200 7 Paul VanRaden, George Wiggans, Jeff O’Connell, Paul VanRaden, George Wiggans, Jeff O’Connell, John Cole, John Cole, Animal Improvement Programs Laboratory Tad Sonstegard, and Curt Van Tassell Bovine Functional Genomics Laboratory USDA Agricultural Research Service, Beltsville, MD, USA [email protected]200 9 Dairy Cattle Breeders Have Dairy Cattle Breeders Have Adopted Genomic Selection Adopted Genomic Selection
31
Embed
2007 Paul VanRaden, George Wiggans, Jeff O’Connell, John Cole, Animal Improvement Programs Laboratory Tad Sonstegard, and Curt Van Tassell Bovine Functional.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
2007
Paul VanRaden, George Wiggans, Jeff O’Connell, John Cole, Paul VanRaden, George Wiggans, Jeff O’Connell, John Cole, Animal Improvement Programs Laboratory Tad Sonstegard, and Curt Van TassellBovine Functional Genomics LaboratoryUSDA Agricultural Research Service, Beltsville, MD, USA [email protected]
2009
Dairy Cattle Breeders Have Dairy Cattle Breeders Have Adopted Genomic Selection Adopted Genomic Selection
Gordon Conference on Quantitative Genetics, Feb. 2009 (2) Paul VanRaden200
9
How’s Your Genome?How’s Your Genome?
Gordon Conference on Quantitative Genetics, Feb. 2009 (3) Paul VanRaden200
9
AcknowledgmentsAcknowledgments
Genotyping and DNA extraction:• USDA Bovine Functional Genomics Lab, U.
Missouri, U. Alberta, GeneSeek, Genetics & IVF Institute, Genetic Visions, and Illumina
Computing: • AIPL staff (Mel Tooker, Leigh Walton, Jay
Megonigal) Funding:
• National Research Initiative grants– 2006-35205-16888, 2006-35205-167012006-35205-16888, 2006-35205-16701
• Agriculture Research Service• Holstein and Jersey breed associations• Contributors to Cooperative Dairy DNA Repository
(CDDR)
Gordon Conference on Quantitative Genetics, Feb. 2009 (4) Paul VanRaden200
9
CDDR ContributorsCDDR Contributors
National Association of Animal Breeders (NAAB, Columbia, MO)• ABS Global (DeForest, WI)• Accelerated Genetics (Baraboo, WI)• Alta (Balzac, AB, Canada)• Genex (Shawano, WI)• New Generation Genetics (Fort Atkinson, WI)• Select Sires (Plain City, OH)• Semex Alliance (Guelph, ON, Canada)• Taurus-Service (Mehoopany, PA)
Gordon Conference on Quantitative Genetics, Feb. 2009 (5) Paul VanRaden200
9
Genomics TimelineGenomics Timeline
Event Year
Dairy DNA repository began 1992
Cattle genome sequenced 2004
58,000 SNP selected May 2007
Illumina SNP50 chip sold Dec 2007
Prelim. genomic predictions Apr 2008
Official genomic predictions Jan 2009
Gordon Conference on Quantitative Genetics, Feb. 2009 (6) Paul VanRaden200
9
SNP Edits and CountsSNP Edits and Counts
Illumina SNP50 BeadChip 58,336
Insufficient number of beads 1,389
Unscorable SNP 4,360
Monomorphic in Holsteins 5,734
Minor allele frequency <5% 6,145
Not in H-W equilibrium 282
Highly correlated 2,010
Used for genomic prediction 38,416
Gordon Conference on Quantitative Genetics, Feb. 2009 (7) Paul VanRaden200
9
Repeatability of GenotypesRepeatability of Genotypes
2 laboratories genotyped the same 46 bulls• About 1% missing genotypes per lab• Mean of 98% SNP same (37,624 out of
38,416)– Range across animals of 20 to 2,244 SNP missingRange across animals of 20 to 2,244 SNP missing
• Mean of 99.997% SNP concordance (conflict <0.003%)
• Mean of 0.9 errors per 38,416 SNP– Range across animals of 0 to 7 SNP conflictsRange across animals of 0 to 7 SNP conflicts
Gordon Conference on Quantitative Genetics, Feb. 2009 (8) Paul VanRaden200
9
Old Genetic TermsOld Genetic Terms
Predicted transmitting ability and parent average• PTA required progeny or own records• PA included only parent data• Genomics blurs the distinction
Reliability = Corr2(predicted, true TA)• Reliability of PA could not exceed 50%
because of Mendelian sampling• Genomics can predict the other 50%• Reliability limit at birth theoretically 99%
Gordon Conference on Quantitative Genetics, Feb. 2009 (9) Paul VanRaden200
9
New Genetic TermsNew Genetic Terms
Genomic vs. pedigree relationships • Expected genes in common (A)• Actual genes in common (G)• Several formulas to compute G• Wright’s (1922) correlation matrix or
Henderson’s (1976) covariance matrix Genomic vs. pedigree inbreeding
• Correlated by 0.68
Daughter merit vs. son merit (X vs. Y)
