Mixed logistic regression in genome-wide association studies

BMC Bioinformatics

Table 2 Comparison of powers in simulations based on the coalescent model (5000 individuals), with \(\tau = 1\)

Scenario	\(p_0\)	\(p_1\)	OR	MAF bin	LR	AMLE	Offset
Large cohort effect	0.05	0.30	1.5	(0.20; 0.25]	0.307	0.251	0.202
Large cohort effect	0.05	0.30	1.5	(0.45; 0.50]	0.597	0.544	0.464
No cohort effect	0.30	0.30	1.5	(0.20; 0.25]	0.871	0.859	0.834
No cohort effect	0.30	0.30	1.5	(0.45; 0.50]	0.996	0.989	0.987

\(p_0\) and \(p_1\) are the prevalences in the two strata. Powers at the genome-wide significance level \(5\times 10^{-8}\) are reported for logistic regression (LR), AMLE (equivalent to the score test) and Offset methods, all analyses including 10 PCs with fixed effect

ISSN: 1471-2105