Y. Masuda1*, I. Misztal1, P.M. VanRaden2, and T. Lawlor3
1University of Georgia, Athens, GA
2Animal Genomics and Improvement Laboratory, ARS, USDA, Beltsville, MD
3Hosltein Association USA, Inc., Brattleboro, VT
2018 J. Dairy Sci. (?)
© American Dairy Science Association,
2018. All rights reserved.
Individuals may download, store, or print single
copies solely for personal use.
Do not share personal accounts or passwords
for the purposes of disseminating this article.
The objective of this study was to validate genomic predictability of single-step genomic BLUP for 305-day protein yield for US Holsteins. The genomic relationship matrix was created with the Algorithm of Proven and Young (APY) with 18,359 core animals. The full data set consisted of phenotypes collected from 1989 through 2015 and pedigrees limited to 3 generations back from phenotyped or genotyped animals. The predictor data set was created by cutting off the phenotypes, pedigree animals, and genotypes in the last 4 years from the full data set. Genomic predictions (GPTA2011) were calculated for predicted bulls that had no recorded-daughters in 2011 but had at least 50 such daughters in 2015. We calculated the daughter yield deviations with the full data (DYD2015) for the predicted bulls (N=3,797). We also used the official GPTA published in 2011 with a multi-step method as a comparison, although the official methods have changed since then. Coefficient of determination (R2) and slope (b1 ) were calculated from a linear regression of DYD2015 on GPTA2011. We investigated the effect of different unknown parent groups (UPGs) and a weight (ω) on the inverse of the pedigree relationship matrix for genotyped animals (A22-1) to compensate incomplete pedigree. When applying QP-transformation to A-1, the R2 was 0.52 with ω=1 compared to 0.51 from the official GPTA. The b1 was similar (0.78) to 0.81 from the official GPTA. Using ω=0.90, the R2 was still similar (0.50) but the b1 was greatly improved (0.96). With QP-transformation in H-1, the R2 was less than 0.4 and the b1 was smaller regardless of ω. Without any UPGs, the predictability and the inflation showed the same level as the official GPTA. The GPTA of a young animal is equivalent to the direct genomic value when many genotypes are included in the evaluation. Fixed UPGs in H-1 added an extra value to GPTA of young animal but this additions is likely redundant in genomic prediction. An option is indirect prediction of direct genomic value using the marker effects. An essential solution may be from metafounders that modify the pedigree relationships according to the genomic relationship matrix.
Keywords: genomic evaluation, incomplete pedigree, Holstein