. 2020 Apr 7;11:282. doi: 10.3389/fgene.2020.00282

TABLE 1.

Seven prediction models configured by parameters for starting values of variance components in the parameter file and the main results of GBLUP and GREML for each model.

	var_snp_a	var_snp_d	var_snp_e	var_hap_a
Model 1	var_snp_a NPV	var_snp_d NPV	var_snp_e NPV	var_hap_a NPV
	$\hat{a}$ , $R_{a}^{2}$ , ${\hat{σ}}_{α}^{2}$ , ${\hat{h}}_{α}^{2}$	$\hat{d}$ , $R_{d}^{2}$ , ${\hat{σ}}_{δ}^{2}$ , ${\hat{h}}_{δ}^{2}$	${\hat{σ}}_{e}^{2}$	${\hat{a}}_{h}$ , $R_{ah}^{2}$ , ${\hat{σ}}_{α h}^{2}$ , ${\hat{h}}_{α h}^{2}$ , $\hat{g}$ , $R_{g}^{2}$
Model 2	var_snp_a NPV	#var_snp_d	var_snp_e NPV	var_hap_a NPV
	$\hat{a}$ , $R_{a}^{2}$ , ${\hat{σ}}_{α}^{2}$ , ${\hat{h}}_{α}^{2}$		${\hat{σ}}_{e}^{2}$	${\hat{a}}_{h}$ , $R_{ah}^{2}$ , ${\hat{σ}}_{α h}^{2}$ , ${\hat{h}}_{α h}^{2}$ , $\hat{g}$ , $R_{g}^{2}$
Model 3	#var_snp_a	var_snp_d NPV	var_snp_e NPV	var_hap_a NPV
		$\hat{d}$ , $R_{d}^{2}$ , ${\hat{σ}}_{δ}^{2}$ , ${\hat{h}}_{δ}^{2}$	${\hat{σ}}_{e}^{2}$	${\hat{a}}_{h}$ , $R_{ah}^{2}$ , ${\hat{σ}}_{α h}^{2}$ , ${\hat{h}}_{α h}^{2}$ , $\hat{g}$ , $R_{g}^{2}$
Model 4	#var_snp_a	#var_snp_d	var_snp_e NPV	var_hap_a NPV
			${\hat{σ}}_{e}^{2}$	${\hat{a}}_{h}$ , $R_{ah}^{2}$ , ${\hat{σ}}_{α h}^{2}$ , ${\hat{h}}_{α h}^{2}$
Model 5	var_snp_a NPV	var_snp_d NPV	var_snp_e NPV	#var_hap_a
	$\hat{a}$ , $R_{a}^{2}$ , ${\hat{σ}}_{α}^{2}$ , ${\hat{h}}_{α}^{2}$	$\hat{d}$ , $R_{d}^{2}$ , ${\hat{σ}}_{δ}^{2}$ , ${\hat{h}}_{δ}^{2}$ , $\hat{g}$ , $R_{g}^{2}$	${\hat{σ}}_{e}^{2}$
Model 6	var_snp_a NPV	#var_snp_d	var_snp_e NPV	#var_hap_a
	$\hat{a}$ , $R_{a}^{2}$ , ${\hat{σ}}_{α}^{2}$ , ${\hat{h}}_{α}^{2}$		${\hat{σ}}_{e}^{2}$
Model 7	#var_snp_a	var_snp_d NPV	var_snp_e NPV	#var_hap_a
		$\hat{d}$ , $R_{d}^{2}$ , ${\hat{σ}}_{δ}^{2}$ , ${\hat{h}}_{δ}^{2}$	${\hat{σ}}_{e}^{2}$

A non-zero positive value (NPV) after the parameter activates the genetic value in the prediction model, and a ‘#’ sign in front of the parameter removes the genetic value from the prediction model. $\hat{a}$ = GBLUP of SNP additive value. $R_{a}^{2}$ = reliability of $\hat{a}$ . $\hat{d}$ = GBLUP of SNP dominance value. $R_{d}^{2}$ = reliability of $\hat{d}$ . $\hat{g}$ = GBLUP of total genotypic value, = $\hat{a}$ + $\hat{d}$ + ${\hat{a}}_{h}$ for Model 1, = $\hat{a}$ + ${\hat{a}}_{h}$ for Model 2, = $\hat{d}$ + ${\hat{a}}_{h}$ for Model 3, = $\hat{a}$ + $\hat{d}$ for Model 5. $R_{g}^{2}$ = reliability of $\hat{g}$ . ${\hat{σ}}_{α}^{2}$ = GREML estimate of SNP additive variance. ${\hat{σ}}_{δ}^{2}$ = GREML estimate of SNP dominance variance. ${\hat{σ}}_{e}^{2}$ = GREML estimate of residual variance. ${\hat{σ}}_{α h}^{2}$ = GREML estimate of haplotype additive variance. ${\hat{h}}_{α}^{2}$ = GREML estimate of SNP additive heritability. ${\hat{h}}_{δ}^{2}$ = GREML estimate of SNP dominance heritability. ${\hat{h}}_{α h}^{2}$ = GREML estimate of haplotype additive heritability.