1	Research Plans for Genomics, Crossbreeding, Fertility, etc.
2	AIPL 5-Year Plan 2007-2012 Objectives Collect genotypes, new phenotypes Document current status and effects of management on dairy traits Improve accuracy of predictions by including SNP data, refining models Estimate economic values of traits to maximize lifetime profit
3	Genomic Goals Predict young bulls and cows more accurately Compare actual DNA inherited Use exact relationship matrix G instead of expected values in A Trace chromosome segments Locate genes with large effects
4	How Related are Relatives? Example: Full sibs are expected to share 50% of their DNA on average may actually share 45% or 55% of their DNA because each inherits a different mixture of chromosome segments from the two parents. Combine genotype and pedigree data to determine exact fractions
5	Genomic Relationships Measures of genetic similarity A = Expected % genes identical by descent from pedigree (Wright, 1922) G = Actual % of DNA shared (using genotype data) T = % genes shared that affect a given trait (using genotype and phenotype) Best measure depends on use
6	QTL Relationship Matrix (T) Three bulls each +50 PTA protein. Are their QTL alleles the same? Possibly, but probably not. Bull A could have 10 positive genes. Bull B could have 10 positive genes, not on same chromosomes as bull A. Bull C could have 20 positive and 10 negative genes.
7	Genes in Common at One Locus
8	Alleles Shared by Sibs
9	Unrelated Individuals? No known common ancestors Many unknown common ancestors born before the known pedigree G = Z Z’ / number of loci Elements of Z are –p and (1 – p), where p is allele frequency Relationships in base = 0 +/- LD
10	Traditional Pedigree
11	Genomic Pedigree
12	Example of a SNP haplotype
13	SNP Pedigree
14	Haplotype Pedigree
15	Genotype Pedigree Count number of copies of second allele
16	Reliability from Full Sibs Marker and QTL positions identical, sib REL = 99%
17	Bulls to Genotype 58,533 SNP Project Choose HO bulls with semen at BFGL Genotype 1777 proven bulls Born 1994-1996 with >75% REL NM Plus 172 ancestor bulls born 1952-1993 Predict 500 bulls sampled later Born 2001 with >75% REL NM Include other bulls in gap years? Born 1997-2000 (proven) or >2002 (waiting)
18	Birth Years of Bulls to Genotype
19	Contributors of DNA 500 CDDR bulls to predict, born in 2001
20	Potential Results Simulation of 50,000 SNPs QTLs normally distributed, n = 100 Reliability vs parent average REL 58% vs 36% if QTLs are between SNPs 71% vs 36% if QTLs are located at SNPs (not likely) Higher REL if major loci and Bayesian methods used, lower if many loci (>100) affect trait
21	Reliability from Genotyping Daughter equivalents DE_Total = DE_PA + DE_Prog + DE_Y + DE_G DE_G is additional DE from genotype REL = DE_total / (DE_Total + k) Gains in reliability DE_G could be about 15 for Net Merit More for traits with low heritability Less for traits with high heritability
22	Genomic Computer Programs Simulate SNPs and QTLs Compare SNP numbers, size of QTLs Calculate genomic EBVs Use selection index, G instead of A Use iteration on data for SNP effects Form haplotypes from genotypes Not programmed yet
23	Computing Times Inversion including G matrix Animals² x markers to form G matrix Animals³ to invert selection index 10 hours for 3000 bulls, 50,000 SNPs Iteration on genotype data Markers x animals x iterations 16 hours for 1000 iterations
24	Distribution of Marker Effects
25	Linear vs Non-linear Models
26	All-Breed Model: Goals Evaluate crossbred animals without biasing purebred evaluations Accurately estimate breed differences Compare crossbreeding strategies Compute national evaluations and examine changes Display results without confusion
27	Methods All-breed animal model Purebreds and crossbreds together Relationship matrix among all Unknown parents grouped by breed Variance adjustments by breed Age adjust to 36 months, not mature Within-breed-of-sire model examined but not used
28	Data Numbers of cows of all breeds 22.6 million for milk and fat 16.1 million for protein 22.5 million for productive life 19.9 million for daughter pregnancy rate 10.5 million for somatic cell score Type traits are still collected and evaluated in separate breed files
29	Purebred and Crossbred Data USA milk yield records
30	Crossbred Cows with 1^st parity records
31	Number of Cows with Records (with > 50% heterosis; March 2007)
32	Number of Cows with Records (with > 50% heterosis; March 2007)
33	Crossbred Daughters Added for sires in top 10 NM$ within breed
34	Heterosis for Yield Traits Percent of Parent Breed Average
35	All-Breed Analyses Crossbred animals Now have PTAs, only 3% did before if in breed association grading-up programs Reliable PTAs from both parents Purebred animals Information from crossbred relatives More herdmates (other breeds, crossbreds) Routinely used in other populations New Zealand (1994), Netherlands (1997) USA goats (1989), calving ease (2005)
36	Unknown Parent Groups Look up PTAs of known parents Estimate averages for unknowns Group unknown parents by Birth year Breed Path (dams of cows, sires of cows, parents of bulls) Origin (domestic vs other countries)
37	All- vs Within-Breed Evaluations Correlations of PTA Milk
