Table 3.

Summary of the prior distribution of the hyper-parameters and the parametric values utilized for genomic prediction with different Bayesian methods.

	BayesA	BayesB	BayesC	BL	BRR
Prior distribution of the marker effects	Scaled-t with degree of freedom dfβ and scale Sβ	Scaled-t mixture, for the marker with non-zero effects, i.e., proportion π and 1 − π proportion of the total markers are assumed to have null effects	Gaussian mixture, for the marker with non-zero effects, i.e., proportion π and 1 − π proportion of the total markers are assumed to have null effects	Double exponential with parameter λ2	Gaussian with mean μβ and variance ${\sigma }_{\beta }^2$
Prior distribution of hyper parameters	$S_{\beta }~\mathrm{\Gamma }\left(r,s\right)$	$\begin{array}{c}{S}_{\beta }~\mathrm{\Gamma }\left(r,s\right),\hfill \\ \pi ~B\mathrm{eta}\left(p_0,{\pi }_0\right)\hfill \end{array}$		${\lambda }^2~\mathrm{\Gamma }\left(r,s\right)$	${\sigma }_{\beta }^2~{\chi }^{-2}\left(\nu ,S\right)$
Prior distribution of the variance of the marker effects and residual		$\begin{array}{c}{\sigma }_{\beta }^2~{\chi }^{-2}\left(\nu ,S\right),\hfill \\ {\sigma }_e^2~{\chi }^{-2}\left(\nu ,S\right),\hfill \\ whereS~\mathrm{\Gamma }\left(r,s\right)\hfill \end{array}$
Parametric value considered	$\begin{array}{c}s=1.1,\hfill \\ R^2=0.5,\hfill \\ d{f}_{\beta }=5,\nu =5,S_{\beta }=\frac{var\left(y\right)\times R^2\times \left(d{f}_{\beta }+2\right)}{M{S}_x},r=\frac{\left(s-1\right)}{S_{\beta }}\hfill \end{array}$	$\begin{array}{c}s=1.1,R^2=0.5,d{f}_{\beta }=5,\nu =5,\hfill \\ S_{\beta }=\frac{var\left(y\right)\times R^2\times \left(d{f}_{\beta }+2\right)}{\left(\frac{M{S}_x}{\pi }\right)},\hfill \\ r=\frac{\left(s-1\right)}{S_{\beta }},{\pi }_0=0.5,\hfill \\ p_0=10\hfill \end{array}$		$\begin{array}{c}s=1.1,\hfill \\ R^2=0.5,\hfill \\ v=5,\hfill \\ r=\frac{\left(s-1\right)}{\left(\frac{2\left(1-R^2\right)}{R^2\times M{S}_x}\right)}\hfill \end{array}$	$\begin{array}{c}{\mu }_{\beta }=0,\hfill \\ s=1.1,\hfill \\ R^2=0.5,\hfill \\ d{f}_{\beta }=5,\nu =5,\hfill \\ S_{\beta }=\frac{var\left(y\right)\times R^2\times \left(d{f}_{\beta }+2\right)}{M{S}_x},\hfill \\ r=\frac{\left(s-1\right)}{S_{\beta }}\hfill \end{array}$
Parameter for MCMC	All the five Bayesian models were implemented with 10,000 iterations, burn-in period of 1000 cycles and thin of 15 iterations

MS_x Sum of the variances of markers under study, Γ Gamma, χ⁻² Inverse Chi-square, BL Bayesian LASSO, BRR Bayesian ridge regression.