Control of false discoveries in grouped hypothesis testing for eQTL data

BMC Bioinformatics

Table 1 Summary of the simulation studies with directly simulated z from an AR(1) model with correlation \(\rho \)

True \(\pi _0\)	True \(\sigma \)	True \(\rho \)	Mean \(\hat{\pi _0}\)	Mean \({\hat{\sigma }}\)	SE(\(\hat{\pi _0}\))	SE(\({\hat{\sigma }}\))	Realized	Realized
							FDR (5%)	FDR (10%)
0.20	1	0.10	0.2030	0.9964	0.1841	0.0823	0.0954	0.1236
0.20	2	0.10	0.1865	1.9660	0.0469	0.0374	0.0576	0.1136
0.20	5	0.10	0.1977	4.9383	0.0094	0.0306	0.0507	0.1014
0.20	1	0.50	0.1932	0.9919	0.1613	0.0757	0.0922	0.1252
0.20	2	0.50	0.1873	1.9663	0.0417	0.0352	0.0565	0.1121
0.20	5	0.50	0.1977	4.9383	0.0092	0.0303	0.0508	0.1013
0.20	1	0.80	0.1857	0.9875	0.1308	0.0664	0.0882	0.1245
0.20	2	0.80	0.1894	1.9673	0.0325	0.0317	0.0545	0.1090
0.20	5	0.80	0.1979	4.9388	0.0085	0.0292	0.0507	0.1012

The last two columns show the FDR control performance of the Z-REG-FDR method when the target FDR is 5% and 10%, respectively

ISSN: 1471-2105