38	Display of PTAs Genetic base Convert all-breed base to within-breed bases (or vice versa) PTA_brd = (PTA_all – mean_brd) SD_brd/SD_HO PTA_all = PTA_brd (SD_HO/SD_brd) + mean_brd Heterosis and inbreeding Both effects removed in the animal model Heterosis added to crossbred animal PTA Expected Future Inbreeding (EFI) and merit differ with mate breed
39	All-Breed PTAs – March Test Run Genetic correlations mostly same JE increase .02 for PL and .01 for SCS BS decrease .01 for fat and SCS AY increase .01 for PL USA bulls in top 100 differ little Numbers are averages across all scales JE improve for SCS, fat (26 vs 25) JE decline for milk, protein (59 vs 62) BS decline for yield (10 vs 15) HO improve for yield (17 vs 16)
40	Jersey and Swiss PTAs Base cow means changed little Base cow SD changed little Top bulls for protein dropped by ~9 lbs, bottom bulls dropped by ~4 lbs in both breeds Unknown parent grouping, heterosis may be responsible
41	All-breed Trend Validation 85 tests, 6 were significant (.05) None significant for milk or SCS 1 of 15 for fat and for protein 2 of 15 for PL and for DPR Increase in DPR repeatability made trend more negative, helped tests
42	Daughter Pregnancy Rate Genetic trend on all-breed base
43	Assumed Effects – Other Traits Transmitting ability differences from Holstein
44	Merit of F₁ Holstein Crossbreds 2006 Merit Indexes
45	Later Generation Crosses
46	Three-Breed Crosses Butterfat yield of three breed crosses was greater than from their F₁ crossbred dams. Three breed crosses averaged 14,927 pounds of milk and 641 pounds of butterfat as 2-year-olds in 1947.
47	Crossbreeding Conclusions All-breed model accounts for: Breed effects and general heterosis Unequal variances within breed Implemented in May 2007 PTA converted back to within-breed bases, crossbreds to breed of sire PTA changes larger in breeds with fewer animals
48	Cow Fertility Research Daughter Pregnancy Rate works well, except that Other traits are evaluated by Interbull Other countries don’t use DPR in their indexes, and their calving interval data comes too late Synchronization changes traits
49	Emphasis on Fertility, Longevity (% of total merit)
50	Days Open Genetic Correlations Jorjani, 2005 Interbull Bulletin
51	DPR Results – March Test Run Holstein genetic correlations
52	Daughter Conception Rate Genetic Correlations Jorjani, 2005 Interbull Bulletin
53	Days to 1^st Insemination Genetic Correlations Interbull, May 2007
54	Fertility Trait Indexes % relative emphasis
55	Predict Longevity from Fertility Which cow fertility trait contributes most to longevity? Days to first insemination (DFI), or Non−return rate (NR) Combined longevity includes 23% DFI and 12% NR in CAN Only DFI in NLD Correlations = .33 DFI, .11 NR in USA
56	DPR - Top 100 bulls Born in last 12 years, March 2007 test run
57	Calving Interval Correlations with other traits in the same country
58	Conception Rate (Trait 4 correlations with other traits)
59	Calving to First Insemination (Trait 2 correlations with other traits)
60	Heifer Fertility (Trait 1 correlations with other traits)
61	Cow Fertility Conclusions Fertility and longevity receive a total of 8% to 40% of selection Fertility definitions not uniform Days to 1^st insemination is more important than conception rate? Selection for fertility reduces costs and increases longevity
62	Bull Fertility Research Dr. Melvin Kuhn Multiple services and an expanded service sire (SSR) term “Type” of model: Linear, Threshold Unconfirmed breedings: outcome not known with certainty Edits and Modeling of nuisance variables
63	Service Sire Effects SSR inbreeding Inbreeding of the Mating SSR age at mating Stud and Stud*year Additive genetic effect (very low heritability)
64	Results: Correlations
65	Linear/Threshold Model Conclusions to date: Little, if any, difference in predictions between the 2 models Use of a good estimate of std. dev. of the predictor in thr model probability calculations may improve thr model evaluations Threshold/Linear model is, at most and if anything at all, only a minor issue Linear model will likely be implemented because it is computationally faster, more reliable, and simpler
66	Sexed Semen (S) Matings 22,843 S-matings reported as of April 2007 92% are Holstein, most of remainder are Jersey 61% are on heifers (not eligible for ERCR) 69% are 1^st services 4,040 ERCR-eligible Holstein S-matings 398 bulls Only 2 bulls with at least 300 ERCR-eligible S-matings
67	Bull Fertility Summary Research on use of multiple services and an expanded service sire term is complete Linear/Thr model is, at most, of minor importance only for this trait; will likely implement linear model Expect to delete unconfirmed matings and treat those with positive preg ck as successes but impact will be evaluated Implementation expected January 2008
68	Test Day Model - Potential Benefits Increased accuracy of evaluations Account for lactation curve differences Account for genetic differences by parity Evaluate persistency, rate of maturity Include milk-only records if multi-trait Possible earlier selection of bull dams Promote as state-of-the-art system Management effects more accurate Could provide to DRPCs and herd owners