To evaluate the consequences of the varying conditions for the estimation of MA-BLUP breeding values, accuracies were calculated for proven bulls, cows and young bull candidates.
This result supports the simulation results of Toosi et al. (2010), who showed that training in admixed populations results in genomic estimates of breeding value with accuracies almost equivalent to those achieved from training and validating within the same breed.
To further investigate the impact of the genetic difference between breeds on the accuracy of genomic selection based on crossbred data, Figure 2 plots the difference in average genotypic values of the two breeds against the accuracy of breeding values predicted based on their crossbred data.
However, combining data from many breeds into a reference population has not always improved the prediction accuracy of breeding values in the validation sample [ 8– 12] and has led to lower prediction accuracies than would have been expected based on the increase in the size of the training population [ 13].
The accuracy of breeding value predictions from Bayesian genomic selection models decreases and becomes more similar to the accuracy of GBLUP with decreasing heritability, increasing number of QTL and decreasing number of records, [ 4, 20- 23].
The breeding values, accuracy and inbreeding coefficient of the 258,851 Labrador retrievers can be accessed at the World Wide Web as follows: www.vet.cornell.edu/research/bvhip The authors thank Cathy French and Marie Stewart for the website development to host breeding values on hip dysplasia.
The genomic breeding value accuracy increased with increased trait heritability and with increased relationship between training and validation.
It gives higher accuracy of breeding values and does not require a complete restructuring of the breeding program.
Using a genomic relationship matrix increased the accuracy of breeding values relative to using a pedigree relationship matrix.
Using stochastic simulations, assuming a full-sib data structure typical for aquaculture breeding schemes, we compared the two designs with respect to accuracy of breeding values and their ability to produce accurate estimates of (co variance components of direct and social effects.
