Skip to main content

Table 1 Power and Type I error rates for gene by methylation interactions

From: Longitudinal data methods for evaluating genome-by-epigenome interactions in families

Model Outcome Nominal (p < 0.05) Genome-wide (p < 1.08 × 10− 7)
1 (0.125) 8 (0.10) 6 (0.075) 17 (0.05) 10 (0.025) Type I error 1 (0.125) 8 (0.10) 6 (0.075) 17 (0.05) 10 (0.025) Type I error
LME Pre/Post 0.930 0.790 0.690 0.545 0.290 0.048 0.016 0.020 0.000 0.000 0.000 0.000
Change 0.839 0.715 0.595 0.510 0.210 0.038 0.011 0.005 0.000 0.000 0.000 0.000
GEE Pre/Post 0.935 0.835 0.705 0.555 0.290 0.073 0.022 0.025 0.005 0.000 0.000 0.002
Change 0.855 0.735 0.565 0.535 0.240 0.069 0.016 0.005 0.000 0.000 0.000 0.001
GEE–BC Pre/Post 0.914 0.835 0.630 0.530 0.265 0.051 0.011 0.015 0.000 0.000 0.000 0.001
Change 0.828 0.715 0.540 0.500 0.210 0.052 0.011 0.005 0.000 0.000 0.000 0.001
QIF Pre/Post 0.941 0.845 0.685 0.580 0.410 0.120 0.054 0.055 0.015 0.000 0.005 0.002
Change 0.860 0.735 0.575 0.580 0.345 0.111 0.048 0.015 0.000 0.000 0.000 0.002
QIF–BC Pre/Post 0.882 0.760 0.520 0.460 0.205 0.043 0.005 0.010 0.000 0.000 0.000 0.000
Change 0.780 0.670 0.450 0.450 0.180 0.042 0.005 0.005 0.000 0.000 0.000 0.000
  1. BC, Bias-corrected methods; Change, post-pre change triglyceride score as the outcome; GEE, generalized estimating equations; LME, linear mixed effects; Pre/Post, posttreatment triglyceride as the outcome and pretreatment as a baseline covariate; QIF, quadratic inference functions
  2. The proportion of 200 (or 186 in the case of the chromosome 1 site) replicates that each SNP × CpG interaction was identified as significant for each significance threshold. Presented is the chromosome location with simulated expected heritability (hg2) in parentheses
  3. Causal chromosome sites simulated include: Chr 1 (SNP: rs9661059; CpG: cg00000363); Chr 8 (SNP: rs1012116; CpG: cg18772399); Chr 6 (SNP: rs736004; CpG: cg10480950); Chr 17 (SNP: rs4399565; CpG: cg01242676); and Chr 10 (SNP: rs10828412; CpG: cg00045910)
\