An efficient unified model for genome-wide association studies and genomic selection

Table 2 Comparison of genomic prediction accuracy between the stepwise linear mixed model (StepLMM) and genomic best linear unbiased prediction (GBLUP) based on WTCCC simulated data

\(\varvec{h}^{2}\)	\(\widehat{\varvec{h}}^{2}\)	\(\widehat{\varvec{h}}_{\varvec{q}}^{2} /\widehat{\varvec{h}}^{{2^{\text{a}} }}\)	\(r_{qtl}^{\text{b}}\)	\(r_{poly}^{\text{c}}\)	\(r_{StepLMM}^{\text{d}}\)	\(r_{GBLUP}^{\text{e}}\)	\(b_{StepLMM}^{\text{f}}\)	\(b_{GBLUP}^{\text{g}}\)
0.25	0.25 (0.009)	0.68 (0.026)	0.71 (0.010)	0.27 (0.009)	0.75 (0.007)	0.49 (0.008)	0.89 (0.011)	1.16 (0.055)
0.50	0.52 (0.007)	0.81 (0.013)	0.87 (0.004)	0.23 (0.006)	0.90 (0.002)	0.71 (0.002)	0.94 (0.005)	1.04 (0.042)
0.75	0.76 (0.004)	0.92 (0.007)	0.95 (0.002)	0.18 (0.007)	0.96 (0.001)	0.86 (0.001)	0.98 (0.002)	1.01 (0.022)

^aProportion of phenotypic variance explained by the QTL in the models, \(h_{q}^{2} = \frac{{\sigma_{q}^{2} }}{{\sigma_{q}^{2} + \sigma_{g}^{2} + \sigma_{e}^{2} }}\)
^bCorrelation between true breeding values and genetic values explained by QTL detected with StepLMM
^cCorrelation between true breeding values and genetic values excluding QTL detected with StepLMM
^dCorrelation between true breeding values and genomic breeding values estimated with StepLMM
^eCorrelation between true breeding values and genomic breeding values estimated with GBLUP
^fRegression coefficient of the true on the estimated breeding values with StepLMM
^gRegression coefficient of the true on the estimated breeding values with GBLUP

ISSN: 1297-9686