USDA-ARS US Meat Animal Research Center

Transcription

USDA-ARS US Meat Animal Research Center
Tim Smith
USDA Agricultural Research Service
U.S. Meat Animal Research Center
Clay Center, Nebraska
ICOMST Cape Town, South Africa – August 12th, 2008
1:
2: Bos primigenius
6
4
3
11
7
5: Banteng
6: Bali cattle
2
7: Gaur
9
1
12
10
Bo
s
10: Yak
us
g
a
eph
11: Domestic yak
Po
Bison
Pliocene
8: Mithun
9: Kou-prey
Bibos
Miocene
3: Zebu
4: Taurine cattle
8
5
Eotragus
Bos namadicus
Pleistocene
13
12: Wisent
13: Bison
Holocene - recent
Adapted from M. Felius © 1995
CATTLE BREEDS: An Encyclopedia
Modern cattle derive from several domestications of strains of Aurochs
Cave painting of Aurochs ca. 14,000 BC
Aurochs drawing ca. 1885
“Domestication” involves selection for docility, manageability
The Industrial Revolution and associated urbanization drove specialization
The “ideal” Durham ca. 1819
The “most perfect” North Devon ca. 1835
First principle of genetics was known as early as mid-17th century :
hindquarters inherited from cow, forequarters from bull
Selection for visible traits led to establishment of breeds
“Breeds of cattle” chromolithograph ca. 1879
Phenotypic selection works best for easily measured traits
Most quickly captures variation with large effect
Quantitative Genetics (mid-20th century)
progeny testing
Using statistical theory and performance of progeny to estimate
genetic merit (Expected Progeny Difference, EPD)
Quantitative Genetics works !
Accommodates selection to accumulate multiple variations of small effect
Represents average over all loci contributing to trait
Major phenotypic changes usually have a downside
extreme muscling affects calving ease
extreme milk yield affects
milk quality and reproductive
rate
Classical Quantitative Genetics is animal-centric
Animal 1: EPD = + 8.0
Animal 2: EPD = ??
Estimation on one animal does not tell you anything about another
( if they are only distantly related )
Goals of developing DNA technology
Original goals:
Increase selection accuracy
Reduce cost of progeny testing
Target difficult / expensive to measure traits
Added goals:
Facilitate animal management decisions
DNA markers aim to achieve transportability
Animal 2: EPD = fx(+8.0)
Animal 1: EPD = + 8.0
Genotype: A
Genotype: A
fx
(% of total genetic effect due to A)
Markers that show this behaviour
indicate presence of QTL
If two animals share common marker set(s), a function can be
derived to estimate EPD in the absence of progeny data
Quantitative trait loci (QTL):
Qualitative traits vs. Quantitative traits
Height is a Quantitative trait
“double muscling” is a qualitative trait
(caused by variations in myostatin gene)
milk yield is a quantitative trait
(results from variation at many loci)
Quantitative trait loci (QTL):
Genomic regions harboring DNA sequence variation having
moderate effect on phenotype
Quantitative trait nucleotide (QTN):
Specific DNA sequence variation causing phenotypic effect
Detecting QTL by “gene mapping”
QTL detection
Approximate position of variation
Large half-sib populations, genome scans
of a few hundred markers track sire allele
segregation
Result of QTL study
Our founder animal carries two alleles functionally
different at each identified QTL (frequently misses QTL
of small effect and overestimates effects of QTL
detected)
Resolution: 5-20 cM ( @ 500 animals )
Limited by:
effect size
number of meioses (recombination)
More markers doesn’t really help !!
GENOME SCAN RESULTS IN FOUR RESOURCE FAMILIES
Pied x
Angus
Belg Blue
x MARCIII
BTA1
BTA2
BTA3
MSTN, MSTN
RPYD
FATYD
MARB, RPYD
MARB
BTA6
BWT, W365,
HCW, LMA
BTA8
BTA4
WBS14
Brahman x Angus
Brahman x Hereford
BTA5
RPYD, FATYD
MARB, RPYD
MARB, RPYD
BTA15
HCW, WBS3
BTA27
FAT, YG, RPYD, WBS14
RIBBONE, DP, RIBFAT, BWT
BTA29
FAT, MARB
FAT
WBS14
MARB
WBS3, WBS14
QTL results are limited in their ability to achieve transportability
Animal 1: EPD = + 8.0
QTL alleles
A, C, D
Animal 2: EPD = ??
