Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene...
-
Upload
barbra-hamilton -
Category
Documents
-
view
214 -
download
0
Transcript of Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene...
![Page 1: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/1.jpg)
Future Directions
Pak Sham, HKU
Boulder 2007
![Page 2: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/2.jpg)
Genetics of Complex Traits
Quantitative Genetics Gene Mapping
Functional Genomics
![Page 3: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/3.jpg)
Gene Mapping – GWA Era
The promiseDetect all the “big” genetic playersUnderstand how they interact with each other and with the environmentGeneration of detailed hypotheses regarding etiology
The challengesHow to get enough funding?How to get the most out of them?
Study designData analysis
![Page 4: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/4.jpg)
GWA: Taking the Plunge
Age-related Macular Degeneration
96 cases / 50 controls
100,000 SNPs Klein et al. (2005)
![Page 5: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/5.jpg)
CARD15
IBD5
ATG16L1
IL23R
conf
conf
From Dr Mark Daly946 cases, 977 controls
![Page 6: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/6.jpg)
NOTCH-2 (1)
TCF7L2 (3)
Type 2 Diabetes Mellitus Genome-wide Association Results
From Dr Mark Daly
![Page 7: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/7.jpg)
Design of GWA Studies
Reducing costShared pool of control subjectsSplit sample designs
Two stage: GWA replicationSplit-half: e.g. 250K (Sty / Nsp) in each half
Maximizing informationChoosing most extreme (genetically loaded) subjectsChoosing most accurately and comprehensively phenotyped subjects
![Page 8: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/8.jpg)
Mining GWA Data
AimTo squeeze out all the information from the vast amount of genotype data
Strategy
Optimal genotype calls
Thorough data cleaningSample characterization (stratification) Apply multiple statistical methodsReplication
![Page 9: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/9.jpg)
WGA Analysis Flow
Raw probe signals Copy number variations
Genotype calls
Cleaned genotype data
Expanded genotype data (with imputations)
Sample stratification
CNV –phenotype relationship
Stratified single SNP association testsOther analyses
![Page 10: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/10.jpg)
Deletion CNVs
![Page 11: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/11.jpg)
CNV and Disease
CNVs
are common throughout the genome
can influence gene function (increased / decreased levels)
can influence disease susceptibilityCharcot Marie-Tooth disease type A (PMP22)Early-onset Parkinson’s disease (alpha-
synuclein)Susceptibility to HIV infection (CCL3L1)
![Page 12: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/12.jpg)
Family-based Tests
“DFAM” Test (implemented in PLINK)Break pedigree into nuclear families
Consider count of minor allele (X) among affected offspring
Calculate expectation and variance of (X) conditional on parental genotypes (if available) or sibship genotypes
Calculate single test statistics
![Page 13: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/13.jpg)
Combining Studies
Imputation to establish common SNP set and thenCombine data and use stratified
association analysis
Meta-analysis – combine odds ratios (inverse variance weighting)
![Page 14: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/14.jpg)
Looking for Epistasis
Epistatic components not detectable by single-locus association analyses
Simple methods of epistasis analysesTest of homogeneity of odds ratios or means differences
across genotypic strata Test of interaction in logistic or linear regression modelsTest for correlation between unlinked lociTest for difference in correlation between loci, in cases
and controlsIncreases multiple testing: e.g. 500,000 SNPs leads to
124,999,750,000 possible pairs of SNPs
![Page 15: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/15.jpg)
Epistasis: Is it worth doing?
Marchini et al (2005) compared 4 analytic strategies using simulated data with epistasis
(1) One single-locus analysis(2) Two single-locus analyses(3) All possible pairs of loci(4) Two-stage: pairs of loci with low p-values from single-locus analysis
Results: Strategies (3) and (4) are often more powerful than (1) and (2) even after Bonferroni adjustment
![Page 16: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/16.jpg)
Will GWA catch all?
