Translation Genomics Idaho 10/1/11. Alison Van Eenennaam

Transcription

1 Translation Genomics Idaho 1/1/11 Assessing the accuracy of genomic predictions: Results from the California commercial ranch project.alison Van Eenennaam Animal Genomics and Biotechnology Cooperative Extension Specialist OUTLINE Overview of CA Commercial Ranch Project Objectives of the study Number of samples collected Preliminary analysis of the data Kristina Weber Ph.D. graduate student This project is supported by National Research Initiative Grant no to AVE from the USDA National Institute of Food and Agriculture. Department of Animal Science University of California, Davis, CA animalscience.ucdavis.edu/animalbiotech Research objectives of Integrating DNA information into beef cattle production systems Extension objectives of Integrating DNA information into beef cattle production systems How is DNA information best incorporated into beef cattle production systems? The extension objective is to develop and deliver educational materials to a national audience on the integration of DNA information into beef cattle selection programs. Which of several incorporation methods is best? Which is feasible for commercial ranches to implement? Which provides economic benefit? RESEARCH OBJECTIVE: Compare the current means of genetic prediction (bepds) with 1. whole-genome scan genetic predictions (molecular breeding values, MBVs),. commercial ranch genetic evaluations (repds) based on the actual performance of offspring under field conditions. Alison Van Eenennaam Includes the development of fact sheets, national educational programs including program at BIF 9, brown bagger series, popular press articles, and NBCEC workshop entitled Integrating DNA information into beef cattle production systems to be held in Kansas City, MO March 5th 13 1

2 Translation Genomics Idaho 1/1/11 California Commercial Ranch Project Commercial Angus bulls 4 cows/ year Progeny MBV Location of 11 presentations on data derived from CA commercial ranch project Four ranches: Data collection: AAA EPD & pedigree Sample collection: For genotyping Ranch and harvest data Collection Cowley (9 cows) Kuck (5 cows) Mole-Richardson (7 cows) UC Davis (3 cows) Approximately 1 Angus herd bulls, and,4 cows per year on project Genotyping Paternity Determination Assessment of DNA-enabled approaches for predicting the genetic merit of herd sires on commercial beef ranches Cowley Ranch What does a California Commercial Ranch collaborator look like? ~ bulls/season Alison Van Eenennaam Photo taken in 1949 at Red Bluff Bull Sale, CA Generously provided by Cathy Maas from Crowe Hereford Ranch, Millville, CA.

3 Translation Genomics Idaho 1/1/11 Jack and Barbara Cowley Kuck Ranch Dan Drake ~1 bulls/season Mole-Richardson Farms UC Davis Sierra foothills ~3 bulls Work flow and collaborators Up North Pacific Ocean Back East DNA on all bulls goes for whole genome scan collaboration with Jerry Taylor (UMC) and John Pollak (MARC) Molecular breeding value (MBV) prediction of genetic merit based on MARC training data set collaboration with Dorian Garrick (IA) and U.S. Meat Animal Research Center Ranch data including sire groupings, birth dates and weaning weights on all calves, all EIDed, and DNAed for parentage determination collaboration with Dan Drake and producers Steer feedlot in weights, treatments, and carcass traits, weight, grading information and meat sample collected in the processing plant collaboration with Harris Ranch Compile data and compare three sources of genetic estimates: breed EPDs (bepds), commercial ranch EPDs (repds), and MBVs Hollywood Alison Van Eenennaam 3

4 Translation Genomics Idaho 1/1/11 Sample and phenotype collec/on 8 83% Remaining 7 Ranches WW Feedlot In- Weight Carcass Pre- project ~55 head ~46 head ~6 head Spring 9 Fall 9: ~6 head Fall 9/ Winter 1: ~5 head Spring/Summer 1: ~45 head Fall 9 4 Winter/Spring 1: ~15 head Late Summer/ Fall 1: ~9 head Winter 11: ~85 head Spring 1 Fall 1 Fall 1/ Winter 11 Spring/Summer 11 Fall 1 4 Winter/Spring 11 Late Summer/ Fall 11 Winter 1 Spring 11 Fall 11 Fall 11/ Winter 1 Spring/Summer 1 Fall 11 4 Winter/Spring 1 Late Summer/ Fall 1 Winter 13 Total records 4 7 records > collec/on trips 45 records Sent electronically 45 records >35 collec/on trips 78% % In Weight Carcass s 19 1 Birth Weight Alison Van Eenennaam Completed 1 6 Number of Records Calving Date Wean Weight 4

