GenoMiX: Utilizing crossbred information to accelerate genetic progress
Crossbred performance is the ultimate breeding goal of many livestock breeding programs. At present, selection is however mainly based on performance measured on purebred animals, in highly controlled environments. For many traits, the genetic correlation between purebred and crossbred performance is considerably lower than unity. This indicates that selection based on purebred performance ignores part of the genetic variation, leading to suboptimal selection responses in crossbred performance.
The overall objective of this project is to improve accuracy of prediction of phenotypes of crossbred animals by utilizing the total genetic variance, including additive, dominance and epistatic effects. Few methods thus far have been developed that can simultaneously estimate these effects, and use these for prediction of phenotypes. Likewise, little is known of optimal design of breeding programs using genomic selection that uses both purebred and crossbred performance records of multiple lines and crosses. The project focusses on developing parametric and non-parametric models that efficiently estimate additive and non-additive effects, required to improve accuracy of prediction of crossbred performance. For genomic prediction across multiple lines, we will partition relationships into a component due to close pedigree relationships within line, and relationships due to linkage disequilibrium over short distances, which also acts between lines. Parametric models will include additive and dominance effects, while non-parametric models will also include epistasis. To allow inclusion of epistasis, we will extend Gaussian Process Regression (GPR) methods to predict phenotypes for the future crossbred offspring of purebred breeding animals. The power of GPR to quantify non-additive effects has recently been demonstrated in yeast, where nearly the full broad-sense heritability was accounted for and phenotypes were predicted with significantly lower mean-squared error than a linear mode l. Connected to the research on prediction models, optimal breeding program designs for accurate estimation of additive and non-additive effects will be derived. This combination of development of prediction tools and optimization of genomic breeding programs has the potential to considerably strengthen crossbreeding programs of the industry.
Crossbreeding is abundant in the livestock industry and relies on within-line (or breed) genetic selection of purebred breeding animals that are mated between lines to produce crossbred production animals. Direct selection based on crossbred performance was thus far mostly prohibited because linking performance of crossbred animals to purebred selection candidates was not possible in many cases due to lack of pedigree information. Now, this link can be established using genomics as the sole carrier of information. However, methods and tools to efficiently establish this link are still very premature.
In this project, efficient methods for genomic prediction of crossbred performance will be developed and implemented in software tools that will be made available to the Breed4Food breeding companies. Another important question for the breeding industry, that will be addressed in this project, is how budgets should be spread across genotyping and phenotyping efforts of purebred and crossbred animals, to optimize investment strategies. Results found in this research can be implemented immediately because the development and testing of the methods will make use of data on the commercial breeding populations of the Breed4Food breeding companies. Through the evaluation of breeding program designs, and the accuracy of the newly developed genomic prediction models, this project will generate important knowledge to support selection strategies for the next 10-20 years.“
Breed4Food STW Partnership
The Breed4Food STW Partnership “Predicting Phenotypes” research projects are:
- From sequence to phenotype: detecting deleterious variation by prediction of functionality
- GenoMIX: utilizing crossbred information to accelarete genetic progress
- SmartBreed: Smart animal breeding with advanced machine learning
- Topbreed; towards precision breeding using genomic prediction
Mutual goal is a more sustainable and animal-friendly breeding.