Table 3. Free amino acid traits and pathway combinations for which the MultiBLUP model explained a significant proportion of heritability, improved model fit relative to random gene groups of approximately the same size, and increased predictive ability compared to GBLUP.

			Prop. h2 explained			Likelihood ratio			Predictive ability
Category	Pathway	Trait	Prop. h2	P-value	FDR P-value	LR	P-value	FDR P-value	Δr	P-value	FDR P-value	Δr2/h2	Δslope	ΔRMSE
amino acid	aa_degradation	his	0.202	0.001	0.020	13.000	<0.001	<0.001	0.039	<0.001	<0.001	0.026	−0.022	−1.46E-02
	aa_degradation	his_t	0.390	<0.001	<0.001	9.907	0.001	0.020	0.076	<0.001	<0.001	0.058	−0.036	−7.71E-01
	aa_degradation	ile_t	0.412	0.003	0.060	5.689	0.007	0.100	0.039	0.006	0.037	0.042	−0.045	−1.60E-04
	aa_degradation	met_t	0.268	0.003	0.060	6.382	0.004	0.080	0.027	0.003	0.012	0.015	−0.079	−1.57E-04
	aa_degradation	val_BCAA	0.218	0.002	0.040	8.024	0.004	0.080	0.042	0.001	0.006	0.027	0.003	−4.67E-04
	aa_degradation	his_GluFam	0.468	0.002	0.040	11.780	<0.001	<0.001	0.097	<0.001	<0.001	0.076	−0.081	−4.59E-01
	aa_synthesis	ile	0.529	0.003	0.060	11.790	<0.001	<0.001	0.070	<0.001	<0.001	0.073	−0.101	−1.73E-03
	aa_synthesis	leu	0.339	0.004	0.080	9.529	<0.001	<0.001	0.053	<0.001	<0.001	0.050	−0.040	−2.86E-03
	aa_synthesis	val	0.336	0.003	0.030	4.507	0.007	0.067	0.017	0.006	0.064	0.016	−0.036	−8.14E-04
	aa_synthesis	BCAA	0.439	0.002	0.040	8.823	0.001	0.020	0.044	<0.001	<0.001	0.050	−0.064	−2.16E-03
specialized	isoprenoids	ile_t	0.429	0.008	0.080	4.575	0.010	0.100	0.044	0.001	0.005	0.032	−0.078	−2.04E-04
	phenylpropanoids	tyr_ShikFam	0.332	<0.001	<0.001	8.811	<0.001	<0.001	0.069	<0.001	0.001	0.035	−0.301	−7.71E-04
	S_containing	ShikFam	0.301	0.001	0.020	10.810	0.001	0.020	0.032	<0.001	0.001	0.036	−0.024	−2.39E-04
protein	protein_aa_activation	gly_SerFam	0.700	0.003	0.060	13.810	<0.001	<0.001	0.075	<0.001	<0.001	0.147	−0.076	−1.89E-03
	protein_folding	his	0.146	0.009	0.090	11.520	0.002	0.020	0.021	0.011	0.055	0.013	−0.009	−8.36E-03
	protein_folding	SerFam	0.303	0.005	0.100	6.487	0.001	0.020	0.032	0.001	0.008	0.042	−0.033	−1.11E-04
	protein_synthesis	total	0.535	0.002	0.040	6.418	<0.001	<0.001	0.024	0.001	0.006	0.018	−0.025	−2.00E-04
	protein_synthesis	PyrFam	0.809	0.005	0.100	5.298	0.003	0.060	0.026	0.003	0.029	0.032	−0.030	−1.50E-03

Bolded rows indicate trait and pathway combinations that increased predictive ability by more than 5% compared to a GBLUP model. The difference in slope between the MultiBLUP and GBLUP models was computed as $\left|slop{e}_{MultiBLUP}-1\right|-\left|slop{e}_{GBLUP}-1\right|$. RMSE, root mean squared error.