A comparison of five methods to predict genomic breeding values of dairy bulls from genome-wide SNP markers

Table 1 Cross-validation results for method fixed regression-least squares at different threshold values

Trait	α ^†	nSNP		MSEP		r_EBV,MBV		b_EBV,MBV
ASI	0.1	197.2	(31.6)	1464	(139.2)	0.52	(0.031)	0.49	(0.059)
	0.01	98.2	(7.1)	1235	(62.1)	0.54	(0.043)	0.58	(0.045)
	0.001	33.0	(5.4)	1090	(124.4)	0.53	(0.036)	0.71	(0.048)
	0.0001	15.0	(1.9)	1108	(136.8)	0.50	(0.043)	0.76	(0.084)
PPT	0.1	215.6	(29.3)	0.0214	(0.0023)	0.35	(0.056)	0.32	(0.059)
	0.01	81.6	(5.0)	0.0156	(0.0016)	0.42	(0.059)	0.48	(0.075)
	0.001	30.0	(4.2)	0.0135	(0.0023)	0.43	(0.089)	0.62	(0.155)
	0.0001	15.4	(2.1)	0.0136	(0.0016)	0.39	(0.076)	0.67	(0.173)

Average number of SNP (nSNP) in the model, mean square error (MSEP), correlation (r_EBV,MBV) between EBV and MBV, and regression coefficient (b_EBV,MBV) of EBV on MBV for different threshold levels (α) in five cross-validation samples of the training data set, standard error in parentheses; ASI: Australian Selection Index; PPT: protein percentage; ^† P-value used to select SNPs in or out of model.

ISSN: 1297-9686