frequently asked questions in qtl mapping...1111111222222333333 444444555555666666 1111111222222...
TRANSCRIPT
The 7th Workshop on QTL Mapping and Breeding Simulation, 20-21 Jan., Mexico
Frequently Asked Questions in QTL Mapping
Jiankang WangJiankang WangCIMMYT China and CAAS
E-mail: [email protected] or [email protected]
1
E mail: [email protected] or [email protected]
Q1: What is LOD?Q1: What is LOD?
2
H th i t t i QTL iHypothesis test in QTL mapping
LOD > LOD0, T itiF l iti
LOD LOD
0i.e., accept Ha
True positives, no errors
False positives, Type I errors
False negati esLOD < LOD0, i.e., accept H0
True negatives, no errors
False negatives, Type II errors
H0: there is no QTL at a genomic
iti th
Ha: there is one QTL at the
i3
position on the trait in interest
genomic position
Likelihood ratio test (LRT)
Definition of LRT )ln(2 0
ALLLRT −=
Definition of LOD (likelihood of odd)A
L )log()log()log( 00
LLLLLOD A
A −==
Relationship between LOD and LRT
or61.4)10ln(2
LRTLRTLOD ≈= LODLRT 61.4≈
4
Q2: How to choose a Qthreshold value of LOD?
5
T i h th i t tTwo errors in hypothesis test
Type I error rate = P {Reject H0 |Type I error rate P {Reject H0 | True H0}
T II t P {A t H |Type II error rate = P {Accept H0 |False H0}False H0}
6
Significance level (α) in hypothesisSignificance level (α) in hypothesis test – The control of Type I error
Significance level for N times of gindependent tests: 1-(1- α)N
B f i dj t t / NBonferroni adjustment: ≈ α / N
Problem: Multiple and dependent tests exist in QTL mapping! Permutation test in QTL mappingPermutation test in QTL mapping
7
Choice of the threshold of LODChoice of the threshold of LOD
For one test:α (e.g., 0.1, 0.05, 0.01)N times of independent tests: 1-(1-α)N
Empirical LOD threshold for an overallEmpirical LOD threshold for an overall significance level of 0.05: 2.0 – 3.0
8
Q3: Which method to use?Q3: Which method to use?
9
P f t ti ti l t tPower of a statistical test
The power of a statistical test is the pprobability that the test will reject a false null hypothesis (i e it will notfalse null hypothesis (i.e., it will not make a Type II error).P 1 0 T IIPower= 1.0 – Type II error
10
QTL mapping from IM and ICIMQTL mapping from IM and ICIM
36
2427303336
e
IMICIM (PIN=0.01)
Under LOD threshold 3 0
1215182124
OD score Under LOD threshold 3.0,
IM identified 3 QTL, and ICIM identified 9 QTL
369
12LO
Q
0
1H
1H
1H
1H
1H
1H
2H
2H
2H
2H
2H
2H
2H
3H
3H
3H
3H
3H
4H
4H
4H
4H
4H
4H
5H
5H
5H
5H
5H
5H
5H
6H
6H
6H
6H
6H
6H
7H
7H
7H
7H
7H
7H
Testing position on the barley genome, step = 1 cM
11
g p y g , p
QTL mapping in a simulated population20
10
15
20
D s
core
完备区间作图 ICIM
0
1
2
ted
effe
ct
完备区间作图 ICIM
0
5LOD
‐2
‐1
Estim
a
10
15
20
scor
e
复合区间作图 CIM
0
1
2
ed e
ffect 复合区间作图 CIM
0
5
10
LOD
s
‐2
‐1
0
Estim
ate
15
20
core
区间作图 IM
1
2
d ef
fect
区间作图 IM
0
5
10
LOD
sc
‐2
‐1
0Es
timat
ed
12Testing every 1 cM on six chromosomes Testing every 1 cM on six chromosomes
0
11111112222223333334444445555556666662
1111111222222333333444444555555666666
