Skip to main content
ARS Home » Northeast Area » Beltsville, Maryland (BARC) » Beltsville Agricultural Research Center » Animal Genomics and Improvement Laboratory » Research » Publications at this Location » Publication #351512
Research Project: Improving Dairy Animals by Increasing Accuracy of Genomic Prediction, Evaluating New Traits, and Redefining Selection Goals

Location: Animal Genomics and Improvement Laboratory

Title: Modelling uncertain paternity to address differential pedigree accuracy

Author
item BRADFORD, HEATHER - University Of Georgia
item MASUDA, YATAKA - University Of Georgia
item Cole, John
item MISZTAL, IGNACY - Abbott Nutrition
item Vanraden, Paul

Submitted to: Journal of Dairy Science
Publication Type: Abstract Only
Publication Acceptance Date: 3/8/2018
Publication Date: 6/24/2018
Citation: Bradford, H.L., Masuda, Y., Cole, J.B., Misztal, I., Van Raden, P.M. 2018. Modelling uncertain paternity to address differential pedigree accuracy [abstract]. Journal of Dairy Science. 101(Suppl. 2):181-182(abstr. 170).

Interpretive Summary:

Technical Abstract: The objective was to implement uncertain parentage models to account for differences in daughter pedigree accuracy. Elite sires have nearly all daughters genotyped resulting in correct paternity assignment. Bulls of lesser genetic merit have fewer daughters genotyped creating the possibility for more paternity errors in their daughters. Data were simulated with QMSim for a moderately (0.3) heritable, sex-limited trait. We created missing pedigrees by removing 8% of sires and 18% of dams. In total, 15 bulls were selected each generation, and the daughters of the best 5 bulls had accurate pedigrees. Daughters of the remaining 10 bulls had 9% sire and 3% dam pedigree errors. Data included 164,500 pedigree animals, 90,000 phenotypes, and 18,000 genotypes and were modeled with an overall mean, additive genetic, and residual effects using single-step genomic BLUP (ssGBLUP) with unknown parent groups. The uncertain parentage model partitioned contributions in A-1 to the parent on record (90 to 100%) and to the appropriate unknown parent group (0 to 10%) depending on the type of animal. We validated predictions based on the youngest animals (n=14,950) without phenotypes. Accuracy was the correlation between true and estimated breeding values. Accuracies (SE) were nearly identical with 0.65 (0.01) for ssGBLUP and 0.64 (0.01) for uncertain parentage. Dispersion was the regression of true on estimated breeding values, and no differences existed between the models with dispersion (SE) of 0.83 (0.01) for ssGBLUP and 0.84 (0.01) for uncertain parentage. Bias was the difference between true and estimated breeding value and was scaled by the genetic standard deviation. Both models had bias (SE) of 0.24 (0.01). Similarly, bias differences were small when evaluating subsets of animals based on pedigree accuracy. Dairy datasets are complex, especially with regard to differences in daughter parentage accuracies across bulls. This complexity is difficult to simulate, and uncertain parentage models should be tested because of the potential to mitigate bias differences across bulls.