In this study, the authors examined the following 5 formulae for accuracy of prediction: 0.11×BH+10.53 (cm) from Brodsky et al1; 0.15×BH+3.96 (cm) from Bahk and Oh2; 0.148×BH+3.8 (cm) from Chow et al;3 0.1×BH+12.5 (cm) from Takita et al4; and 0.1977×BH – 4.2423 (cm) (authors' formula).
In fact the deterministic formula for GEBV accuracy predicts quite well the GEBV accuracies we achieve in the Jersey population given the Ne, number of records used and heritability [ 14, 22].
This can also be seen from the formula for the accuracy of GBLUP, as shown by [ 17] and [ 18], which depends on the number of independent segments in the genome, but not on the actual number of QTL.
First, the coalescent model enables the derivation of analytical formulas for imputation accuracy.
Goddard [ 6] and Daetwyler et al. [ 15] proposed predictive formulae for the accuracy of genomic predictions.
The CKD EPI formula was not tested for accuracy in Thai population and thus may be used for relative measurements of GFR and comparisons between vitamin D deficiency and non-vitamin D deficiency.
More details on our simulations for this formula can be found in B.1 Additional graphs for accuracy of level and for power.
This indicates that formulas for estimating the accuracy of genomic prediction using multiple populations need further investigation to define the best way to calculate genomic relationships across populations.
Finally, the empirical accuracies and estimated accuracies (averaged across animals in the testing set) were compared with the analytical expectation for accuracy of genomic predictions, calculated according to a formula proposed by [ 32] (i.e. Equation 1 in that study).