QTL alleles
A, C, D
Not sure if phase is the same and if
underlying variation segregating
To use QTL data: fine mapping to obtain markers with predictive merit
Example: meat tenderness QTL on chromosome 29 (BTA29)
μcalpain,
CAPN1
Q
T
L
Bovine CAPN1 gene
mu-calpain
Compared the gene sequence of CAPN1 in a panel of cattle from 8 beef breeds
Identified > 164 SNP among the different cattle
Thoroughly tested some of these SNP for association with meat tenderness
using three populations (approx. 500 animals each) specifically designed to
test for marker association
Found two markers (314 and 4751) that consistently associate with more
tender meat
Unlikely these are causative markers (QTN), but their consistent association
indicates they can be predictive of genetic merit
Predictive markers: genotype is correlated with genotype of QTN
r2 -- the squared correlation between two loci
Essentially, the degree to which the genotype of a marker at position “A”
predicts the genotype of a marker at position “B”
When r2 = 1 , perfectly predictive
When r2
0 , randomly correlated
Note: r2 between A and B may vary between populations
Uses of markers with predictive merit
Marker assisted selection (MAS):
marker information improves EBV accuracy
reduces number of progeny required for progeny test
reduces number of generations to fix favorable alleles
permits management of non-additive alleles
Marker assisted management (MAM):
marker information used to sort animals by genetic potential
markers used to match cows to bulls for maximum gain
Need to explain more of the variation than at present
Search for QTN or markers with high r2 with QTN
Genome scan
QTL
fine mapping is one way
Genome Wide Association (GWA) is another way:
assay tens of thousands of markers across whole genome at once
Genome-Wide Association (GWA)
Example: Wellcome study using 500K chip (about 400,000 useful SNP)
Identified loci associated with:
coronary artery disease
Crohn’s disease
rheumatoid arthritis
type I diabetes
type II diabetes
Requirements for GWA
Need to have markers that capture enough information about the genome
to track the majority of functional variation
Goal is to have markers in each “Block” of Linkage Disequilibrium (LD)
LD blocks of about 100 kb in cattle genome achieve r2 around 0.3
3 Gb / 0.1 Mb = 30,000 LD blocks
Goals for a bovine SNP array
Achieve > 30,000 SNP
Spread evenly across genome
Highly informative across cattle breeds and populations
(average minor allele frequency [MAF] >20%)
Cattle SNP-chip Collaboration
• USDA-ARS U.S. Meat Animal Research Center
• USDA-ARS Beltsville Agricultural Research Center (BARC)
• University of Missouri
• Illumina / Solexa
Illumina iSelect™ assay (60,800 bead types)
-- hope for about 53,000 useful SNP
SNP Content on the Chip
SNPs With MAF (68%)
Reduced Representation Library
25,125
Bovine Hapmap Consortium
12,641
Parentage Markers
118
Others (UA-IFASA,US-MARC,DPI)
60
Insilico SNPs (32%)
Assembly SNPs
Inter Breed
BACend Derived
INRA, DIAS
10,075
6,200
1,484
310
Total assays
56,947
Failed to produce genotype
2,939
Total successful assays on chip
54,008
Monomorphic in all animals tested 2,622
Total number informative SNP
51,386
Gap Distribution of SNPs along chromosomes
0.18
0.16
Illumina BovineSNP50
0.12
Affymetrix Targeted Genotyping
Bovine 25K SNP Panel
0.1
0.08
0.06
0.04
0.02
362500
342500
322500
302500
282500
262500
242500
222500
202500
182500
162500
142500
122500
102500
82500
62500
42500
22500
0
2500
Frequency
0.14
QC of chip: animals of variety of breeds
Bos taurus
Bos Indicus
Holstein
64
Brahman
Angus
62
Nellore
Limousin
44
Gir
Hereford
31
Jersey
28
African
Charolais
26
N'Dama
Brown Swiss
24
Sheko
Peidmontese
24
Ramagnola
24
Composite
Guernsey
21
Beef Master
Norwegian Red 21
Santa Getrudis
Red Angus
15
Gelbvieh
Only 3 of these two breeds
Simmental
3
25
24
24
25
20
24
24
Out Group
Bubalus depressicornis
Bos gaurus
Bos bison
Bos javanicus
Syncerus caffer
Bos grunniens
4
4
4
2
2
2
Assay utility: MAF by Breed
(all 54,008 markers)
MAF of composite / tropically adapted cattle
Performance in outgroups
Breed
% at least one genotype call
% Heterozygous
Lowland Anoa
96.92
5.18
Gaur
97.72
4.58
North American Bison
97.14
3.10
Banteng
96.86
2.52
Cape Buffalo
86.42
2.24
Yak
96.65
1.75
Goals for a bovine SNP array
Achieve > 30,000 SNP
Spread evenly across genome
Highly informative across cattle breeds and populations
(average minor allele frequency >20%)
Can we detect previously known variation via GWA using
bovSNP50?
