Detection of genegene interactions in genomewide association studies

1
Detection of gene-gene interactions in
genome-wide association studies
Manuel A R Ferreira
Center for Human Genetic Research
Massachusetts General Hospital
Harvard Medical School
2
What is epistasis or GxG?
Phenotype
5
3
Locus B
1
0
1
1
Locus A
Epistasis defined as the extent to which the
joint contribution of two loci towards a
phenotype deviates from that expected under a
purely additive model.
Fisher (1918)
3
Is it expected to be important for complex
traits/diseases?
Not much evidence
Growing evidence
Model organisms
Model organisms
Brem et al. 2005 Nature 436 701703 Yeast Li
et al. 1997 Genetics 145, 453465
Rice Montooth et al. 2003 Genetics 165, 623635
Drosophila Carlborg et al. 2003 Genome Res.
13, 413421 Chicken Shimomura et al. 2001
Genome Res. 11, 959980 Mice
Xu Jia 2007 Genetics 175, 1955-1963
Barley Zeng et al. 2000 Genetics 154, 299310
Drosophila Flint et al. 2004 Mamm. Genome 15,
7782 Mice
Humans
Humans
Maller et al. 2006 Nat Genet. 381055-9
Schadt Lum 2006 J Lipid Res 47
26012613 (Gjuvsland et al. 2006 Genetics 175
411420)
Can we detect it in genome-wide association
studies?
Technically challenging
Astronomical number of tests (how to perform,
analyze and correct for them, power)
Plausible for certain models of interaction
Marchini et al. 2005 Nat Genet 37, 413417 Evans
et al. 2006 PLoS Genet 2, e157
No reports as yet (in humans)
4
Traditional methods to detect epistasis
1. Regression
Flexible framework
Slow
m3. (LocusA LocusB)
y m1.LocusA m2.LocusB
2. Linkage Disequilibrium or allelic-association
Powerful (eg. case-only)
Less flexible, phasing
a d
OR
b c
3. Transmission distortion
More robust
Less powerful
All allele-based!
5
New methods
Allele-based test
Faster standard tests (eg. logistic regression),
useful for whole-genome screens
Collapse B
Collapse A
a d
OR
b c
ORcases ? ORcontrols
Test for epistasis
6
New methods
Gene 1
Gene 2
A
B
1
C
2
35 allele-based tests
SNPs
D
3
E
4
F
5
A
B
1
C
2
A single gene-based test
D
3
E
4
F
5
7
New methods
2. Gene-based test
Reduce tests, capture haplotypic variation,
analysis of pathways or networks
Case-only sample
Powerful
Less robust
1. Canonical correlation analysis of Gene 1 and
Gene 2
p canonical correlations
2. Estimate the significance of all correlations
using Bartletts (1941) test
Case-control sample
Flexible
Less powerful
1. Canonical correlation analysis of Gene 1 and
Gene 2
Store composite variables for Gene 1 and Gene 2
associated with the largest canonical correlation
2. Test for interaction between these composite
variables using standard linear or logit
regression
m3. (Gene1 Gene2)
y m1.Gene1 m2.Gene2
8
New methods
?
3. Performance
Gene 1
Gene 2
Type-1 error
(a 0.05)
9
New methods
?
3. Performance
Gene 1
Gene 2
Power
(a 0.05)
10
http//pngu.mgh.harvard.edu/purcell/plink/
11
Application to a bipolar disease GWAS
Poster 560
New Bioinformatic and Computational Methods
3.15 5.00pm
12
Acknowledgements
MGH
University College London
Shaun Purcell Pamela Sklar Mark Daly Ed
Scolnik Laurie Weiss Douglas Ruderfer Yan
Meng Jennifer Stone Matt Ogdie
Hugh Gurling
WTCCC
Nick Craddock
Funding
NHMRC Sidney Sax post-doctoral fellowship
STEP-BD
Jordan Smoller Roy Perlis Vishwajit
Nimgaonkar Nan Laird Matt McQueen Steve Faraone