1. WiggansCornell University, ANSC 3310, March 10, 2015 (1) Dr. George R. Wiggans Animal Genomics and Improvement Laboratory Agricultural Research Service, USDA Beltsville, MD 20705-2350 301-504-8407 (voice) 301-504-8092 (fax) george.wiggans@ars.usda.gov US genomic evaluation system

2. WiggansCornell University, ANSC 3310, March 10, 2015 (2) History of genomic evaluations l BovineSNP50 BeadChip availableDec. 2007 l First unofficial evaluation releasedApr. 2008 l Official evaluations for Holsteins and JerseysJan. 2009 l Official evaluations for Brown SwissAug. 2009 l Monthly evaluation Jan. 2010 l Official 3K evaluationsDec. 2010 l BovineLD BeadChip availableSept. 2011 l Official evaluations for Ayrshires Apr. 2013 l Weekly evaluation Nov. 2014

3. WiggansCornell University, ANSC 3310, March 10, 2015 (3) Collaboration with industry l Council on Dairy Cattle Breeding (CDCB) responsible for receiving data and for computing and delivering US genetic evaluations for dairy cattle l Animal Genomics and Improvement Lab (AGIL) responsible for research and development to improve the evaluation system l CDCB and AGIL employees co-located in Beltsville Dr. João Dürr is CDCB CEO

4. WiggansCornell University, ANSC 3310, March 10, 2015 (4) Staff l Research team w 4 senior scientists w 6 support scientists w 4 information technology specialists w 1 administrative assistant l On-site collaborators w Council of Dairy Cattle Breeding (CEO, systems administrator, and 2 consultants)

5. WiggansCornell University, ANSC 3310, March 10, 2015 (5) Funding l CDCB evaluation calculation and dissemination funded by fee system w Based on animals genotyped w 87% of revenue from bulls w Higher fees for herds that contribute less information l USDA research on evaluation methodology funded by US Federal Government $

6. WiggansCornell University, ANSC 3310, March 10, 2015 (6) Council on Dairy Cattle Breeding l 3 members from each organization l Total of 12 voting members l 2 nonvoting industry members Purebred Dairy Cattle Association National Association of Animal Breeders Dairy Records Processing Centers Dairy Herd Information

7. WiggansCornell University, ANSC 3310, March 10, 2015 (7) Genomic data flow DNA samples genotypes genomic evaluations nominations, pedigree data genotype quality reports genomic evaluations DNA samples genotypes DNA samples Dairy Herd Information (DHI) producer Council on Dairy Cattle Breeding (CDCB) DNA laboratory AI organization, breed association AI organization, breed association

8. WiggansCornell University, ANSC 3310, March 10, 2015 (8) Evaluation flow l Animal nominated for genomic evaluation by approved nominator l DNA source sent to genotyping lab (2014)  SourceSamples (no.)Samples (%) Blood 10,7274 Hair113,45539 Nasal swab2,9541 Semen3,4321 Tissue149,30151 Unknown12,3014

9. WiggansCornell University, ANSC 3310, March 10, 2015 (9) Evaluation flow (continued) l DNA extracted and placed on chip for 3-day genotyping process l Genotypes sent from genotyping lab to CDCB for accuracy review  

10. WiggansCornell University, ANSC 3310, March 10, 2015 (10) Genotype chips ChipSNP (no.)ChipSNP (no.) 50K54,001GP219,809 50K v254,609ZLD11,410 3K2,900ZMD56,955 HD777,962ELD9,072 Affy648,875LD26,912 LD6,909GP326,151 GGP8,762ZL217,557 GHD77,068ZM260,914

11. WiggansCornell University, ANSC 3310, March 10, 2015 (11) Laboratory quality control l Each SNP evaluated for w Call rate w Portion heterozygous w Parent-progeny conflicts l Clustering investigated if SNP exceeds limits l Number of failing SNP indicates genotype quality l Target of <10 SNP in each category

12. WiggansCornell University, ANSC 3310, March 10, 2015 (12) Parentage validation and discovery l Parent-progeny conflicts detected w Animal checked against all other genotypes w Reported to breeds and requesters w Correct sire usually detected l Maternal grandsire checking w SNP at a time checking w Haplotype checking more accurate l Breeds moving to accept SNP in place of microsatellites

