Combining host and rumen metagenome profiling for selection in sheep: prediction of methane, feed efficiency, production, and health traits

Table 6 Variance components and prediction accuracies for residual feed intake on lucerne pellets using genotypes and metagenomes

Metagenome profile	Model^a	Component^b	Proportion of variance^c	Accuracy^d	Regression coefficient^e
Lucerne lamb	G	G_G	0.39 ± 0.07	0.29 ± 0.08	1.06 ± 0.22
	M	M_M	0.77 ± 0.07	0.47 ± 0.06	1.15 ± 0.52
	G + M	G_G+M	0.30 ± 0.07	0.19 ± 0.13^f	1.15 ± 0.63^f
		M_G+M	0.67 ± 0.07	0.46 ± 0.07^f	1.20 ± 0.61^f
		G + M_G+M	0.98 ± 0.08	0.49 ± 0.06^f	1.14 ± 0.52^f
Grass lamb	G	G_G	0.43 ± 0.08	0.29 ± 0.09	1.25 ± 0.48
	M	M_M	0.74 ± 0.21	0.19 ± 0.05^f	0.90 ± 0.44^f
	G + M	G_G+M	0.40 ± 0.08	0.28 ± 0.08^f	1.28 ± 0.45^f
		M_G+M	0.57 ± 0.19	0.18 ± 0.06^f	1.04 ± 0.56^f
		G + M_G+M	0.97 ± 0.19	0.31 ± 0.07^f	1.05 ± 0.35^f
Grass adult	G	G_G	0.39 ± 0.07	0.28 ± 0.10	1.20 ± 0.50
	M	M_M	0.69 ± 0.20	0.21 ± 0.16	1.08 ± 0.93
	G + M	G_G+M	0.37 ± 0.07	0.27 ± 0.09^f	1.27 ± 0.54^f
		M_G+M	0.48 ± 0.18	0.19 ± 0.16^f	1.32 ± 1.14^f
		G + M_G+M	0.85 ± 0.19	0.31 ± 0.12^f	1.16 ± 0.62^f

^aModel fitting the random effects of animal genetics (G), metagenome profile (M) or both (G + M)
^bLarge letter(s) represent the component (genomics = G; metagenome = M; or sum of the two = G + M), and the subscript represents the model used
^cProportion of the phenotypic variance (adjusted for fixed effects) explained by animal genetics (G, heritability), the metagenome profile (M, microbiability) or both (G + M)
^dCorrelation between phenotype adjusted for fixed effects and either breeding value (G), metagenome value (M) or the sum of both (G + M) when using cross-fold validation
^eRegression coefficient of phenotype adjusted for fixed effects on either breeding value (G), metagenome value (M) or the sum of both (G + M) when using cross-fold validation with each Cohort a different fold
^fAt least one of the models contributing to this value had convergence or singularity issues

ISSN: 1297-9686