Power from 3 simulated l i h f 200 RILpopulations, each of 200 RILs
Pop Chr. Pos. Effect LOD PVE (%) CI=10 cM( )1 2 25.1 0.19 2.56 3.48 False QTL
5 51.1 0.29 6.05 8.14 IQ56 60.0 0.30 6.72 8.86 IQ67 40.0 0.20 2.94 3.71 False QTL7 70 0 0 42 11 87 16 64 IQ77 70.0 0.42 11.87 16.64 IQ7
2 2 30.5 0.27 5.35 7.78 IQ25 45 0 0 27 5 25 7 94 False QTL5 45.0 0.27 5.25 7.94 False QTL6 59.1 0.26 4.94 7.50 IQ67 59.4 0.38 9.84 15.61 False QTL
3 2 30.0 0.21 2.50 3.96 IQ26 55.4 0.29 4.47 7.81 IQ6
13
7 70.0 0.28 4.42 7.14 IQ77 90.0 0.25 3.39 5.41 False QTL
Power comparison: ICIM vs. IMPower comparison: ICIM vs. IM
90100 20 40
60 80100 120
4050607080
ower (%
)
100 120140 160180 200220 240260 280300 320
010203040
Po
300 320340 360380 400420 440460 480500 520
100 20 40
1 2 3 4 5 10 20 30 FDRPercentage of phenotypic variation explained (PVE)
500 520540 560580 600
5060708090
er (%
)
60 80100 120140 160180 200220 240
1020304050
Powe 260 280
300 320340 360380 400420 440460 480
14
0
1 2 3 4 5 10 20 30 FDRPercentage of phenotypic variation explained (PVE)
460 480500 520540 560580 600
Pop lation si e eq i edPopulation size required
ProbabilityPVE (%) 0 9 0 8 0 7 0 6PVE (%) 0.9 0.8 0.7 0.6
1 >600 5402 600 420 340 2802 600 420 340 2803 430 280 230 2004 340 250 190 1605 280 200 160 1305 280 200 160 130
10 160 120 100 8020 100 80 60 50
15
20 100 80 60 5030 100 60 50 40
FDR: false discovery ratey(False QTL out of all positives)
90100 20 40
60 80100 120
607080
%)
100 120140 160180 200
405060
ower (% 220 240
260 280300 320
203040P 300 320
340 360380 400420 440
010 420 440
460 480500 520
16
1 2 3 4 5 10 20 30 FDR
Percentage of phenotypic variation explained (PVE)
540 560580 600
Q4: What are the ways that can improve mapping efficiency?improve mapping efficiency?
17
Possible ways
Large populationg p pPrecision phenotyping and
t igenotyping Efficient methodEfficient methodHigh marker density?!Two-stage mapping strategy?!
18
The length of empirical 95%The length of empirical 95% confidence intervals of QTLconfidence intervals of QTL
PVE (%) PS 200 PS 400PVE (%) PS=200 PS=400MD=5 cM
MD=10 cM
MD=20 cM
MD=40 cM
MD=5 cM
MD=10 cM
MD=20 cM
MD=40 cM
1 93.30 104.55 103.10 120.03 47.00 61.94 76.83 88.242 54.14 62.37 73.66 86.08 37.55 32.38 38.34 47.593 52 65 47 12 50 02 48 18 25 64 21 95 18 78 33 363 52.65 47.12 50.02 48.18 25.64 21.95 18.78 33.364 38.89 46.14 41.94 56.72 25.01 18.46 22.74 36.575 25.99 37.83 44.73 59.74 16.35 16.39 22.54 36.30
10 10.31 8.35 11.29 46.45 3.72 4.78 7.92 22.7420 8.55 10.19 14.78 26.97 4.90 6.04 8.70 15.5630 5 33 8 23 11 56 18 62 3 18 4 78 6 62 12 62
19
30 5.33 8.23 11.56 18.62 3.18 4.78 6.62 12.62
Q5: How to calculate theQ5: How to calculate the contribution of individual QTL?Q
20
Contribution of a QTL onContribution of a QTL on phenotypic variationp yp
PVE = Phenotypic variation explained (%)PVE Phenotypic variation explained (%)PVEg =Vg/Vp * 100%
BC, DH and RIL,Vg=a2 (a is the additive effect)F2, Vg=a2/2+d2/4 (d is the dominance effect), g / / ( )
21
D hi h ff t hi h PVE?Does high effect mean high PVE?