5 Translation Genomics Idaho 1/1/11 Assessing the Genomic Predic/ons: Results from the California Commercial Ranch Project Kris/na Weber, PhD Candidate PD: Alison Van Eenennaam UC Davis Background: Several sets of MBV for quan/ta/ve growth and carcass traits have been developed for beef cawle based on 5K SNP genotypes Commercial tests: IGENITY ( ) and Pfizer Animal GeneScs (MVP) Angus GeneScs Inc. Genomic Enhanced EPDs. Iowa State University and the University of Missouri- Columbia (ISU/UMC) U.S. Meat Animal Research Center (USMARC; Clay Center, NE). At UCD, we have a popula/on of Angus bulls purchased as yearlings with many progeny records for weaning weight, feedlot in- weight, and carcass traits which we can use to assess the gene/c merit of these bulls in a Northern California environment ObjecSve: In this study, the accuracies of 5K- derived MBV were assessed rela/ve to ranch- based breeding values calculated from commercial progeny phenotypes of purebred Angus bulls. Weber, K.L., D.J. Drake, J.F. Taylor, D.J. Garrick, L.A. Kuehn, R.M. Thallman, R.D. Schnabel, W.M. Snelling, E.J. Pollak, and A.L. Van Eenennaam. 1. The accuracies of DNA- based essmates of genesc merit derived from Angus- or muls- breed beef cadle training populasons. J. Anim. Sci. (submi.ed). DNA Test Number of tested bulls WW ADG CW, MS, RE ISU/UMC MVP AN HH Total Bulls MBV Considered ISU/UMC: Iowa State University and University of Missouri- Columbia, Angus, 5K, training: GBLUP with up to 3,57 records : IGENITY, Angus, 384 SNP panel MVP: Pfizer, Angus, 5K, training: Bayesian model with up to 1,445 records 11 natural service bulls from four ranches were 5K genotyped. ISU/UMC predicsons were available for 99 bulls at the Sme of publicason. Due to the cost of purchasing DNA test results, IGENITY and Pfizer predicsons were purchased for the 9 bulls with the highest number of progeny records. DNA Test Number of tested bulls WW ADG CW, MS, RE ISU/UMC MVP AN HH Total Bulls MBV Considered : USMARC, Germplasm EvaluaSon Program Cycle VII and new crossbred, 5K, training: BayesCπ with up to 3,358 phenotypic records : USMARC, Bull Project muls- breed, 5K, training: BayesCπ with up to,6 records AN : USMARC, Angus, 5K, training: BayesCπ with 373 records HH : USMARC, Hereford, 5K, training: BayesCπ with 463 records Published essmates of MBV Accuracy DNA Test Reference Accuracy(±SE where available) Angus Northcud, MVP Pfizer Technical Summary Northcud, AN Weber et al., Mul/- breed Weber et al., Weber et al., Weber, K.L., R.M. Thallman, J.W. Keele, W.M. Snelling, G.L. Benned, T.P.L. Smith, T.G. McDaneld, M.F. Allan, A.L. Van Eenennaam, and L.A. Kuehn. 1. genomic breeding values in muls- breed beef cadle populasons derived from deregressed breeding values and phenotypes. J. Anim. Sci. (submi.ed). HH HH Weber et al., 1.4. Weber, K.L., R.M. Thallman, J.W. Keele, W.M. Snelling, G.L. Benned, T.P.L. Smith, T.G. McDaneld, M.F. Allan, A.L. Van Eenennaam, and L.A. Kuehn. 1. genomic breeding values in muls- breed beef cadle populasons derived from deregressed breeding values and phenotypes. J. Anim. Sci. (submi.ed). Alison Van Eenennaam 5

