biomedical studies to explore the genetic and epigenetic mechanisms that mediate a disease causing process. Epigenetic studies for complex diseases have shown that the effect of many risk factors are likely to be mediated in a collaborative fashion through multiple probes in a genomic region, such as CpG islands or shores. We propose a Wald-type test for the overall natural indirect effect targeting the multiple mediator problem. Kernel machine based approach is employed to account for potential interactions and nonlinearity among mediators. Under certain regularity conditions, we develop a simple variance estimator when comparing coefficients from different models. The proposed test is evaluated on data simulated from various underlying mediation pathway structures and demonstrates substantial gain in power when the effects are nonlinear. We also apply the proposed test on the Normal Aging Study to investigate the effect of smoking on lung function measures that mediated by the DNA methylation level on regions over the whole genome.

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

Back to the full JSM 2016 program

Copyright © American Statistical Association