In DH or RIL, when there is segregation di idistortion,
Vg=(1-q)*a2+q*a2-[(1-2q)*a]2=4q(1-q)a2Vg (1 q) a +q a [(1 2q) a] 4q(1 q)a
Vg depends on effect and allele frequency
When p=q=0.5, Vg is maximized; otherwise, smaller than that of non-distortion
It is possible that one higher-effect QTL has lower PVEhas lower PVE
22
Non-additive PVE
For two random variables X and YE(X+Y) = E(X) + E(Y)V(X+Y) = V(X) + V(Y) +2Cov(X, Y)( ) ( ) ( ) ( , )
When QTL are unlinked, PVE of multiple QTL i th f i di id l PVEQTL is the sum of individual PVEWhen QTL are linked PVE of multipleWhen QTL are linked, PVE of multiple QTL is not equal to the sum of
d d lindividual PVE23
PVE can be more than 100%PVE can be more than 100%基因型 频率 基因型值
AABB )1(21 r− a1+a2
AAbb r1 a1-a2AAbb r2a1 a2
aaBB r21 -a1+a2
In the RIL population
aabb )1(21 r− -a1-a2
In the RIL populationGenetic variance of A-a: a1
2
Genetic variance of B-b: a22
Total genetic variance : a12 +a2
2+2(1-2r)a1a2ota ge et c a a ce a1 a2 ( )a1a2
24
A simulated populationA simulated population of 200 DH lines
30405060
scoreTwo QTL are located
t 25 M d 36 M01020
LOD
1 2
at 25 cM and 36 cMon a chromsome of 120 M Th i dditi
‐0 40
0.40.81.2
Effect
120 cM. Their additive effects are 1.0 and 1 0 R d
‐1.2‐0.80.4E
100120
−1.0. Random error variance is 0.4. M k i t l i 2 M
20406080100
PVE (%
)Marker interval is 2 cM.
25
0
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
105
110
115
Scanning every 1 cM
Linkage in couplingA: a1=1, a2=1, r=0.5; V1=1, V2=1, Vg=2, Ve=0.4; H2=0.83 B: a1=1, a2=1, r=0.1; V1=1, V2=1, Vg=3.6, Ve=0.4; H2=0.90Estimated R2 from regression = 0.78 Estimated R2 from regression = 0.82
60 60
1020304050
LOD score
1020304050
LOD score
0
00.40.81.2
fect
0
00.40.81.2
fect
‐1.2‐0.8‐0.4
0
Eff
100120
‐1.2‐0.8‐0.4
0
Eff
100120
020406080100
PVE (%)
020406080100
PVE (%)
26
0
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
105
110
115 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
105
110
115
Scanning every 1 cM
0
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
105
110
115
Scanning every 1 cM
Linkage in repulsionC: a1=1, a2=-1, r=0.5; V1=1, V2=1, Vg=2, Ve=0.4; H2=0.83 D: a1=1, a2=-1, r=0.1; V1=1, V2=1, Vg=0.4, Ve=0.4; H2=0.50Estimated R2 from regression = 0.89 Estimated R2 from regression = 0.51
60 60
102030405060
LOD score
102030405060
LOD score
0
00.40.81.2
ect
0
00.40.81.2
ect
‐1.2‐0.8‐0.4
0
Effe
100120
‐1.2‐0.8‐0.4
0
Effe
100120
020406080100
PVE (%)
020406080100
PVE (%)
27
0
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
105
110
115 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
105
110
115
Scanning every 1 cM
0
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
105
110
115
Scanning every 1 cM
Q6: How to determine the Q6 o o dsource of favorable alleles?
28
Source of favorable allelesSource of favorable alleles
Definition of additive and dominanceDefinition of additive and dominance genetic effects
Coding in QTL mapping: 2 (P1) 0 (P2) 1 (F1)Coding in QTL mapping: 2 (P1), 0 (P2), 1 (F1)P1: m+a; F1: m+d; P2: m-a
基因型 A2A2 A1A2 A1A1
中亲值m基因型值 G22 G12 G11
加性效应a 加性效应a
29显性效应d
Q7: Is selective genotypingQ7: Is selective genotyping still useful?still useful?