Bob Schnabel, University of Missouri
QTL on BTA 14 for FAT % in milk
Due to amino acid substitution in bovine DGAT1 gene
(diacylglycerol acyltransferase)
Trial: Analyze BTA14 markers from the first 2,000 dairy cattle run with chip
• 64 sire families
• 1,843 sons (avg 29 sons/sire)
• 1,340 markers (1,217 used)–MAF ≥ 0.03
AND ≥ 95% complete genotypes
WGA scan of BTA 14 for FAT% in milk
Extreme close-up of peak region of association
BTA 14 (
)
Genome Wide Association population
Large population of small family size, test for
association of marker and phenotype
Uses “historical” recombination events to localize position of variation
More recombination = higher resolution = more markers is better
WGA has identified large number of marker-phenotype
associations for numerous production traits from meat
quality to reproduction
Problem 1: what is effect of selection for a marker / trait
on non-target traits?
Problem 2: How to apply marker information
in comprehensive way?
Whole Genome Selection (WGS)
Can use historical records and genotypes to develop methods and
determine accuracy
Using WGS to predict merit in dairy animals (BARC)
Paul van Raden, George Wiggans, Curt van Tassell
Beltsville Agricultural Research Center (BARC)
Genotypes of 2130 Holstein bulls with high accuracy EPD (born
before 1997 and having a high number of daughters) used to “train”
the prediction algorithm for desirable haplotypes
The training set used to predict performance of younger bulls (born
2001, but having extensive progeny test data)
WGS increases reliability of merit predictions
WGS in beef cattle
The “2000 bull” project
Genotyping industry bulls including some used in USMARC herds
-- Provides connection of our data to industry herds
-- Determines if WGS results can be moved across breeds
Summary
1. Selection in cattle generally makes use of old or ancient
variation
2. Advent of the bovSNP50 chip promises leap forward
3. High density SNP analysis may obviate progeny testing
Limitations
1. Can only make use of extant variation
2. bovSNP50 has limited application outside Bos taurus
3. Higher SNP density may be necessary to proceed to QTN
4. Still need to develop efficient methods to bring results to
beef cattle producers
5. Intensive selection reduces diversity
“Personal genomics”
23andMe:
$1,000 scan of your DNA for 550,000 SNP
30andMoo
Intelligent management decisions from informed genomics
8,000 B.C. Cattle domesticated in Africa, Mideast and Asia
6,000 B.C. Migration to Europe
15th century: development of specialized types
2009: 30andMoo introduces Individualized Genome Management for beef cattle
Unlock the potential of your herd. Today.
1493: first domesticated cattle introduced to the New World
18th century: beginning of modern breed development
Welcome to 30andMoo, a web-based service that helps you
read and understand the genetic potential of your beef and
dairy cattle. Simply provide a tissue sample of blood or
saliva with an easy-to-use kit and determine parentage,
establish farm-to-fork traceback, predict growth potential,
examine breed composition, or estimate genetic merit for a
wide variety of cattle production traits from reproduction to
meat quality.
What’s new at 30andMoo
June 3, 2009: 30andMoo introduces new Whole Genome
Selection feature, enhances search and breed registry
database features, and adds new Gene Journal content for
arthropod resistance.
What is the true breed
background of your
prize bull?
What does your bull’s
DNA indicate about
performance?
Who are the
correct sire and
dam for your
new calf?
DPR:
+ 4.8
Growth rate:
+ 3.6
Marbling:
- 1.4
Tenderness:
- 0.8
Cavling ease:
- 2.2
% milk fat:
+ 1.2
Feed efficiency: + 0.1
Incorporate genomic
information into
genetic evaluation via
genome-enabled EBV
Acknowledgements
USMARC
BARC
Mark Allan
Lakshmi Matukumalli
Tad Sonstegard
Paul van Raden
Curt van Tassell
George Wiggans
Renee Godtel
Larry Kuehn
Bob Lee
Tara McDaneld
Steve Simcox
Kevin Tennill
Warren Snelling
U. of Missouri
Bob Schnabel
Jerry Taylor
Illumina
Cindy Lawley-Taylor
Marylinn Munson
Identification of QTN advantageous, but may not be
all that is required
C313Y
Asn-Tyr-Cys-Ser-Gly-Glu-Tyr-Glu-Phe-Val-Phe
Piedmontese
Normal
aat tac tgc tct gga gaa tAt gaa ttt gta ttt
aat tac tgc tct gga gaa tGt gaa ttt gta ttt
Asn-Tyr-Cys-Ser-Gly-Glu-Cys-Glu-Phe-Val-Phe
Amino acid
313
A/A
C313Y genotypes
predict phenotype in
Piedmontese
A/A
A/G
G/G
A/G genotypes do not
predict phenotype in
Belgian Blues
WT
BB
11 bp
deletion
A/G
Expect to see many more predictive
markers in near future
Keys to good predictive markers:
Thorough testing
Desirable allele not at extremely low (difficult to apply) or
extremely high (not worthwhile) frequency
5% chromosomes have the
desirable allele – difficult or
slow to increase frequency
without over-reliance on
single bull or small set of
bulls
95% of chromosomes have the
desirable allele – not worth the
genotyping investment to
identify the small number of less
desirable animals