Skip to main content

Table 1 Type I error rates of GxEsum to detect GxE at a significance threshold of p-value <‚ÄČ0.05

From: GxEsum: a novel approach to estimate the phenotypic variance explained by genome-wide GxE interaction based on GWAS summary statistics for biobank-scale data

Scenarios Type l error rate
Var(GxE) = 0, var(RxE) = 0 0.066
Var(GxE) = 0, var(RxE) = 0, G-E correlation = 0.1 0.064
Var(GxE) = 0, var(RxE) = 0, R-E correlation = 0.1 0.044
Var(GxE) = 0, var(RxE) = 0, G-E correlation = 0.1, R-E correlation = 0.1 0.056
Var(GxE) = 0, var(RxE) = 0.1 0.044
Var(GxE) = 0, var(RxE) = 0.1, G-E correlation = 0.1 0.034
Var(GxE) = 0, var(RxE) = 0.1, R-E correlation = 0.1 0.028
Var(GxE) = 0, var(RxE) = 0.1, G-E correlation = 0.1, R-E correlation = 0.1 0.054
Average 0.049
  1. We simulated phenotypic data based on a real genotypic dataset (ARIC GWAS) including 7263 participants with 583,085 SNPs, using various scenarios. The phenotypes were standardized such that the phenotypic mean was 0 and the phenotypic variance was 1. Type I error rate (i.e., false-positive) was estimated from 500 replicates for each scenario
  2. GxE genotype-by-environment interaction, RxE residual-by-environment interaction, G-E correlation genotype-environment correlation, R-E correlation residual-environment correlation