Gordon Conference on Quantitative Genetics, Feb. 2009 (10) Paul VanRaden200
9
Differences in Differences in GG and and AA GG = genomic and = genomic and AA = pedigree relationships = pedigree relationships
Detected clones, identical twins, and duplicate samples
Detected incorrect DNA samples
Detected incorrect pedigrees
Identified correct source of DNA by genomic relationships with other animals
Gordon Conference on Quantitative Genetics, Feb. 2009 (11) Paul VanRaden200
Proposed by Nejati-Javaremi, Smith, Gibson, 1997 J. Anim Sci. 75:1738
Nonlinear regression, haplotyping or only slightly more accurate
Gordon Conference on Quantitative Genetics, Feb. 2009 (12) Paul VanRaden200
9
Worldwide Dairy GenotypingWorldwide Dairy Genotypingas of January 2009as of January 2009
Countries Animals
United States and Canada 22,344
France 8,500
Netherlands, New Zealand1 6,000
New Zealand and Ireland 4,500
Germany 3,000
Australia 2,000
Denmark, Finland, Sweden 2,0001Using a customized Illumina 50K chip (different markers)
Gordon Conference on Quantitative Genetics, Feb. 2009 (13) Paul VanRaden200
9
PhenotypesPhenotypes
26 traits plus the Net Merit index
The 6,184 bulls genotyped have >10 million phenotyped daughters (average 2,000 daughters per bull)
Most traits recorded uniformly across the world
Foreign data provided by Interbull
Gordon Conference on Quantitative Genetics, Feb. 2009 (14) Paul VanRaden200
9
Genotyped Animals (n=22,344)Genotyped Animals (n=22,344)In North America as of February 2009In North America as of February 2009
0
500
1000
1500
2000
2500
1950
1970
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
Year of Birth
Nu
mb
er
of
An
ima
ls Predictor
Predictee
Young
Gordon Conference on Quantitative Genetics, Feb. 2009 (15) Paul VanRaden200
9
Experimental Design - UpdateExperimental Design - UpdateHolstein, Jersey, and Brown Swiss breedsHolstein, Jersey, and Brown Swiss breeds
HOL JER BSW
Predictor:
Bulls born <2000 4,422 1,149 225
Cows with data 947 212
Total 5,369 1,361 225
Predicted:
Bulls born >2000 2,035 388 118
Data from 2004 used to predict independent data from 2009
Gordon Conference on Quantitative Genetics, Feb. 2009 (16) Paul VanRaden200
9
Reliability GainReliability Gain11 by Breed by BreedYield traits and NM$ of young bullsYield traits and NM$ of young bulls
Trait HO JE BS
Net merit 24 8 3
Milk 26 6 0
Fat 32 11 5
Protein 24 2 1
Fat % 50 36 10
Protein % 38 29 5
1Gain above parent average reliability ~35%
Gordon Conference on Quantitative Genetics, Feb. 2009 (17) Paul VanRaden200
9
Reliability Gain by BreedReliability Gain by BreedHealth and type traits of young bullsHealth and type traits of young bulls
Trait HO JE BS
Productive life 32 7 2
Somatic cell score 23 3 16
Dtr pregnancy rate 28 7 -
Final score 20 2 -
Udder depth 37 20 3
Foot angle 25 11 -
Trait average 29 13 N/A
Gordon Conference on Quantitative Genetics, Feb. 2009 (18) Paul VanRaden200
9
Value of Genotyping More AnimalsValue of Genotyping More AnimalsActual andActual and predicted gains predicted gains for 27 traits and for Net Meritfor 27 traits and for Net Merit
Bulls Reliability Gain
Predictor Predicted NM$ 27 trait avg
2130 261 13 17
3576 1759 23 23
4422 2035 24 29
6184 7330 31 30
Cows:
947
1916
Gordon Conference on Quantitative Genetics, Feb. 2009 (19) Paul VanRaden200
9
SimulationSimulation Results ResultsWorld Holstein PopulationWorld Holstein Population
40,360 older bulls to predict 9,850 younger bulls in Interbull file
50,000 or 100,000 SNP; 5,000 QTL
Reliability vs. parent average REL• Genomic REL = corr2 (EBV, true BV) • 81% vs 30% observed using 50K• 83% vs 30% observed using 100K
Gordon Conference on Quantitative Genetics, Feb. 2009 (20) Paul VanRaden200
9
Marker Effects for Net Merit Marker Effects for Net Merit
Gordon Conference on Quantitative Genetics, Feb. 2009 (21) Paul VanRaden200
9
Significance Tests are StupidSignificance Tests are Stupid
Gordon Conference on Quantitative Genetics, Feb. 2009 (22) Paul VanRaden200
9
Insignificant SNP EffectsInsignificant SNP Effects
Traditional selection on PA• 50 : 50 chance of better chromosome
1 SNP with tiny effect• 50.01 : 49.99 chance
38,416 SNPs with tiny effects• 70 : 30 chance
Only test overall sum of effects!