13. WiggansCornell University, ANSC 3310, March 10, 2015 (13) Before clustering adjustment 86% call rate

14. WiggansCornell University, ANSC 3310, March 10, 2015 (14) After clustering adjustment 100% call rate

15. WiggansCornell University, ANSC 3310, March 10, 2015 (15) Evaluation flow (continued) l Genotype calls modified as necessary l Genotypes loaded into database l Nominators receive reports of parentage and other conflicts l Pedigree or animal assignments corrected l Genotypes extracted and imputed to 61K l SNP effects estimated l Final evaluations calculated       

16. WiggansCornell University, ANSC 3310, March 10, 2015 (16) Evaluation flow (continued) l Evaluations released to dairy industry w Download from CDCB FTP site with separate files for each nominator w Weekly release of evaluations of new animals w Monthly release for females and bulls not marketed w All genomic evaluations updated 3 times each year with traditional evaluations 

17. WiggansCornell University, ANSC 3310, March 10, 2015 (17) 2014 genotypes by chip SNP density Chip SNP densityFemaleMale All animals Low 239,07129,631268,702 Medium9,09814,20223,300 High14028168 All248,309 43,861292,170

18. WiggansCornell University, ANSC 3310, March 10, 2015 (18) 2014 genotypes by breed and sex BreedFemaleMale All animals Female: male Ayrshire 1,4852091,69488:12 Brown Swiss9448,6419,58510:90 Guernsey1,7773332,11084:16 Holstein212,76530,883243,64887:13 Jersey31,3233,79335,11689:11 Milking Shorthorn21367:33 Normande0100:100 Crossbred130 100:0 All248,30943,861292,17085:15

19. WiggansCornell University, ANSC 3310, March 10, 2015 (19) Genotypes by animal age (last 12 months)  

20. WiggansCornell University, ANSC 3310, March 10, 2015 (20) Growth in bull predictor population BreedJan. 201512-mo gain Ayrshire 71129 Brown Swiss6,112336 Holstein26,7592,174 Jersey4,448245

21. WiggansCornell University, ANSC 3310, March 10, 2015 (21) Holstein prediction accuracy *2013 deregressed value – 2009 genomic evaluation TraitBias*Reliability (%) Reliability gain (% points) Milk (kg)−80.369.230.3 Fat (kg)−1.468.429.5 Protein (kg)−0.960.922.6 Fat (%)0.093.754.8 Protein (%)0.086.348.0 Productive life (mo)−0.773.741.6 Somatic cell score 0.064.929.3 Daughter pregnancy rate (%) 0.253.520.9 Sire calving ease 0.645.819.6 Daughter calving ease−1.844.222.4 Sire stillbirth rate 0.228.2 5.9 Daughter stillbirth rate0.137.617.9

22. WiggansCornell University, ANSC 3310, March 10, 2015 (22) Holstein prediction accuracy *2013 deregressed value – 2009 genomic evaluation TraitBias*Reliability (%) Reliability gain (% points) Final score0.158.822.7 Stature−0.268.530.6 Dairy form−0.271.834.5 Rump angle0.070.234.7 Rump width−0.265.028.1 Feed and legs0.244.012.8 Fore udder attachment −0.270.433.1 Rear udder height −0.159.422.2 Udder depth −0.375.337.7 Udder cleft−0.262.125.1 Front teat placement −0.269.932.6 Teat length−0.166.729.4

23. WiggansCornell University, ANSC 3310, March 10, 2015 (23) Reliability gains Reliability (%)Ayrshire Brown SwissJerseyHolstein Genomic37546170 Parent average2830 Gain 9243140 Reference bulls 6805,767 4,207 24,547 Animals genotyped1,7889,01659,923469,960 Exchange partnersCanadaCanada, Interbull Canada, Denmark Canada, Italy, UK Source: VanRaden, Advancing Dairy Cattle Genetics: Genomics and Beyond presentation, Feb. 2014

24. WiggansCornell University, ANSC 3310, March 10, 2015 (24) Gene tests (imputed and actual) l Bovine leucocyte adhesion deficiency (BLAD) l Complex vertebral malformation (CVM) l Deficiency of uridine monophosphate synthase (DUMPS) l Syndactyly (mulefoot) l Weaver Syndrome, spinal dismyelination (SDM), spinal muscular atrophy (SMA) l Red coat color l Polledness

