Predicting crossbred performance using Breed-of-Origin of Alleles

Published on
June 15, 2017

Genetic improvement should be expressed in crossbred performance at the farms of clients. For some traits, where the genetic correlation between purebred performance and crossbred performance is low, some of the realized genetic improvement is lost along the way. That is why Wageningen University & Research Animal Breeding and Genomics is working together with the Breed4Food breeding companies to include more crossbred animals in their genomic evaluations, which until recently, considered only purebred animals.

Tracing back to te purebred line

Current research from Wageningen University & Research Animal Breeding and Genomics in partnership with University of Vi├žosa, Topigs Norsvin, and Topigs Norsvin do Brasil, is producing a model for genomic prediction that integrates crossbred information accounting for breed-specific effects of alleles.

If it is possible to trace back a specific trait to the purebred line, breeding values can be predicted more accurate. Ultimately, this will result in more efficient breeding programs and better health, resilience and welfare of the animals involved.

Breed-of-Origin of Alleles in crossbred animals

To estimate the effects of alleles that originate from different breeds in a crossbred population, the breed-of-origin of alleles (BOA) in crossbred animals must be known. The BOA model traces accurately the breed-of-origin of alleles in three-way crosses. BOA was tested in practical data of a three-way crossbred pig population, where on average 95% of the alleles were assigned a breed-of-origin.

With the BOA model we can estimate genomic breeding values of purebred animals for crossbred performance, based on the effect on crossbred performance of alleles coming from that specific breed. In this way we can predict the merit of animals for breeding.

Further research

The BOA model tends to yield increased genomic prediction accuracy for traits with low genetic correlation between purebred and crossbred performance (<0.30), low heritabilities (<0.20), and when breeds were distantly related. However, the BOA model does not always perform better than standard models. In those situations the standard models may be preferred because it can be more easily implemented than the BOA model. In further research we will investigate if the added value of the BOA model is worth the extra steps.