Transcript Slide 1

Multiple-Trait
Selection in a
Single-Gene World
David Notter
Department of Animal and Poultry
Sciences
Virginia Tech
Genetic Markers and NCE
• Genetic Markers have the potential to
improve the effectiveness of NCE
• However, for most traits, genetic markers
will not account for enough of the genetic
variation to allow them to be used as the
only selection criterion
• Instead, methods must be developed to
combine information on genetic markers
with performance data
Types of Marker-Assisted Selection
• Gene-Assisted Selection (GAS)
– A DNA sequence variant exists within the gene
– May be the actual causal mutation or just
associated with it
• Linkage-Disequilibrium MAS (LD-MAS)
– Marker is not a part of the gene, but is very
tightly linked with the favorable form of the QTL
• Linkage-Equilibrium MAS (LE-MAS)
– Marker is loosely linked to the QTL. The
association can differ among families (sires)
Garrick and Johnson, 2003
A GAS
Marker
A MAS
Marker
THE GENE
A GAS
Marker
A MAS Marker
LD
THE GENE
→
LE
Flanking
MAS Markers
THE GENE
Judging the Importance of a Marker
• The Size of Marker Effect
– How different are the different marker genotypes?
• Degree of Dominance of the Marker
– Are heterozygotes intermediate or do they resemble
one of the homozygotes?
• The Marker Frequency
– Is the marker common or rare within a breed?
• The Proportion of the Genetic Variance in
the Trait Accounted for by the Marker
– How much variation (opportunity for improvement)
exists independent of the marker?
The Marker Effect
Genotype
MM
Effect
+a
Marker
mm
-a
= 2a
Effect
For the GeneStar marbling (thyroglobulin) marker, the
difference in marbling score between homozygotes is
3.5 to 11% (Hetzel, 2003). In Angus cattle (AAA, 2004),
with a mean marbling score of about 6.0, this would give
2a = 0.48, or about one half of a marbling score.
Degree of Dominance of the Marker
Genotype
MM
Effect
+a
Mm
Mm
Mm
+a → Dominant marker
0 → Co-dominant marker
-a → Recessive marker
mm
-a
The GeneStar marbling marker is approximately
codominant in its effect on marbling
Genestar Genetic Marker
Percentage Choice
Percent Choice
80
70
60
57
60
64
50
40
30
No Stars
One Star
No. of "Stars"
Two Stars
Genetic Variance Accounted for by
a Codominant Marker
2
σ
2
h
M
A-M
=
=
2
2p(1-p)a
2
[2p(1-p)a ]
/
2
σ
P
• h2M is the marker heritability
• a is the marker effect
• p is the frequency of the marker
• σ2P is the phenotypic variance
Frequency of the Marker
p
0.50
h2
0.50
h2M
0.125
h2M = .25 h2 at p = .50
Frequency of the Marker
p
0.50
h2
0.50
h2M
0.125
0.75
0.48
0.100
h2M = .25 h2 at p = .50
Frequency of the Marker
p
0.50
h2
0.50
h2M
0.125
0.75
0.48
0.100
0.90
0.46
0.050
For GeneStar Marbling with p = 0.50, h2M ~ 0.04,
which accounts for 11% of the additive variance in
marbling score. Thus 89% is associated with
other, currently unidentified, genes.
Frequency of the GeneStar Marbling
Marker in Various Breeds
Breed
B. Angus
Frequency*
0.30
R. Angus
0.45
Simmental
0.29
Wagyu
0.65
* From Hetzel, 2003--Approximate
Overview of Issues Involved in
Marker Assisted Selection
1. The size of the effect: what is the
difference (2a) between individuals
homozygous for alternative marker alleles?
• Must be estimated and validated
2. The importance of the effect: what is the
economic effect of a change in marker
genotype?
3. The mechanism of gene action: is the
marker dominant, recessive, or
co-dominant?
Overview of Issues Involved in
Marker Assisted Selection
4. The importance of other genes: compare
the marker heritability (h2M) to the overall
h2 to determine the need for continued data
recording and gene discovery.
5. The frequency of the favorable marker:
•
•
•
Frequencies near 0.5 support the most rapid
and immediate improvement
High frequencies imply limited impact
Low frequency markers result in a lag period,
and have lots of potential, but raise concern
about loss of genetic diversity and impact on
other traits
Integrating Marker Information into
National Genetic Evaluations
• Genes and markers will continue to be
discovered
• Many will not be of general utility, but
some will be useful
• Comprehensive genotyping of many
animals may be possible but is not yet a
reality
• Partial genotyping of subsamples of
animals is more realistic for the immediate
future.
How Might Breed Associations
Respond?