30
Selective genotypingSelective genotyping
Low tail High tail
LLHH
LH
ppppppt
)1()1( −+
−−
=
31L
LL
H
HH
Npp
Npp
2)(
2)(+
Comparison of SGM with IM and ICIMp(PVE=5%, MD=5cM and both tails have the
selected proportion of 10%)p p )
SGM has higher detection power than theSGM has higher detection power than the conventional IM but lower detection power
than ICIMthan ICIMSGM may still be useful!
Q8: Can composite traits be Q8 o pos sused in QTL mapping?
33
Genetic effects of composite traitsGenetic effects of composite traitsEffect Trait I Trait II Addition Subtraction Multiplication DivisionEffect Trait I Trait II Addition Subtraction Multiplication DivisionMean 25 20 45 5 500 1.2563A1 1 0 1 1 20 0.0503A 1 0 1 1 20 0 0503A2 1 0 1 1 20 0.0503A3 0 1 1 -1 25 -0.0631A4 0 1 1 -1 25 -0.0631A 0 0 0 0 0 0A12 0 0 0 0 0 0A13 0 0 0 0 1 -0.0025A14 0 0 0 0 1 -0.0025A23 0 0 0 0 1 -0.0025A24 0 0 0 0 1 -0.0025A34 0 0 0 0 0 0.006334A123 0 0 0 0 0 0A124 0 0 0 0 0 0A134 0 0 0 0 0 0.0003
34
A134 0 0 0 0 0 0.0003A234 0 0 0 0 0 0.0003A1234 0 0 0 0 0 0
Composite traits reduced powerComposite traits reduced power and increased FDRQTL Trait I Trait II Addition Subtraction Multiplication Division
Model I Power (%) Q1 95.10 69.60 69.30 55.20 50.50Q2 94 80 69 80 70 40 54 10 50 90Q2 94.80 69.80 70.40 54.10 50.90Q3 92.50 67.20 65.30 76.90 75.20Q4 94.50 68.40 65.40 77.80 75.20
(%)FDR (%) 21.63 22.98 27.42 28.05 28.07 29.68Model II Power (%) Q1 95.40 67.40 65.60 54.80 49.90
Q2 92.90 62.40 66.00 50.00 49.90 Q3 93.70 69.90 67.00 79.20 74.90 Q4 91.90 62.40 64.90 73.50 72.90
FDR (%) 21.35 22.18 28.76 28.59 28.07 28.89( )Model III Power (%) Q1 95.20 66.60 52.40 53.60 37.70
Q2 95.00 69.20 51.60 54.70 36.40 Q3 92 90 63 40 47 80 69 70 56 20
35
Q3 92.90 63.40 47.80 69.70 56.20 Q4 92.60 61.50 49.90 72.60 58.00
FDR (%) 19.78 23.44 28.83 27.71 29.74 30.18
Q9: Does the phenotype haveQ9: Does the phenotype have to be normally distributed?
