|O'Connell, J -|
Submitted to: Journal of Dairy Science
Publication Type: Abstract Only
Publication Acceptance Date: February 24, 2011
Publication Date: June 30, 2011
Citation: Van Raden, P.M., O'Connell, J.R. 2011. Overview of genomic selection in dairy cattle populations. Journal of Animal Science 89(E-Suppl. 1)/Journal of Dairy Science 94(E-Suppl. 1):664(abstr. 597). Technical Abstract: Genomic selection is most successful for traits recorded over many years in large populations. Holstein breeders have reference populations >10,000 proven bulls via cooperation among major countries, and countries with smaller Holstein populations can contribute additional bulls. Scandinavian red dairy cattle breeders have 8,000 reference bulls, and Brown Swiss breeders have a global population of 4,500 reference bulls at Interbull. Jersey breeders have genotyped but have not yet merged their 6,000 reference bulls. Denser chips can transfer genomic information across breeds if all breeds are in the same dataset. Less dense chips with imputation to higher densities allow affordable selection for smaller populations or more recently recorded traits. The North American database now includes Illumina 2,900 marker (3K) or 50,000 marker (50K) genotypes for 74,389 Holsteins, 8,905 Jerseys, and 2,008 Brown Swiss, plus 777,000 marker (HD) genotypes for 435 animals. To determine how many HD animals within each breed may be needed for imputation, 600,000 marker genotypes were simulated for either the youngest animals or for older bulls with highest reliability, and the other animals had 40,000 markers. After imputation using findhap.f90 version 2, percentages of estimated genotypes that matched true genotypes ranged from 96.1 to 98.7% when numbers of HD genotypes ranged from 250 to 1000 within each of the three breeds. Imputation accuracy was about 1% less if the youngest animals instead of the older bulls had HD. The value of matching cow phenotypes to their own genotype instead of to their sire’s genotype was demonstrated by excluding bulls and using only the 13,935 cows in the Holstein reference population instead of all 25,131 reference bulls and cows (official). For milk yield of young animals, the correlation was 0.86 between cow-only and official evaluations vs. 0.71 between parent average and official. Smaller populations can increase genomic reliability by exchanging information with large populations and by lower cost genotyping.