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

Table 2 Type I error rates of GxEsum when using binary disease traits with various population prevalence

Scenarios	Population prevalence (k)	Type I error rate
Var(GxE) = 0, Var(RxE) = 0	0.025	0.052
	0.05	0.042
	0.1	0.076
	0.5	0.054
Var(GxE) = 0, Var(RxE) = 0.1 (on the liability scale)	0.025	0.044
	0.05	0.036
	0.1	0.050
	0.5	0.052
Average		0.050

We simulated quantitative phenotypic data based on a real genotypic dataset (ARIC GWAS) including 7263 individuals with 583,085 SNPs. The phenotypes were standardised such that the mean was 0 and variance was 1, for which we applied the liability threshold model to generate affected or unaffected disease status for each individual, using various values for the population prevalence (k = 0.025, 0.05, 0.1 or 0.5). Type I error rate at a significance threshold of p-value < 0.05 was estimated from 500 replicates for each scenario and population prevalence

ISSN: 1474-760X