36
Quantitative traits are normally distributed under the polygene hypothesis F2
h 2 0 90 h 2 0 10B1
h 2 0 00B2
h 2 0 95 h 2 0 05F2:3
2 2RIL
h 2 0 88 h 2 0 12
h mg2=0.90, h pg
2=0.10 h mg2=0.00
h pg2=1.00
h mg2=0.95, h pg
2=0.05 h mg2=0.89, h pg
2=0.11 h mg2=0.88, h pg
2=0.12
F2h mg
2=0.80, h pg2=0.10
B1h mg
2=0.002
B2h mg
2=0.87, h pg2=0.05
F2:3h mg
2=0.822
RILh mg
2=0.84, h pg2=0.13
h pg2=0.38 h pg
2=0.12
F2h mg
2=0.70, h pg2=0.10
B1h mg
2=0.00h pg
2=0.23
B2h mg
2=0.78, h pg2=0.05
F2:3h mg
2=0.74h pg
2=0.12
RILh mg
2=0.78, h pg2=0.13
Phenotypic di t ib ti
F2h mg
2=0.60, h pg2=0.10
B1h mg
2=0.00h pg
2=0.17
B2h mg
2=0.69, h pg2=0.05
F2:3h mg
2=0.66h pg
2=0.12
RILh mg
2=0.72, h pg2=0.14
distribution under one major
F2h mg
2=0.50, h pg2=0.10
B1h mg
2=0.00h pg
2=0.13
B2h mg
2=0.59, h pg2=0.05
F2:3h mg
2=0.57h pg
2=0.13
RILh mg
2=0.65, h pg2=0.16
gene model
F2h mg
2=0.40, h pg2=0.10
B1h mg
2=0.00h pg
2=0.11
B2h mg
2=0.49, h pg2=0.05
F2:3h mg
2=0.47h pg
2=0.13
RILh mg
2=0.57, h pg2=0.17Random errors
have to beF2
h mg2=0.30, h pg
2=0.10B1
h mg2=0.00
h pg2=0.09
B2h mg
2=0.38h pg
2=0.06
F2:3h mg
2=0.37h pg
2=0.14
RILh mg
2=0.47, h pg2=0.19
have to be normally di t ib t d d
37
F2h mg
2=0.20h pg
2=0.10
B1h mg
2=0.00h pg
2=0.08
B2h mg
2=0.26h pg
2=0.06
F2:3h mg
2=0.26h pg
2=0.15
RILh mg
2=0.34h pg
2=0.21
distributed and independent!
One QTL with PVE = 80%One QTL with PVE = 80%Located at 25 cM and a=1.0
40
50
cy
0
10
20
30Freq
uen
0
8.5 9 9.5 10 10.5 11 11.5 12Phenotypic value
1.2100
0.40.60.81
mated
effect
ICIMIM
20
40
60
80
LOD score
ICIMIM
‐0.20
0.2
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
105
110
115
120
125
130
135
140
145
150
155
160
Estim
1D‐scanning on one chromosome, step=1cM
0
20
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
105
110
115
120
125
130
135
140
145
150
155
160
1D‐scanning on one chromosome, step=1cM
38
Q10: Can the precision be Q pimproved by adding more
k ?markers?
39
Empirical marker density
In linkage mappingg pp g10-20 cM, covering the whole genomeM k d it l l tiMarker density + large population
In linkage mappingIn linkage mappingThe more, the better to exploit the
i i LD i th i l tiremaining LD in the mapping population
40
Power and FDR for two marker d i i 10 M ( ) d 20 M (d )densities: 10 cM (up), and 20 cM (down)
(Confidence interval is the whole chromosome)
60708090
100
er
20406080100120 8
10
12
14
overy rate
1020304050
Powe 140
160180200220240260 0
2
4
6
False disco
0
1 2 3 4 5 10 20 30Phenotypic variation explained (PVE) (%)
260280300320340
0
20 80 140
200
260
320
380
440
500
560
Population size
60708090
100
r
20406080100120 8
10
12
14
very rate
102030405060
Powe 140
160180200220240260 0
2
4
6
False discov
41
010
1 2 3 4 5 10 20 30Phenotypic variation explained (PVE) (%)
260280300320340
020 80 140
200
260
320
380
440
500
560
Population size
Power and FDR for two marker densities: 10 cM (up), and 20 cM (down)
(10 cM confidence interval, true QTL at the center of CI)
708090
100 20406080100 60
708090
ery rate
2030405060
Power
120140160180200220240 10
20304050
False discove
010
1 2 3 4 5 10 20 30Phenotypic variation explained (PVE) (%)
240260280300320340
0
20 80 140
200
260
320
380
440
500
560
Population size
708090
100 20406080100 60
708090
ry rate
203040506070
Power
100120140160180200220240 10
2030405060
False discover
4201020
1 2 3 4 5 10 20 30Phenotypic variation explained (PVE) (%)
240260280300320340
010
20 80 140
200
260
320
380
440
500
560
Population size
Q11: What is effect of missing markers?
