Transcript Results from QTL analyses - Institute for Behavioral Genetics

QTL studies: past,
present & (bright?)
future
Overview
• A brief history of ‘genetic variation’
• Summary of detected QTL
– plants
– livestock
– humans
• Modelling distribution of QTL effects
• From QTL to causal mutations
• Three success stories
[Galton, 1889]
(early 1900s)
Inheritance of quantitative traits
Biometricians
vs.
Mendelians
(Pearson)
(Bateson)
The height vs. pea debate
Do continuously varying traits have the same
hereditary and evolutionary properties as
discrete characters?
Yes!
Trait
Qq
qq
t
m-a
QQ
m+d
m+a
[Fisher, Wright]
Multiple-factor hypothesis
• (Many) independently segregating loci
– Continuous (Gaussian) distribution of
genotypes
• Environmental variation
– ‘Regression towards mediocrity’ [Galton, 1889]
• trait in progeny is not the average of trait in parents
• R = h2 S
• Linear models & multivariate normality
– Livestock breeders [Henderson]
– BLUP(A)
Three bi-allelic additive loci
0.35
0.30
Frequency
0.25
0.20
0.15
0.10
0.05
0.00
-3
-2
-1
0
Genotype value
1
2
3
©Jeremy
Stockton
©Roslin
Institute
Lynch & Walsh (1998)
• Summary of 52 experiments (222 traits),
mostly from inbred founder lines
– in 45% of traits a QTL explaining >20% of
phenotypic variation
– in 84% of traits all QTLs explained >20% of
the phenotypic variation
– in 33% of traits all QTLs explained >50% of
the phenotypic variation
Reported QTL in pigs
• 15 experimental crosses
– N from 200 to 1000
[Bidanel & Rothschild 2002]
• multiple QTL for growth, fatness, carcass
traits and reproduction
• nearly all chromosomes covered
• QTL explain 3 to 20% of F2 variance
SSC
1
2
0
3
4
IGF2
A85k
A70k
5
6
7
8
9
10 11 12 13 14 15 16 17 18
T100k
A
A100k
A115k
A70k
A90k
20
A90k
X280k
T115k
A115k
A13w
L115k
A40k
R26w
40
A40k
A115k
A17w
Length (cM)
A40k
60
80
A90k
MC4R
L100k
L14w
L115k
100
L100k
A70k
A60k
A80k
A90k
S,M,
A80k
A110k
A115k
R 115k
L115k
A40k
A17w
RYR1
A90k
L14w
R100k
A95k
A22w
L115k
A100k
L100k
R115k
RN
L14w
F100k
R115k
M95k
A60k
A95k
120
A80k
A17w
A105k
A80
A90
HFAB
L115k
A115k
A13w
L14w
R14w
A17w
A13w
R14w
A70k
140
160
A,L,
T,F
T115k
A90k
Backfat thickness
Xyz : X = A (average), L (lumbar), R (last rib), T (tenth-rib), S (shoulder), M (mid-back), F (first-rib) backfat
thickness at xx kg (k) or xx weeks (w) of age; Locus names (in bold characters) : MC4R = melanocortin-4
receptor locus; IGF2 = insulin growth factor 2; RYR1 = ryanodine receptor locus ; HFAB = heart fatty acid
binding protein locus; PIT1 = regulatory factor locus; RN = “acid meat” locus.
[Bidanel & Rothschild 2002]
X
How many QTLs are there and
how many can we detect?
• Theory
– Distribution of effects & experimental sample
size (Otto & Jones, 2000)
• Data
– Model reported QTL effects from experiments
(Hayes & Goddard 2001)
[Otto & Jones 2000]
Potential distributions of allelic effects. Each curve describes a gamma distribution
with mean µ = 1 but with different coefficients of variation (C). The QTL underlying
a particular phenotypic difference represent draws from the appropriate distribution,
as illustrated by the circles under the x-axis. Only those QTL above the threshold of
detection (q = 0.8, thin vertical line) are likely to be detected (solid circles). Those
below the threshold are likely to remain undetected (open circles).
[Otto & Jones 2000]
The expected number of detected loci as
a function of the number of underlying
loci. The expected number of detected
loci is equal to n times the fraction of
the probability density function, g[x, µ,
C] given by (13), that lies above . It is
plotted as a function of the number of
underlying loci for a bell-shaped
distribution (C = 0.5; dot-dashed curve),
an exponential distribution (C = 1; solid
curve), and an L-shaped distribution (C
= 2; dashed curve). (A) q = 10% of D,
as was typical in our studies with a
large number of QTL and 200 F2's. (B)
