Genomic information has the potential to increase the accuracy of estimating breeding values, but requires large reference populations containing animals with both genotype and phenotype information. In numerical small breeds or populations, establishing such large reference populations is challenging. Therefore, the potential to use information from one population to estimate breeding values for individuals from another population was investigated in a recently published paper.
Due to differences in the genetic background across populations, differences in the effect of a gene on a specific trait might exist. Those differences might hinder the use of information across the populations. The correlation between the effects of all genes across populations can be considered as the genetic correlation. The results of this study show that there is a linear relationship between the genetic correlation and the possibility to use information across populations; the higher the genetic correlation, the higher the potential to use information across the populations. As long as the genetic correlation is different from zero, useful information can be shared across the populations. Moreover, the results show that the genetic correlation can be accurately estimated using a multi-trait model including the same trait in the different populations as different traits.
In the paper, a formula was derived to enable the estimation of the prediction accuracy including one or more populations in the reference population. The genetic correlation across the populations is an important parameter in this formula, but also genomic relationships and trait parameters are included. With this formula, the potential accuracy can accurately be estimated before phenotypic data of the selection candidates is available.