Fast empirical Bayesian LASSO for multiple quantitative trait locus mapping

BMC Bioinformatics

Table 3 Summary of results for the simulated data with only main effects

Algorithm	Parameters^◇	PE ± STE*	Number of effects^†‡		CPU time (sec)
	(0.001, 0.001)	11.52 ± 0.5677	14/0	11.1	1.2
	(0.01, 0.01)	11.52 ± 0.578	16/0	10.53	1.3
	(0.05, 0.05)	11.36 ± 0.6088	17/0	10.32	1.1
	(0.1, 0.1)	11.23 ± 0.5571	17/0	10.32	1.1
	(0.5, 0.5)	11.32 ± 0.5937	17/0	10.34	1.1
EBLASSO	(1, 1)	11.4 ± 0.5929	16/0	10.64	1.1
	(0.5, 0.1)	11.57 ± 0.5593	15/0	10.83	1.3
	(-0.5, 0.1)	10.87 ± 0.5599	17/0	10.31	1.6
	(-0.75, 0.1)	10.78 ± 0.5646	20/5	9.52	1.5
	(-0.95, 0.1)	11.09 ± 0.5045	22/20	8.71	1.4
	(-1, 0.0001)	17.73 ± 2.0244	9/0	16.07	1491.9
	(-1, 0.0005)	15.81 ± 2.5732	16/0	11.66	1676.0
EB	(-1, 0.001)	12.21 ± 1.7635	17/2	10.65	1657.9
	(-1, 0.01)	10.69 ± 0.9903	19/4	9.05	1954.9
	(-1, 0.1)	11.63 ± 0.5743	20/20	7.29	2222.7
RVM	-	-	20/42	7.81	268.7
	0.1347	10.77 ± 0.4583	16/27	9.47	0.7
	0.0850	10.52 ± 0.4442	20/49	8.89	0.7
LASSO	0.0675	10.50 ± 0.5248	20/48	8.63	0.7
	0.0536	10.52 ± 0.4382	19/35	8.28	0.7
	0.0338	10.59 ± 0.4434	17/2	7.35	0.7

^◇ Parameters are (a, b) for the EBLASSO, (τ, ω) for the EB and λ for the LASSO.
*The average PE and the standard error were obtained from ten-fold cross validation.
^†The number of detected effects and residual variance were obtained using all 1000 samples not from cross validation.
^‡The first number is the number of true positive effects; the second number is number of false positive effects. All the effects counted have a p-value ≤ 0.05.

ISSN: 1471-2105