6 Translation Genomics Idaho 1/1/11 Popula/on Structure and Rela/onship to Training Popula/ons Birth year ranged from - 9 The UCD bull populason included: 3 sets of full siblings sets of paternal half siblings 1 pair of maternal half siblings These families ranged in size from - 9, with siblings present on up to 3 different ranches Popula/on Structure and Rela/onship to Training Popula/ons Rela/onship to training popula/on ISU/UMC Data available to AAA by the Sme of bull sale (i.e. no progeny data) was included in ISU/UMC training set for 87 UCD bulls. 79 bulls sires were present in the ISU/UMC training populason Of the remaining bulls tested, 15 had grandsires and/or great- grandsires present in the ISU/UMC training populason 71 bulls sires were present in the training populason Of the remaining 5 bulls, 44 had grandsires and/or great- grandsires present in the training populason 1 UCD bulls were related to animals in the training populason through sharing a common sire Frequency Frequency Expected ISU/UMC PredicSons WW Sire not in training Sire in training GBLUP Accuracy MS GBLUP Accuracy Frequency Frequency CW GBLUP Accuracy RE GBLUP Accuracy Number of UCD Bulls with Phenotyped Progeny and the Number of Progeny Per Bull DNA Test ISU/ UMC MVP AN HH Mean progeny number (range) WW 44 ( ( (1-15 ADG 44 (15-1 CW, MS, RE 6 (1-13) 48 (11-13) (1-13) The bulls for which the IGENITY and Pfizer DNA tests were purchased had 31 more progeny WW records and 3 more carcass records than average for the complete dataset. Number of UCD Bulls with Phenotyped Progeny and the Number of Progeny Per Bull DNA Test ISU/ UMC MVP AN HH Mean progeny number (range) WW 44 ( ( (1-15 ADG 44 (15-1 CW, MS, RE 6 (1-13) 48 (11-13) (1-13) The bulls for which the IGENITY and Pfizer DNA tests were purchased had 31 more progeny WW records and 3 more carcass records than average for the complete dataset. Progeny phenotypes Angus sires Progeny Units Mean SD Min Max phenotypes Weaning weight (WW) 19 4,7 lb Feedlot average daily 75 1,9 lb/day gain (ADG) Carcass weight (CW) 136,865 lb Marbling score (MS) 136,864 * Ribeye area (RE) 136,864 in *3=traces, 4=slight, 5=small, 6=modest, 7=moderate, 8=slightly abundant, 9=moderately abundant WW was adjusted for age at weaning and age of dam prior to analysis ADG was essmated using rate of gain from feedlot in- weight to essmated feedlot final weight derived from CW, backfat thickness, and RE. Fixed effects: Contemporary group: hys for WW, hys+feedlot lot for ADG, and hys+harvest lot for HCW, MS, and RE Age for carcass traits Sex for WW, HCW, and MS. Fixed effects were tested for significance (p<) as computed by ASREML from incremental Wald F stassscs (Gilmour et al., 9). Alison Van Eenennaam 6

7 Translation Genomics Idaho 1/1/11 The importance of collecsng meat samples and verifying live animal- carcass idensficason The importance of collecsng meat samples and verifying live animal- carcass idensficason Error rates in abawoir reported IDs Average error rate of 1.8% across 5 consecusve cohorts from one ranch: 3.5% in 165 head 19.3% in 9 head 6.4% in 167 head 8.1% in 16 head 16.5% in 14 head Reasons: Rail outs Inversions Failure to record animals, leading to a sequence of records offset by one or two records from the correct ID Example Gang Tag Carcass ID Expected live Actual live animal ID animal ID Not recorded Linear Model for EsSmaSng MBV Accuracy Accuracy = genesc correlason between the MBV and the ranch- based essmate of the genesc merit of the bulls (Kachman, 8). For WW: u Aσ Aσ Aσ y uy uyumbv uyum u σ σ σ MBV A A A uyumbv umbv umbvum var u = m Aσ u σ σ y u m A u MBV u m A u m σ e I e ε MBV Iσ ε MBV For all other traits, the same model was used, with Z m u m excluded. Variance components σ A ±SE 663±15.7± 1556±86.384±.5.4±.7 σ AM ±SE 199±81 σ M ±SE 843±94 σ E ±SE 1994±17.18± 4±3.398±.6.77±.6 Ranch and AAA EBV Accuracies Sire EBV±SE AAA EPD±SE.8± (-.5 4).3 (.-.5 ).7± (-.6 3).16±.3± (-.6 6) 8).± (-.6 h A ±SE.179±.4.67±.7.393±.7.59±.7.35±.6 Ranch and AAA EBV Accuracies Sire EBV±SE Young bulls that have been genotyped but don t have phenotyped progeny yet.8± (-.5.3 (.-.5 ).7± (-.6 3).3± (-.6 6).± (-.6 Ranch and AAA EBV Accuracies Sire EBV±SE.8± (-.5.3 (.-.5 ).7± (-.6 3).3± (-.6 6).± (-.6 AAA EPD±SE 4).16± 8) AAA EPD±SE 4).16± 8) Pedigree only +Bull s own phenotype Alison Van Eenennaam 7