Certainly not! Alleles of small effects Rare variants
“Residual” genetic variation
![Page 17: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/17.jpg)
Using Functional Information
ABCDEFGHIJ
12345678
Canonical correlation analysis
Gene-gene interaction
Gene-based tests: Haplotypic / Allelic
Pathways –based tests
![Page 18: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/18.jpg)
Population-based Linkage
Linkage is possible only in families, BUT
Every pair of individuals are related if traced back far enough
Genetic relationship (overall genomic sharing) can be estimated from GWA data
Local IBD sharing can be also estimated from GWA data Therefore IBD sharing can be correlated with phenotypic
similarity in GWA data Likely to be useful for rare phenotypes with rare variants
of moderately strong effects
![Page 19: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/19.jpg)
Estimating Genome-wide IBD
0 1 2
0 2 (piqi)2 4 piqi(1-2piqi) N-2 piqi(2-3piqi)
1 0 2 piqi N - 2 piqi
2 0 0 N
IBD
Expected number of SNPs with IBS =
N : Number of SNPs; p, q : allele frequencies
![Page 20: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/20.jpg)
Estimating Genome-wide IBD
N0 = Number of IBS0 SNPs
N1 = Number of IBS1 SNPs
N2 = Number of IBS2 SNPs
iiiiii
iiiiii
ii
qpfqpqpfNNE
qpfqpqpfNE
qpfNE
102
101
00
2)32(2)(
2)21(4)(
2)(
f0 = Proportion of genome IBD0f1 = Proportion of genome IBD1f2 = Proportion of genome IBD2
![Page 21: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/21.jpg)
Estimating Genome-wide IBD
The estimated genome-wide IBD proportions, obtained by solving the linear equations, are:
Boundary conditions Small sample (rare allele) adjustment Inbreeding adjustment
212
011
00
1
2/)21(4
2/
fff
qpqpqpfNf
qpNf
iiiiii
ii
![Page 22: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/22.jpg)
HapMap Relationships
B
C
A 2
3
1
3
1 2
B
C
A
P(IBD=0) P(IBD=1) P(IBD=2)
A 2 0.005 0.995 0.000
A 3 0.436 0.564 0.000
C 2 0.388 0.612 0.000
Two Yoruba Trios
![Page 23: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/23.jpg)
Distribution of Genomic Sharing: Among CEPH Founders
Average sharing ~ 1.6%(Average kinship ~ 0.8%)
![Page 24: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/24.jpg)
Estimating Segmental Sharing
Prune SNPs to reduce LD relationships Use allele frequencies to calculate
likelihood of IBD given single SNP genotypes of the pair of individuals
Use genome-wide IBD to estimate the least number of meioses that separate the two genomes
Use number of meioses to calculate transition matrix of IBD states
Use Hidden Markov Model to calculate “multipoint” IBD probabilities
![Page 25: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/25.jpg)
Transition Matrix
0 1
0
1
12
111
2
m
m 12
111
1
2
m
m
211 m 21 m
22 )1( : Recombination fraction
m : Number of meioses (2)
IBD(i)
IBD(i+1)
![Page 26: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/26.jpg)
Estimated Segmental Sharing
YRI: NA19130, NA191940 (Half aunt / Half niece)
Overall: IBD0 = 0.76, IBD1 = 0.24, IBD2 = 0
0.0 e+00 5.0 e+07 1.0 e+08 1.5 e+08 2.0 e+08
51
01
52
0
Index
tmp
3[, 7
]
![Page 27: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/27.jpg)
0.0 e+00 5.0 e+07 1.0 e+08 1.5 e+08 2.0 e+08
51
01
52
0
Index
tmp
3[, 7
]
Estimated Segmental Sharing
YRI: NA18913, NA19240 (Grandparent / Grandchild)
Overall: IBD0 = 0.44, IBD1 = 0.56, IBD2 = 0
![Page 28: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/28.jpg)
Other Data Mining Methods
The possibilities are endless!
Neural networks
CART
MARS etc …….
![Page 29: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/29.jpg)
Beyond GWA
Incorporating measured genotypes into quantitative genetic-epidemiological analysis
Functional genomic studies – gene expression profiles, cell biology, etc.
![Page 30: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/30.jpg)
Summary
The technology for GWA is reaching maturity
GWA is already yielding novel susceptibility loci for complex diseases
GWA are increasing in number and in size
GWA data offer interesting analytical and computational challenges
The results from GWA studies will revolutionize quantitative genetics and functional genomics
![Page 31: Future Directions Pak Sham, HKU Boulder 2007. Genetics of Complex Traits Quantitative GeneticsGene Mapping Functional Genomics.](https://reader035.fdocuments.in/reader035/viewer/2022062518/5697bf761a28abf838c80e16/html5/thumbnails/31.jpg)
The End