• We are effectively being told that there is
something outside NCE that makes an
animals better or worse than his EPDs might
indicate
• Yet for proven sires, the EPD is a more
definitive predictor of progeny performance
and genetic worth
– Markers are valuable mainly for young
(unproven) animals, for traits not included in the
EPDs, or for traits that take a long time to
evaluate accurately
How Might Breed Associations
Respond?
• Explicitly identify the genes and markers of
interest to the breed
• Develop a DNA collection strategy
• Develop a genotyping strategy
• Develop validation strategies
• Incorporate marker information into NCE
How Might Breed Associations Respond?
Explicitly identify the genes and
markers of interest to the breed
• Identify the known genes and LD markers
of interest to the breed
• Might also identify a set of informative
microsatellite markers for use in gene
discovery
• This will be an evolving array, but provides
guidance for the genes and markers that
will be supported in NCE
How Might Breed Associations Respond?
Develop a DNA collection strategy
• Evaluate simple techniques for DNA
acquisition and physical storage:
fluoroacetate papers, hair, etc.
• Don’t extract DNA until you need it.
• Capacity for repeated extractions.
• Identify high-priority animals, but don’t
necessarily rule out storage of (for
example) blood on all registered or
performance-recorded animals
Blood samples on Perforated
FTATM Cards
How Might Breed Associations Respond?
Develop a genotyping strategy
• Breed associations need to ensure access
to genotypes on their animals and become
repositories for those genotypes
• “Multiplex” genotyping capacity is needed to
allow efficient genotyping of individual
animals for many genes/markers
• Develop a genotyping plan for high-use
(“legacy”) sires, and perhaps samples of
their calves (i.e., to screen for segregating
markers)
How Might Breed Associations Respond?
Develop validation strategies
• New markers must be validated to
determine if initial results are repeatable
• New markers must be validated in different
breeds
• Markers must be validated for both the
primary trait and for correlated traits
• Genotyping strategies can be designed to
support validation strategies
How Might Breed Associations Respond?
Incorporate marker information
into NCE
• We know very little about how this will happen!
• We do know that marker information will
continue to evolve—we will always be behind!
• Must be able to continuously incorporate new
markers into NCE
• Marker information will enhance, but certainly
not replace, performance data and EPDs
How Might Breed Associations Respond?
Take Control of the Use of Genetic
Markers in NCE
• Knowledge and resources to allow
breeders and their organizations to impact
marker detection and development
• Rapid evaluation of frequencies of new
genetic variants and markers
• Rapid and efficient validation of newly
proposed markers
How Might Breed Associations Respond?
Issues in the Incorporation of
Marker Information into NCE
• Are marker effects “fixed” or “random”?
• What is the genetic base for a marker effect?
• What are the effects of a marker on other
traits? How do we estimate these accurately?
• How do we validate marker effects in different
environments and management systems?
• How do we check if a marker “stops working”?
• How to handle animals that are not genotyped
or genotyped for only a few markers?
Using Markers in NCE
Are the marker effects “fixed” or
“random”?
• If fixed, then we make a constant adjustment to
the EPD based on marker genotype
• If random, then even if we know the genotype
exactly, we still hedge our bets to allow for
recombination, interactions of the marker with
the environment or the background genotype,
or other unknown variations in the gene
• But HOW do we hedge? Estimate sire x
marker or marker x environment interaction
variances?
A GAS Marker can, at least
hypothetically, be considered a
fixed effect
(but somehow it seems too simple!)
A GAS
Marker
THE GENE
Genotype
MM
Mm
Mm
EPD
+α
d
-α
A GAS
Marker
THE GENE
Other, unknown sequence variants
could be present in some animals and
invalidate the effects of the known marker.
We need to prepare ourselves for things
like this!
LE-MAS Markers effectively must be considered
random effects to allow for recombination
LE-MAS Markers
must be reconfirmed
in such sires, which
appears likely to
limit their appeal
???
Using Markers in NCE
What is the genetic base for a
marker effect?
• Depends on, and changes with, the frequency
of the marker
• As marker approaches fixation, the favorable
form becomes less and less useful.
Using Markers in NCE
What are the effects of a marker
on other traits? How do we
estimate these accurately?
• Major validation issue
• We will immediately credit an animal for the
known, favorable marker effect, but only
slowly identify that animal as possibly inferior
for correlated effects
• Linkage with performance records is
mandatory, as is adequate genotyping of
offspring of both sexes
Using Markers in NCE
What about animals that are not
genotyped or are genotyped for
only a few markers?
• Many animals will likely not be genotyped
• We will therefore need to infer the possible
genotypes of such animals using the
genotypes of their relatives
• Thallman has developed one methodology to
accomplish this—there may be others
Conclusions
• The search for markers will continue
• The bovine gene map will accelerate the
search for and the rate of discovery of
genetic markers
• BIF is facing a developmental effort to use
these DNA technologies that may rival the
implementation of BLUP EPDs
• The BIF Guidelines are going to get
thicker again!