Table 3.

The predictive ability of four basic methods and ELPGV, and the comparison p-value between ELPGV and others in T1D, T2D, BD, RA, CAD, HT with WTCCC dataset

Method	T1D		T2D		BD		RA		CAD		HT
	Predictive ability	p-value	Predictive ability	p-value	Predictive ability	p-value	Predictive ability	p-value	Predictive ability	p-value	Predictive ability	p-value
ELPGV	0.8879 ± 0.0008	—	0.8471 ± 0.0009	—	0.9195 ± 0.0006	—	0.8843 ± 0.0008	—	0.8705 ± 0.0008	—	0.9132 ± 0.0006	—
BayesA	0.8664 ± 0.0009	9.037E−69	0.8022 ± 0.0014	9.330E−72	0.8984 ± 0.0007	5.191E−73	0.8628 ± 0.0010	1.454E−68	0.8459 ± 0.0010	6.011E−61	0.8879 ± 0.0007	2.525E−75
BayesB	0.8841 ± 0.0008	4.330E−27	0.8396 ± 0.0010	1.517E−33	0.9167 ± 0.0005	4.416E−32	0.8810 ± 0.0008	1.425E−27	0.8651 ± 0.0008	2.272E−28	0.9100 ± 0.0005	9.319E−29
BayesCπ	0.8863 ± 0.0008	1.827E−23	0.8447 ± 0.0010	6.458E−31	0.9168 ± 0.0006	9.400E−39	0.8827 ± 0.0008	9.640E−20	0.8688 ± 0.0008	1.720E−24	0.9111 ± 0.0006	9.426E−32
GBLUP	0.4957 ± 0.0030	1.597E−114	0.4390 ± 0.0032	1.090E−112	0.6311 ± 0.0023	2.207E−112	0.5337 ± 0.0025	4.853E−118	0.5459 ± 0.0028	7.565E−109	0.5596 ± 0.0028	3.421E−110

ELPGV is the ensemble learning based on BayesA, BayesB, BayesCπ and GBLUP

— Represents no explicit result was found in this method

type 1 diabetes (T1D), type 2 diabetes (T2D), bipolar disorder (BD), rheumatoid arthritis (RA), coronary artery disease (CAD) and hypertension (HT)