q = 5% of D, as was typical in our
studies with a large number of QTL and
500 F2's.
Distribution of QTL effects in
livestock
5
Density
4
3
2
1
0
0
0.25
0.5
0.75
1
Effect (phenotypic SD)
1.25
1.5
[Hayes and Goddard, 2001]
Proportion of genetic variance
explained by QTLs
Prop. variance explained by QTL
above this size
100
90
80
70
60
50
40
30
20
10
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Size of QTL (phenotypic SD)
0.8
0.9
1
[Hayes and Goddard, 2001]
From QTL to gene
• Paradigm
–
–
–
–
Linkage
Fine-mapping (IBD/LD)
Association
Function
Positional Cloning of Complex Traits
Genetics
Sib pairs
Chromosome Region
Association Study
Genomics
Physical Mapping/
Sequencing
Candidate Gene Selection/
Polymorphism Detection
Mutation Characterization/
Functional Annotation
Identified causal polymorphisms
• 41 (< March 2004)
– 31 in mammals
• 17 outbred populations
– 14 in humans
– 2 in pigs (RN, IGF2)
– 1 in dairy cattle (DGAT1)
• Few ‘proven’ with functional assays or
through transgenics
[Korstanje & Piagen 2001; Glazier et al. 2002]
Identified QTLs in mammals
[Korstanje & Piagen 2002]
[Korstanje & Piagen 2002]
[Glazier et al. 2002]
Botstein & Risch (2003), Nature Genetics
Is the nature of genetic variation for quantitative traits
different???
Three success stories of QTL
identification in farm animals
• IGF2 in pigs
• DGAT in dairy cows
• Callipyge in sheep
Van Laere et al. (2003).
Nature 425:832-836
• QTL Linkage peak on chr. 2p for muscle
mass
– Wild Boar x Large White cross
– Pietran x Large White cross
• IGF2 = candidate
– IGF2 is paternally imprinted in mice and man
• QTL = paternally imprinted
– Sire’s allele expressed
[Nezer et al. 1999; Jeon et al. 1999]
Effects etc.
• Wild boar cross
– 20-30 % of variance explained
– ~3% difference in Lean Meat %
• Pietran cross
– ~2% difference in % Lean Cuts
– ~5 mm difference in backfat
• Confidence interval ~4 cM (= small!!!)
• No sequence variants in coding parts of IGF2
could explain the observed effects
[Nezer et al. 1999]
Fine-mapping using haplotype
sharing (Nezer et al. 2003)
• Marker-assisted segregation analysis
– Assume bi-allelic QTL
– Assume that ‘favourable’ allele Q appeared by mutation
or migration ~50-100 years ago
– Assume known effect (2% of ‘lean cuts’)
– Determine QTL genotype status of 20 boars
– Look for shared haplotype on Q chromosomes
• Identified shared haplotype of ~250 kb
– Contained 2 paternally imprinted genes (INS and IGF2)
Qq boars
Q
q
QQ or qq boars
Genotype deduced
From Qq haplotypes
All Q chromosome share a 90 kb common haplotype not
present on q chromosomes
[Nezer et al. 2003]
Resequencing 3 Q and 8 q chromosomes for 28.5 Kb
spanning INS-IGF2 identifies 33 putative QTN
[M. Georges]
Resequencing a heterozygous, non-segregating
Hampshire sire identifies a recombination excluding
TH-IGF2(I1) (- 9 candidate QTN)
[M. Georges]
Resequencing a heterozygous, non-segregating
Large White x Meishan sire identifies the QTN
[M. Georges]
TH
INS
14
12 3
SWC9
IGF2
Genes
1
2
3
4a 4b
5
6
7
8
9
%(G+C)
P208 (ref.)
CpG
DMR1 island
LW3
Q
LRJ
H205
H254
M220
LW1224
LW1461
LW209
q
LW419
LW197
EWB
LW33361
LW463
JWB
Pig-q
Pig-Q
Human
Mouse
AGCCAGGGACGAGCCTGCCCGCGGCGGCAGCCGGGCCGCGGCTTCGCCTAGGCTCGCAGCGCGGGAGCGCGTGGGGCGCGGCGGCGGCGGGGAG
.......................................................A......................................
....G.....T.......T.C...T...G..TC...............................AG...A.........A.T....AG......
...T.........T......C.......T...T....C..A................G...TCT...............A.G............
QTN is guanine to adenine substitution in IGF2-intron3 nucleotide
3072
Van Laere, Fig. 1A
DGAT in dairy cows
• Genome scan suggested QTL for fat% in
milk on chromosome 14
• IBD fine-mapping reduced region to 3 cM
• Association / linkage disequilibrium
identifies causative mutation
• Mutation is an amino acid changing SNP in
the DGAT1 gene
There are large QTL out there!
QTL explains > 50% (!) of genetic variance in fat%
QTL allele is common
QTL acts additively
Callipyge mutation in sheep
(major gene, not QTL)
Gene action: “Polar overdominance”
[1st allele from dad 2nd from mum]
[Freking et al. 1998]
Callipyge summary
• Gene action impossible to work out without
genetic markers
• Causal mutation is non-coding
• How common is imprinting for QTL?
[Glazier et al. 2002]