ESTIMATION OF SNP EFFECTS IN LACAUNE DAIRY SHEEP DEPENDING ON THE REFERENCE POPULATION COMPOSITION
Résumé
The Lacaune dairy sheep breed split in 1972 into two subpopulations with no exchange of genetic material but a single genetic evaluation and same selection objectives. Previous work has shown that this led to the creation of two disconnected but genetically close subpopulations. Previous work also demonstrated that the currently performed combined genomic evaluation of both subpopulations is slightly advantageous, in terms of accuracy, as opposed to within-subpopulation genomic evaluations. This paper focuses on the study of the estimated SNPs effects related to the three training populations: composed of one, the other or both subpopulations. The estimated SNP effects are strongly correlated across years within the training population. When subpopulations are predicted separately, there is low correlation between estimated SNP effects, but when they are predicted jointly, there is a strong correlation of the joint estimate with subpopulation estimates. The regression of "early" (only based on genomic information) on "late" (including progeny information) SNP predictions is lower than one for one of the subpopulations but not for the other, and close to one for the joint prediction. This shows some bias in this particular subpopulation whose origin is not understood.
Domaines
Génétique animaleOrigine | Fichiers produits par l'(les) auteur(s) |
---|