The impact of genotyping strategies and statistical models on accuracy of genomic prediction for survival in pigs

Table 3 Regression coefficient of the EBV from whole data on the EBV from reference data

Trait¹	Genotyping scenario²	a⁴			m⁴			a + m
Trait¹	Genotyping scenario²	LM³	LG	PM	LM	LG	PM	LM	LG	PM
T_4/4	G_all	1.046	1.082	1.051	0.895	0.902	0.924	0.983	1.008	0.974
	G80_ran	1.050	1.086	1.057	0.898	0.906	0.928	0.983	1.011	0.977
	G_alive	1.120	1.132	1.107	0.908	0.909	0.938	1.017	1.026	1.000
	G_none	1.052	1.075	1.047	0.866	0.870	0.898	0.985	1.008	0.966
T_2/4	G_all	1.001	1.081	1.077	1.070	1.057	1.080	1.098	1.082	1.109
	G80_ran	1.007	1.090	1.088	1.071	1.057	1.080	1.099	1.083	1.110
	G_alive	1.037	1.126	1.118	1.079	1.061	1.085	1.117	1.094	1.122
	G_none	1.049	1.133	1.115	1.073	1.077	1.054	1.122	1.122	1.101
T_2/2	G_all	0.947	0.996	0.983	1.000	1.016	1.000	0.960	0.996	0.985
	G80_ran	0.944	0.992	0.984	1.007	1.016	1.006	0.961	0.993	0.988
	G_alive	0.974	1.019	1.009	1.034	1.043	1.028	0.983	1.013	1.004
	G_none	0.991	1.006	0.997	1.003	0.982	0.970	0.989	0.984	0.977

¹T_4/4: trait with h_a² = 0.04, h_m² = 0.04 and lit² = 0.04, T_2/4: h_a² = 0.02, h_m² = 0.04 and lit² = 0.04; T_2/2: h_a² = 0.02, h_m² = 0.02 and lit² = 0.02, in liability scale
²G_all: all pigs were genotyped; G80_ran: 80% of pigs randomly selected from the whole population were genotyped; G_alive: only alive pigs (80%) were genotyped; G_none: no pig was genotyped
³LM linear model, LG logit model, PM probit model
⁴a: direct additive genetic effect; m: maternal additive genetic effect

ISSN: 2049-1891