43
Mi i d t i QTL iMissing data in QTL mapping
Missing markersss g a e sImputation using the linkage map
Missing phenotypelMean replacement
DeletionDeletion
44
Effect of missing markersec o ss g a e s(First simulated F2 population from QTL distribution
model I and population size 500)
No missing markers 5% of missing
10% of missing 15% of missing
45
Power analysis of various ylevels of missing markers
A
80
100
%)
0 5% 10% 15% 20% 25% 30% C
80
100
)
20
40
60
Po
wer
(%
20
40
60
Po
wer
(%)
B
0qPH1-1qPH1-2qPH3-1qPH3-2 qPH4 qPH5 qPH6 qPH7 qPH12 FDR
100 D
0qHD1 qHD3 qHD4 qHD5 qHD8 qHD11 FDR
100
40
60
80
Po
wer
(%
)
40
60
80
100
ow
er
(%)
C
0
20
qPH1-1qPH1-2qPH3-1qPH3-2 qPH4 qPH5 qPH6 qPH7 qPH12 FDR
P
0
20
40
qHD1 qHD3 qHD4 qHD5 qHD8 qHD11 FDR
Po
A,QTL for plant height, population size=180B,QTL for plant height, population size=500
C, QTL for heading days, population size=180D, QTL for heading days, population size=500
Effect of missing markers is similar gto the reduction in population size
Population size 180, with various missing marker levels 100100
40
60
80 180 171 162 153 144 135 126
40
60
80
ower (%
)
0% 5% 10% 15% 20% 25% 30%
0
20
qPH1‐1qPH1‐2qPH3‐1qPH3‐2 qPH4 qPH5 qPH6 qPH7 qPH12 FDR
0
20
qPH1‐1 qPH1‐2 qPH3‐1 qPH3‐2 qPH4 qPH5 qPH6 qPH7 qPH12 FDR
Po
Population size 500, with various missing marker levels
80
100 500 475 450 425 400 375 350
80
100
%)
0% 5% 10% 15% 20% 25% 30%
20
40
60
20
40
60
Power (%
47
0
qPH1‐1qPH1‐2qPH3‐1qPH3‐2 qPH4 qPH5 qPH6 qPH7 qPH12 FDR
0
qPH1‐1 qPH1‐2 qPH3‐1 qPH3‐2 qPH4 qPH5 qPH6 qPH7 qPH12 FDR
Q12: What is the effect of Q s osegregation distortion?
48
Segregation distortion
P1 (AA) X P2 (aa)无选择
P1BC1 AA A 1 1P1BC1: AA:Aa=1:1P2BC1: Aa:aa=1:1F2: AA:Aa:aa=1:2:1DH RIL: AA:aa=1:1DH, RIL: AA:aa=1:1
Reasons for distortion Random driftSelection in gametes and zygotes g yg
49
Ratio of AA:aa in a barleyRatio of AA:aa in a barley DH population
1 0
1.2
1.4
1.6
1.8
0.0
0.2
0.4
0.6
0.8
1.0
0.0
Act8
A
aHor
2
ABG
464
iP
gd2
A
tpbA
A
BC26
1
Aga
7
MW
G84
4
WG
516
M
WG
520A
B
CD
351B
BC
D11
1
MW
G86
5
MW
G88
2
ABG
317
AB
G61
3
ABC
171
AB
G47
1
Ugp
1
MW
G84
7
MW
G20
40
ABC
166
M
WG
838
A
BC17
2
WG
622
C
DO
669
M
WG
716
AB
G71
5
MW
G65
5C
ABG
601
AB
A306
B
MW
G50
2
MW
G92
0A
Dor
5
BC
D41
0B
ksuA
1B
mSr
h
MW
G91
4
RZ
404
C
DO
504
Tu
bA3
BC
D29
8
ABG
463
A
BC30
9
PSR
167
AB
G37
8
PSR
106
M
WG
916
BC
D10
2
MW
G82
0
Am
y1
ABG
711
M
WG
658
dR
pg1
BC
D12
9
Prx1
A
ABG
380
Br
z
MW
G51
1
VAt
p57A
M
WG
889B
P
SR12
9
ABG
608
M
WG
635B
50
