genomic prediction genetic architecture

News

Genetic architecture of traits determine success of genomic prediction

Published on
July 6, 2015

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.

Genomic selection

In numerical small breeds or populations, establishing such large reference populations may simply be impossible. Therefore, combining animals from different populations in one reference populations, known as multi breed genomic prediction, is an attractive option to increase reference population sizes and, potentially, the accuracy of breeding value estimation. One of the factors that could influence the accuracy of both single and multi-breed genomic prediction is the genetic architecture of a trait, which is investigated in a recently published paper in Genetics Selection Evolution in collaboration of the Animal Breeding and Genomics Centre with the Department of Environment and Primary Industries in Australia.

Genetic architecture

Our study showed that the accuracy of both single and multi-breed genomic prediction depends on the genetic architecture underlying a trait, such as the number of genes, the allele frequencies of the genes, and the size of the effect of each of the genes. When the average allele frequency of the genes decreases, the accuracy decreases as well. Moreover, the accuracy is lower when genes at low frequency have a large effect. For multi breed genomic prediction, the number of breed specific genes increased when the average allele frequency decreased. This limited the benefit of adding another breed to the reference population. In general, genes underlying a trait are expected to have a low allele frequency, therefore, the results of this study provide a possible explanation why empirical studies show only a limited or no benefit in combining different breeds in one population as opposed to simulation studies where increases in accuracy were much higher.

For more information, please see the full article 'Impact of QTL properties on the accuracy of multi-breed genomic prediction'.

This study is a result from a collaboration between Animal Breeding and Genomics Centre and the Department of Environment and Primary Industries in Australia.