25. WiggansCornell University, ANSC 3310, March 10, 2015 (25) New fertility haplotype for Jerseys (JH2) l Chromosome 26 at 8.8 – 9.4 Mbp l Carrier frequency w 14 – 28% in decades before 1990 w Only 2.6% now l Estimated effect on conception rate of – 4.0% ± 1.5% l Additional sequencing needed to find causative genetic variant

26. WiggansCornell University, ANSC 3310, March 10, 2015 (26) Parent ages for marketed Holstein bulls

27. WiggansCornell University, ANSC 3310, March 10, 2015 (27) Inbreeding for Holstein cows – Inbreeding – Expected future inbreeding

28. WiggansCornell University, ANSC 3310, March 10, 2015 (28) Active AI bulls that were genomic bulls

29. WiggansCornell University, ANSC 3310, March 10, 2015 (29) Marketed Holstein bulls Year entered AI Traditional progeny- tested Genomic marketed All bulls 20081,7681701,938 20091,4743461,820 20101,3883931,781 20111,2546481,902 20121,2397061,945 2013 9077471,654 2014 6617921,453

30. WiggansCornell University, ANSC 3310, March 10, 2015 (30) Genetic merit of marketed Holstein bulls Average gain: $19.42/year Average gain: $47.95/year Average gain: $87.49/year

31. WiggansCornell University, ANSC 3310, March 10, 2015 (31) Stability of genomic evaluations l 642 Holstein bulls w Dec. 2012 NM$ compared with Dec. 2014 NM$ w First traditional evaluation in Aug. 2014 w  50 daughters by Dec. 2014 l Top 100 bulls in 2012 w Average rank change of 9.6 w Maximum drop of 119 w Maximum rise of 56 l All 642 bulls w Correlation of 0.94 between 2012 and 2014 w Regression of 0.92

32. WiggansCornell University, ANSC 3310, March 10, 2015 (32) Haplotypes affecting fertility l Rapid discovery of new recessive defects w Large numbers of genotyped animals w Affordable DNA sequencing l Determination of haplotype location w Significant number of homozygous animals expected, but none observed w Narrow suspect region with fine mapping w Use sequence data to find causative mutation

33. WiggansCornell University, ANSC 3310, March 10, 2015 (33) Haplotypes affecting fertility *Causative mutation known Name BTA chromo- some Location* (Mbp) Carrier frequency (%)Earliest known ancestor HH1563.2*3.8Pawnee Farm Arlinda Chief HH2194.9 – 96.63.3Willowholme Mark Anthony HH3895.4*5.9 Glendell Arlinda Chief, Gray View Skyliner HH411.3*0.7Besne Buck HH5992.4 – 93.94.4Thornlea Texal Supreme JH11515.7*24.2Observer Chocolate Soldier JH2268.8 – 9.42.6Liberators Basilius BH1742.8 – 47.013.3West Lawn Stretch Improver BH21910.6 – 11.715.6Rancho Rustic My Design AH11765.9*26.0Selwood Betty’s Commander

34. WiggansCornell University, ANSC 3310, March 10, 2015 (34) Haplotype tracking of known recessives *Causative mutation known Recessive Haplo- type BTA chromo- some Tested animals (no.) Concord- ance (%) New carriers (no.) BrachyspinaHH021??? BLADHHB 1*11,78299.9314 CVMHHC 3*13,226—2,716 DUMPSHHD 1*3,242100.03 Mule footHHM15*8797.7120 PolledHHP1345—2,050 Red coat colorHHR18*4,137—5,927 SDMBHD11*10894.4108 SMABHM24*56898.1111 WeaverBHW416396.332

35. WiggansCornell University, ANSC 3310, March 10, 2015 (35) Recent accomplishments l Introduction of imputed indicators for inherited defects of dairy cattle l Introduction of genomic evaluations for Ayrshires l Discovery of additional haplotypes that affect fertility l Improved accuracy of genomic evaluations by an increase to 60,671 DNA markers l Improved weighting of cow evaluations l Multitrait traditional evaluations for heifer and cow conception rates