8 Translation Genomics Idaho 1/1/11 Ranch and AAA EBV Accuracies + Many phenotyped progeny Number of Sire EBV±SE AAA EPD±SE.8± (-.5 4).3 (.-.5 ).7± (-.6 3).16±.3± (-.6 6) 8).± (-.6 Gene/c correla/on with UCD Data bulls Accuracy±SE WW- d ADG CW MS RE AAA data 1.15±.8.14±.19.6±..53±.13 ISU/UMC 99.9±.14.7±.14.64±.1.64± ±..33±..9±.3.44±.18.3±.1 MVP 9.79±.1 -.3±.4.9±..68±.1.68±.13 AN 11.4±.13.15±.14.4±.1.3± ±.18.19±.15.18±.17.1±.13 ALL 11.6±.13.19±.14.37±.1.17±.14 HH 11 ± ± Gene/c correla/on with UCD Data Gene/c correla/on with UCD Data Number of bulls Accuracy±SE WW- d ADG CW MS RE AAA data 1.15±.8.14±.19.6±..53±.13 ISU/UMC 99.9±.14.7±.14.64±.1.64± ±..33±..9±.3.44±.18.3±.1 MVP 9.79±.1 -.3±.4.9±..68±.1.68±.13 AN 11.4±.13.15±.14.4±.1.3± ±.18.19±.15.18±.17.1±.13 ALL 11.6±.13.19±.14.37±.1.17±.14 HH 11 ± ± Number of bulls Accuracy±SE WW- d ADG CW MS RE AAA data 1.15±.8.14±.19.6±..53±.13 ISU/UMC 99.9±.14.7±.14.64±.1.64± ±..33±..9±.3.44±.18.3±.1 MVP 9.79±.1 -.3±.4.9±..68±.1.68±.13 AN 11.4±.13.15±.14.4±.1.3± ±.18.19±.15.18±.17.1±.13 ALL 11.6±.13.19±.14.37±.1.17±.14 HH 11 ± ± Gene/c correla/on GeneSc correlason with UCD Data AAA data MVP ISU/UMC AN HH WW- d ADG CW MS RE Gene/c correla/on GeneSc correlason with UCD Data MVP ISU/UMC WW- d CW MS RE Alison Van Eenennaam 8

9 Translation Genomics Idaho 1/1/11 Pairwise Gene/c Correla/ons between MBV Conclusions ISU/UMC x MVP ISU/UMC x MVP x WW.7±.13.49±.14.31±.17 ADG.15±.18 CW.64±.1.51±.1.4±.16 MS.77±.13.69±.6.43±.15 RE.64±.1.5±.13.48±.15 MBV accuracies for commercially available tests were similar to those reported for the Angus breed but for traits in which ranch EPD were not well correlated with AAA EPD, there was a trend of lower than expected MBV accuracy MBV that were not derived from Angus were less accurate than Angus- derived MBV High correlasons observed between ISU/UMC, MVP and for all traits except ADG. Future DirecSons Illumina BovineHD genotyping and imputason up to HD from 5K for the training and assessment populasons has begun Preliminary results suggest that there is some improvement in muls- breed MBV accuracy when training on HD genotype data Finish collecsng all of the data and graduate! USDA Integrated Grant Collaborators Integrating DNA information into Beef Cattle Production Systems Producer Collaborators: Jack Cowley, Cowley Ranch, Siskiyou County, CA Dale, Greg, and Richard Kuck, Kuck Ranch, Siskiyou County, CA Matt Parker, Mole-Richardson Farms, Siskiyou County, CA Processor Collaborators: Harris Ranch Beef Company, Coalinga, CA Los Banos Abattoir, Los Banos, CA Software Collaborators: Jim Lowe, Cow Sense Herd Management Software, NE Other Contributors/Collaborators Dr. Darrh Bullock, Extension Professor, University of Kentucky, KY Dr. Jerry Taylor, University of Missouri, MO Dr. Daniel Drake, University of California Cooperative Extension Livestock Advisor, CA Dr. Dorian Garrick, Professor, Iowa State University, IA Dr. John Pollak, US Meat Animal Research Center, Clay Center, NE Dr. Larry Kuehn, US Meat Animal Research Center, Clay Center, NE Dr. Mark Thallman, US Meat Animal Research Center, Clay Center, NE Dr. Warren Snelling, US Meat Animal Research Center, Clay Center, NE Dr. Matt Spangler, University of Nebraska, NE Dr. Bob Weaber, Kansas State University, KS Alison Van Eenennaam 9