Emily Slade1, Peter Kraft. 1. Department of Biostatistics, Harvard T.H. Chan School of Public Health, Boston, Mass., USA.
Abstract
OBJECTIVE: The association between DNA methylation and a trait of interest may depend on an environmental exposure, and incorrectly accounting for this dependence can lead to a reduction in power of the standard tests used in epigenome-wide association studies. We present the M-ME test to jointly test for the main effect of DNA methylation and methylation-environment interaction. METHODS: Through simulation, we compare the power and type 1 error of the M-ME test to a standard marginal test (M test) and a standard interaction test (ME test) under 1,800 different underlying models. These models allow for methylation-environment correlation and measurement error in the exposure. RESULTS: In many true underlying models, either the M test or the ME test has very low power, but the M-ME test has optimal or nearly optimal power to detect a DNA methylation effect in all models considered, including those with methylation- environment dependence and measurement error in the exposure. Type 1 error inflation occurs in the tests when the exposure is measured with error and correlated with DNA methylation. CONCLUSION: The M-ME test is an attractive choice for studies aiming to detect any DNA methylation association when little is known about the epigenetic associations a priori.
OBJECTIVE: The association between DNA methylation and a trait of interest may depend on an environmental exposure, and incorrectly accounting for this dependence can lead to a reduction in power of the standard tests used in epigenome-wide association studies. We present the M-ME test to jointly test for the main effect of DNA methylation and methylation-environment interaction. METHODS: Through simulation, we compare the power and type 1 error of the M-ME test to a standard marginal test (M test) and a standard interaction test (ME test) under 1,800 different underlying models. These models allow for methylation-environment correlation and measurement error in the exposure. RESULTS: In many true underlying models, either the M test or the ME test has very low power, but the M-ME test has optimal or nearly optimal power to detect a DNA methylation effect in all models considered, including those with methylation- environment dependence and measurement error in the exposure. Type 1 error inflation occurs in the tests when the exposure is measured with error and correlated with DNA methylation. CONCLUSION: The M-ME test is an attractive choice for studies aiming to detect any DNA methylation association when little is known about the epigenetic associations a priori.
Authors: F Ferfouri; F Boitrelle; I Ghout; M Albert; D Molina Gomes; R Wainer; M Bailly; J Selva; F Vialard Journal: Andrology Date: 2013-08-28 Impact factor: 3.842
Authors: Lakshitha P Gunawardhana; Peter G Gibson; Jodie L Simpson; Miles C Benton; Rodney A Lea; Katherine J Baines Journal: Epigenetics Date: 2014-08-11 Impact factor: 4.528
Authors: Pan Du; Xiao Zhang; Chiang-Ching Huang; Nadereh Jafari; Warren A Kibbe; Lifang Hou; Simon M Lin Journal: BMC Bioinformatics Date: 2010-11-30 Impact factor: 3.169
Authors: Roberto A Lleras; Richard V Smith; Leslie R Adrien; Nicolas F Schlecht; Robert D Burk; Thomas M Harris; Geoffrey Childs; Michael B Prystowsky; Thomas J Belbin Journal: Clin Cancer Res Date: 2013-07-26 Impact factor: 12.531
Authors: Leanne K Küpers; Xiaojing Xu; Soesma A Jankipersadsing; Ahmad Vaez; Sacha la Bastide-van Gemert; Salome Scholtens; Ilja M Nolte; Rebecca C Richmond; Caroline L Relton; Janine F Felix; Liesbeth Duijts; Joyce B van Meurs; Henning Tiemeier; Vincent W Jaddoe; Xiaoling Wang; Eva Corpeleijn; Harold Snieder Journal: Int J Epidemiol Date: 2015-04-10 Impact factor: 7.196
Authors: Nicolas F Schlecht; Miriam Ben-Dayan; Nicole Anayannis; Roberto A Lleras; Carlos Thomas; Yanhua Wang; Richard V Smith; Robert D Burk; Thomas M Harris; Geoffrey Childs; Thomas J Ow; Michael B Prystowsky; Thomas J Belbin Journal: Cancer Med Date: 2015-01-26 Impact factor: 4.452
Authors: Brock C Christensen; E Andres Houseman; Carmen J Marsit; Shichun Zheng; Margaret R Wrensch; Joseph L Wiemels; Heather H Nelson; Margaret R Karagas; James F Padbury; Raphael Bueno; David J Sugarbaker; Ru-Fang Yeh; John K Wiencke; Karl T Kelsey Journal: PLoS Genet Date: 2009-08-14 Impact factor: 5.917
Authors: Loukia G Tsaprouni; Tsun-Po Yang; Jordana Bell; Katherine J Dick; Stavroula Kanoni; James Nisbet; Ana Viñuela; Elin Grundberg; Christopher P Nelson; Eshwar Meduri; Alfonso Buil; Francois Cambien; Christian Hengstenberg; Jeanette Erdmann; Heribert Schunkert; Alison H Goodall; Willem H Ouwehand; Emmanouil Dermitzakis; Tim D Spector; Nilesh J Samani; Panos Deloukas Journal: Epigenetics Date: 2014-10 Impact factor: 4.528