36. WiggansCornell University, ANSC 3310, March 10, 2015 (36) December 2014 changes l Net merit update l Grazing index l Genomic mating program l Base change l Weekly evaluations l New computer programs for traditional evaluations l New definition of daughter pregnancy rate

37. WiggansCornell University, ANSC 3310, March 10, 2015 (37) Weekly evaluations l Released to nominators, breed associations, and dairy records processing centers at 8 am each Tuesday l Calculations restricted to genotypes that first became usable during the previous week l Computing time minimized by not calculating reliability or inbreeding

38. WiggansCornell University, ANSC 3310, March 10, 2015 (38) Application to more traits l Animal’s genotype good for all traits l Traditional evaluations required for accurate estimates of SNP effects l Traditional evaluations not currently available for heat tolerance or feed efficiency l Research populations could provide data for traits that are expensive to measure l Will resulting evaluations work in target population?

39. WiggansCornell University, ANSC 3310, March 10, 2015 (39) What’s already planned l Genomic evaluations for new traits w Health w Feed efficiency l Genomic mating programs w Selection of favorable minor alleles w Reduction of genomic inbreeding l Adding SNP for causative genetic variants

40. WiggansCornell University, ANSC 3310, March 10, 2015 (40) What’s already planned (continued) l BARD project (Volcani Center, Israel) w A posteriori granddaughter design (APGD) w Identification of causative variants for economically important traits l International collaboration on sequencing w United States, United Kingdom, Italy, Canada w Bulls selected using APGD w Participation in 1000 Bull Genomes project

41. WiggansCornell University, ANSC 3310, March 10, 2015 (41) GeneSeek 77K chip (GHD) l Designed to include the most informative SNP l 76,934 SNP typically provided, including wYwY w Single-gene tests l About 28,200 50K SNP included (mostly low MAFs excluded) l Other SNP selected from HD based on MAF and magnitude of effect

42. WiggansCornell University, ANSC 3310, March 10, 2015 (42) Current SNP set for genomic evaluations l 60,671 SNP used after culling on w MAF w Parent-progeny conflicts w Percentage heterozygous (departure from HWE) l SNP for HH1, BLAD, DUMPS, CVM, polled, red, and mulefoot included w JH1 included for Jerseys l Some SNP eliminated because incorrect location  haplotype non-inheritance

43. WiggansCornell University, ANSC 3310, March 10, 2015 (43) New GHD version (Expected this month) l Around 143,000 SNPs expected l Include 16,248 among 60,671 SNPs currently used that are not on GHD l Many added SNPs have low to moderate minor allele frequency w Increasing to 85,000 SNP improves evaluation accuracy

44. WiggansCornell University, ANSC 3310, March 10, 2015 (44) Low-cost chip (expected this month) l ~4,100 SNPs l Built-in validation l Single-gene tests l Lower imputation accuracy if neither parent genotyped l Imputation accuracy within 1% of LD chip if at least 1 parent genotyped

45. WiggansCornell University, ANSC 3310, March 10, 2015 (45) Mating programs l Match genotypes of parents to minimize genomic inbreeding l Avoid mating carriers l Consider nonadditive gene action l May attempt to increases variance to get outliers

46. WiggansCornell University, ANSC 3310, March 10, 2015 (46) Why genomics works for dairy cattle l Extensive historical data available l Well-developed genetic evaluation program l Widespread use of AI sires l Progeny-test programs l High-value animals worth the cost of genotyping l Long generation interval that can be reduced substantially by genomics

47. WiggansCornell University, ANSC 3310, March 10, 2015 (47) Future l Discovery of causative genetic variants w Do not have linkage decay w Added to chips as discovered w Used when enough genotypes exist to support imputation w Accelerated by availability of sequence data at a lower cost l Evaluation of benefit from larger SNP sets as cost per SNP genotype declines l Application of genomics to more traits l Across-breed evaluation l Accounting for genomic pre-selection

48. WiggansCornell University, ANSC 3310, March 10, 2015 (48) Conclusions l Genomic evaluation has dramatically changed dairy cattle breeding l Rate of gain has increased primarily because of large reduction in generation interval l Genomic research is ongoing w Detect causative genetic variants w Find more haplotypes that affect fertility w Improve accuracy

49. WiggansCornell University, ANSC 3310, March 10, 2015 (49) Questions?