Gordon Conference on Quantitative Genetics, Feb. 2009 (23) Paul VanRaden200
Gordon Conference on Quantitative Genetics, Feb. 2009 (24) Paul VanRaden200
9
Net Merit by Chromosome for O-ManNet Merit by Chromosome for O-ManTop bull, +$778 Lifetime Net Merit Top bull, +$778 Lifetime Net Merit
-40
-20
0
20
40
60
80
X 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Chromosome
NM
$
NM$
Gordon Conference on Quantitative Genetics, Feb. 2009 (25) Paul VanRaden200
9
Progeny Tested BullProgeny Tested BullO-ManO-Man
Semen sales ~200,000 units / year Semen price $40 / unit Income > $5 million / year 40,144 daughters already milking
• 29,811 in United States• 1,963 in France, 1,895 in Denmark,
1,716 in Italy, 839 in Holland, etc.
Gordon Conference on Quantitative Genetics, Feb. 2009 (26) Paul VanRaden200
9
O-Man Daughters O-Man Daughters vs. Average Cowsvs. Average Cows
TraitO-Man
daughterAverage Holstein
Milk (gallons/day) 10.4 10.0
Protein (lbs/day) 2.78 2.58
Cell count (1000/ml) 205 262
Productive life (mo) 33.8 27.7
Pregnancy rate (%) 25.7 23.1
Calving difficulty (%) 3% 8%
Gordon Conference on Quantitative Genetics, Feb. 2009 (27) Paul VanRaden200
9
Genomic Tested BullsGenomic Tested BullsAvailable Jan 2009Available Jan 2009
Age (yrs) Reliability Net Merit
Freddie 4 69 918
Al 1 67 914
Russell 1 65 854
Alan 1 68 841
O-Man 10 99 778
Gordon Conference on Quantitative Genetics, Feb. 2009 (28) Paul VanRaden200
9
Adoption of Genomic TestingAdoption of Genomic TestingUS young bulls purchased by AI companiesUS young bulls purchased by AI companies
Birth Year
Bulls Sampled
Bulls Tested
Genomic Tested %
2008* 193 166 86
2007* 1455 896 62
2006 1657 717 43
2005 1642 818 50
2004 1638 742 45
* 2007-2008 counts are incomplete
Gordon Conference on Quantitative Genetics, Feb. 2009 (29) Paul VanRaden200
9
Genetic ProgressGenetic Progress
Assume 60% REL for net merit• Sires mostly 1-3 instead of 6 years old• Dams of sons mostly heifers with 60% REL
instead of cows with phenotype and genotype (66% REL)
Progress could increase by >50%• 0.37 vs. 0.23 genetic SD per year• Reduce generation interval more than
accuracy
Gordon Conference on Quantitative Genetics, Feb. 2009 (30) Paul VanRaden200
9
Low Density SNP ChipLow Density SNP Chip
Choose 384 marker subset• SNP that best predict net merit• Parentage markers to be shared
Use for initial screening of cows• 40% benefit of full set for 10% cost• Could get larger benefits using
haplotyping (Habier et al., 2008)
Gordon Conference on Quantitative Genetics, Feb. 2009 (31) Paul VanRaden200
9
ConclusionsConclusions
High accuracy requires very many genotypes and phenotypes
Most traits are very quantitative (few major genes)
Genomic reliability > traditional • 30-40% with traditional parent average• 60-70% using 8,100 genotyped Holsteins • 81-83% from 40,000 simulated bulls