Multi-breed Genomics of Beef and Dairy Cattle – GEMBAL

Adapting selection tools and objectives to efficiently manage French cattle meat and milk productions is a major challenge of the next years. Genomic selection provides a fantastic opportunity to reorient bovine selection towards a more sustainable breeding. The Gembal project aims at developing a multi-breed genomic selection to extend its use to all beef and dairy breeds, including the small ones. Special attention will be paid for functional traits and maternal traits: calving ease, fertility and longevity of cows in both beef and dairy breeds. At national level, this project should be a common foundation for all breeding schemes, thus avoiding a multiplication of too small and inefficient initiatives.

The core of the project is the making-up of the technical basis for the development of multi-breed genomic selection in beef and dairy cattle. The basic idea is that a sample - so-called imputation population - will be genotyped with a high density chip in each breed, whereas most other individuals will be genotyped at a lower cost for a medium density chip. The condition required to build the imputation populations is an extensive use of a new molecular tool, a high density chip with 800,000 SNP developed by Illumina with a consortium including INRA and UNCEIA. Task 2 is dedicated to this technical part of the project.

In Task 3, the large multi-breed resource cattle population generated in Task 2 will be the basis for academic researches aimed at characterizing the genetic diversity across breeds and the history of each population submitted to its own context, i.e. drift and selection. This task will also be useful to detect the conserved chromosomal segments across breeds that can be used in multi-breed genomic selection as it will be envisioned in Task 5.

Task 4 corresponds to imputation, i.e the statistical procedure to infer missing genotypes in most individuals from the complete genotype information in a limited imputation sample. We will study the quality of the imputation according to breed effective and imputation sample size. We will also develop more computationally efficient algorithms, as imputation will be very demanding with the fast development of genomic selection.

Then, a genomic prediction model, using linkage disequilibrium information across breeds, will be developed in Task 5. The methodological challenges are the development of powerful and robust statistical approaches as well as and computing tools for the prediction in a multi-breed context, especially for functional traits with correlated direct and maternal genetic effects. The applications regarding functional traits will be carried out in Task 6 and Task 7 for dairy and beef breeds, respectively.

In Task 6, the existence of three breeds in France for which reference populations of reasonable to very large size are available and for which genomic selection programs are already implemented will allow us to undertake reliable comparisons of within vs multi-breed genomic evaluations, hopefully revealing what are the underlying conditions for a successful implementation of multi-breed evaluation. An alternative strategy will consist in checking whether the conserved genome fragments corresponding to favourable haplotypes of QTL detected in any large breed are also segregating in the smaller breeds. Then a genomic evaluation based on these haplotypes could be implemented for the smaller breeds.

In Task 7, the multi-beef breed reference population will be composed of the 2,300 bulls that also constitute the beef imputation populations. If a sufficient number of QTL are commonly detected across beef and dairy breeds, a QTL detection and a computation of prediction equations from the beef and dairy pooled reference populations will be undertaken for maternal functional traits.