Table 6.

Accuracy and bias of genomic predictions for milk production^a traits using different reference populations and different analysis methods and when validated in Holstein, Jersey or Australian Red animals

Analysis methodb	Reference dataset	Validation dataset	Accuracyc			Bias
			FY	MY	PY	FY	MY	PY
BayesRd	Holstein	Holstein	0.63	0.62	0.58	1.22	0.89	1.02
BayesR_LC	Holstein	Holstein	0.65	0.62	0.57	1.17	0.91	0.99
BayesMV	Holstein	Holstein	0.65	0.63	0.59	1.21	0.89	1.03
BayesRd	Hol_Jer	Holstein	0.65	0.63	0.58	1.25	0.89	0.99
BayesR_LC	Hol_Jer	Holstein	0.65	0.62	0.58	1.14	0.90	0.97
BayesMV	Hol_Jer	Holstein	0.66	0.63	0.58	1.17	0.87	0.97
BayesRd	Jersey	Jersey	0.56	0.70	0.72	0.89	0.98	1.24
BayesR_LC	Jersey	Jersey	0.57	0.70	0.72	0.70	1.05	1.17
BayesMV	Jersey	Jersey	0.55	0.70	0.71	0.81	1.00	1.11
BayesRd	Hol_Jer	Jersey	0.56	0.69	0.71	0.93	0.95	1.18
BayesR_LC	Hol_Jer	Jersey	0.58	0.69	0.73	0.92	1.00	1.20
BayesMV	Hol_Jer	Jersey	0.55	0.66	0.69	0.92	0.96	1.15
BayesRd	Hol_Jer	Aust Red	0.26	0.22	0.10	0.89	0.56	0.38
BayesR_LC	Hol_Jer	Aust Red	0.28	0.20	0.12	0.87	0.53	0.41
BayesMV	Hol_Jer	Aust Red	0.26	0.14	0.07	0.75	0.34	0.25

^aMilk production traits were fat yield (FY), milk yield (MY) and protein yield (PY)

^bMethods were either BayesR on raw phenotypes (BayesR), linear combinations of traits analyzed with univariate BayesR (BayesR_LC) or the multivariate BayesMV method

^cStandard errors are approximately 0.062 for Holstein, 0.098 for Jersey and 0.074 for Australian Red predictions

^dUnivariate results from Kemper et al. [1]