Genomic Predictability of Single-Step GBLUP in US Holstein Cows
Explore the genomic predictability of single-step GBLUP for production traits in US Holstein cows. The study validates genomic predictions for young bulls using different UPG configurations and discusses models to handle UPG in ssGBLUP. Full data analysis, validation studies, and comparison of DYD2015 vs GPTA2011 results are presented.
Download Presentation
Please find below an Image/Link to download the presentation.
The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author. If you encounter any issues during the download, it is possible that the publisher has removed the file from their server.
You are allowed to download the files provided on this website for personal or commercial use, subject to the condition that they are used lawfully. All files are the property of their respective owners.
The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author.
E N D
Presentation Transcript
Genomic predictability of single-step GBLUP for production traits in US Holstein Y. Masuda1, P. M. VanRaden2, H. L. Bradford2, A. Legarra3, I. Misztal1, and T. J. Lawlor4 1 University of Georgia, USA; 2 AGIL, USDA, USA; 3 INRA, France; 4 Holstein Association USA, Inc., USA ADSA 2018, June 24-27, Knoxville, TN
Background Genomic prediction with single-step GBLUP (ssGBLUP) Genotyped + non-genotyped animals Accountability for pre-selection APY: dimensionality reduction in marker genotypes Compatibility among relationship matrices ? 1= ? 1+0 0 1 ? 1 ??22 0 Reasonable in complete pedigree Missing pedigree: adjustment of ?22 How to use unknown parent groups (UPG)? 1by ?
Objectives To validate genomic predictions for young bulls by different UPG configurations in US Holstein To discuss possible models to handle UPG in ssGBLUP: a simulation study
Full data Number of records/animals Description Milk, fat, and protein yield (305-d basis) for US Holstein cows recorded between Jan. 1990 and Apr. 2015 Phenotype 37,259,427 Cows with phenotype(s) 15,891,366 Animals born in Apr. 2015 or earlier (3-gen. back from phenotyped cows) 185 UPGs Pedigree 22,963,255 Animals born in Apr. 2015 or earlier (60,671 markers) Genotype 764,029
Validation study 2011 Dec. 2015 Apr. Validation Bulls: Genotyped young bulls with no tested daughters in 2011 but with at least 50 tested daughters in 2015 (N=3,797) 1990 2000 Full For Daughter Yield Deviation (DYD2015) Phenotype Pedigree 2011 Dec. 2015 Apr. 1990 2000 Trunc2011 For GPTA using ssGBLUP (GPTA2011) Phenotype Genotype Pedigree ???2015 = ?1 ????2011 + ?0 R2 : validation reliability Slope (?1): Inflation of prediction
Different UPG in ?1 1: 0.9 or 1.0 1. Weight (?) on ?22 2. UPG: pedigree + genomic UPG, pedigree UPG only, or no UPG (genomic UPG) (pedigree UPG) 0 0 0 0 0 0 0 0 0 0 0 0 0 1?2 1)?2 ? 1 ??22 ?2 ? = ? + 1 ? 1 ??22 0 + (? 1 ??22 1) (? 1 ??22 ?2
DYD2015 vs GPTA2011 (Protein) Data Official GPTA 2011 R2 b1 0.51 0.81 =0.9 =1.0 Data GPTA2011 UPG Genomic UPG different inb. different UPG Pedigree UPG No UPGs R2 b1 R2 b1 0.39 0.39 0.74 0.75 0.32 0.34 0.32 0.52 0.50 0.51 0.51 0.50 0.78 0.78 0.50 0.96
Low accuracy with exact UPG GPTA for young genotypes PedigreeUPG: Genomic UPG: ???? = ?1?? + ?2??? ?3?? ???? = ?1?? + ?2??? ?3?? + ?4??? ??? + ??? ??? Larger weights with many genotypes Too large for young animals Possible solutions Discounting UPG effects Removing double counting between DGV and UPG Scaling ? to ? ( metafounders ) 0 0 0 0 0 1 1?2 1)?2 ? 1 ?22 ?2 ? 1 ?22 ?2 ? = ? + (? 1 ?22 1) (? 1 ?22
1 with UPG Modified ?22 Indirect inversion 1= ?22 ?21?11 1?12 ?22 where ? 1=?11 ?12 ?22 ?21 With UPG ?11 ?31 ?13 ?12 ?32 = ?22 ?21 ?23 ?22 ?33+ ? ?11 ?21 ?12 ?22 ?11 ?21 ?12? ?22? ?11 ?21 ?31 ?12 ?22 ?32 ?13 ?23 ?33 where A = = ? ?11 ? ?12 ? ? 1? ?21 ?22
Missing parents in ssGBLUP Genomic UPG 0 0 0 0 0 ?22 0 1?2 ?22 0 0 0 0 0 ? 1 0 0 0 0 0 0 0 0 ? 1?2 ?2 ?22 ?2 ? = ? + ? ? 1 ??? 0 + ? 1?2 1 1?2 ?2 ?2 Genomic UPG without Q GQ 0 0 0 Pedigree UPG 0 0 0 0 0 0 0 0 ?22 0 1?2 ?22 0 0 0 0 0 ? 1 0 0 0 0 0 ? 1?2 ?2 ?22 ?2 ? = ? + ? ? 1 ??? 0 + ? 1?2 1 1?2 ?2 ?2 0 0 0 0 ? = ? + ? ? 1 ??? 0
With ?22 Genomic UPG 0 0 0 0 0 ?22 0 1?2 ?22 0 0 0 0 0 ? 1 0 0 0 0 0 0 0 0 ? 1?2 ?2 ?22 ?2 ? = ? + ? ? 1 ??? 0 + ? 1?2 1 1?2 ?2 ?2 Genomic UPG without Q GQ 0 0 0 Pedigree UPG 0 0 0 0 0 0 0 0 ?22 0 1?2 ?22 0 0 0 0 0 ? 1 0 0 0 0 0 ? 1?2 ?2 ?22 ?2 ? = ? + ? ? 1 ??? 0 + ? 1?2 1 1?2 ?2 ?2 0 0 0 0 ? = ? + ? ? 1 ??? 0
Simulation study Structure h2 = 0.3 Sex-limited trait (n = 90,000) EBV selection 10 generations (n = 164,500) Ne: 200 theoretical; 25 realized Mean F in last generation: 0.11 Genotypes 18,674 total 5108 in gen. 10 for validation Assignment of UPGs UPG1 for generation 0-4 UPG2 for generation 5-7 UPG3 for generation 8-10 Non genotyped 0 Category Top bulls Genotyped 0 Top cows Bottom bulls 30% (dam) Bottom cows 30% (dam) 5% (dam) 0 10% (dam) 10% (dam)
Results from simulation ? Modified ??? R2 0.63 0.61 0.63 Standard ??? R2 0.63 0.53 0.62 b1 b1 Pedigree UPG Exact UPG without Q GQ 1.06 0.86 1.05 1.06 1.01 1.04 R2 b1 Metafounders 0.63 1.08 * Genotyped young animals without records
Summary Missing pedigree may reduce the accuracy of genomic prediction in single-step GBLUP. Specific data structure with many missing parents Double-counting in UPG effects We have several options to discount the possible double-counting. Removal of ? 1 contribution Use of Modified ?22 Metafounders 1
Acknowledgement USDA NIFA (2015-67015-22936) and Holstein Association USA for financial support. Council of Dairy Cattle Breeding for phenotype, genotype, and pedigree data. John Cole and Melvin Tooker (USDA-AGIL) for preparing the initial data sets and a computing environment.