Segregation distortion in anSegregation distortion in an actual rice F2 population
RP12920
RP178
RM159RM147
91
16
20
(-lo
g 10P
)
RM143
4 RM
304 R
M49
8
12
nce
leve
l
RM
44 R RM552
4
8
sign
ifica
n
0
1 1 1 1 2 2 2 2 3 3 3 3 4 4 5 5 5 5 6 6 6 6 7 7 8 8 9 910 10 11 11 11 12 12
Markers on the 12 rice chromosomesMarkers on the 12 rice chromosomes
When segregation distortion markers are not linked with QTL
AA
80
100
%)
No distortion SDM1 SDM2 SDM3SDM4 SDM5 SDM6 SDM7SDM8 SDM9A,QTL for
l h i h
20
40
60Po
wer
(%plant height, population i 180
B
0
20
qPH1-1 qPH1-2 qPH3-1 qPH3-2 qPH4 qPH5 qPH6 qPH7 qPH12 FDR
size=180
B QTL f B
80
100B,QTL for heading d
40
60
Pow
er (%
)days, population i 180
0
20
qHD1 qHD3 qHD4 qHD5 qHD8 qHD11 FDR
Psize=180
When segregation distortion g gmarkers are linked with QTL
A
80
100 No distortion SDM1 SDM2 SDM3SDM4 SDM5 SDM6 SDM7SDM8 SDM9
C
80
100
)
20
40
60
Pow
er (%
)
20
40
60
Pow
er (%
)
B
0qPH1-1 qPH1-2 qPH3-1 qPH3-2 qPH4 qPH5 qPH6 qPH7 qPH12
80
100 D
0qHD1 qHD3 qHD4 qHD5 qHD8 qHD11
100
120
40
60
80
Po
wer
(%)
40
60
80
Pow
er (%
)
0
20
qPH1-1 qPH1-2 qPH3-1 qPH3-2 qPH4 qPH5 qPH6 qPH7 qPH12
0
20
qHD1 qHD3 qHD4 qHD5 qHD8 qHD11
A,QTL for plant height, population size=180B,QTL for plant height, population size=500
C, QTL for heading days, population size=180D, QTL for heading days, population size=500
Effect of segregation distortionEffect of segregation distortion markers (SDM) on QTL mapping
If the SDM is not closely linked with any l t h i ht h di d t QTLplant height or heading date QTL, no
significant effects were observed on the detection power.Otherwise SDM may increase or decreaseOtherwise, SDM may increase or decrease the QTL detection power.
l l fIn large-size populations, say size of 500, the effect of SDM was minor even the SDM was closely linked with QTL.
奇异分离对作图功效的影响奇异分离对作图功效的影响(F2群体)
2211021
220202 )()(2])([ dffadfffaffffVG −+−−−−+= 110210202 )()(2])([ dffadfffaffffVG ++
a b dc
k=1k 1
r=0Prob(k>1) ≈ 47%
r=0.5Prob(k>1) ≈ 51%
r=0.75 Prob(k>1) ≈ 52%
r=1 Prob(k>1) ≈ 50%
k=1k=1k=1f qq
k=1 k=1
k=1
k=1
k=1( ) ( ) ( ) ( )
fe g hfQQ fQQfQQ fQQ
r=1.5 r=2 r=5 r=10
f qq
k=1k=1 k=1
k=1
r 1.5 Prob(k>1) ≈ 37%
r 2 Prob(k>1) ≈ 29%
r 5Prob(k>1) ≈ 19%
r 10 Prob(k>1) ≈ 14%
fQQ fQQfQQ fQQ
k=1
In F1 and BC derived DHIn F1 and BC derived DH populations
1 2
1.4
1.0
1.2
ce (k)
0.6
0.8
of varianc
0.4
Ratio o
BC1‐derived DH population
0.0
0.2 F1‐derived DH population
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
0.55
0.60
0.65
0.70
0.75
0.80
0.85
0.90
0.95
1.00
Frequency of QQ