Download Common trait genetics

Common trait genetics
Chris Cotsapas
Traits and (sub?)phenotypes
• Categorical
– Disease/healthy
– Super-skinny/skinny/normal/fat/super-fat
• Continuous
– Weight, height, BMI, WHR, LDL
– Gene expression, methylation, DNase I
Heritability
• Proportion of differences due to genetic
factors
– From family and twin studies
• Broad sense (H2)
– Additive, dominant and epistatic
• Narrow sense (h2)
– Additive only
• How many genetic factors explain
heritability?
– Mendel v Galton
Mendelian – one variant, “all” H2/h2
Necessary and sufficient
(expressivity,
penetrance)
Galton - multigenicity
Common and rare variant
hypotheses
GENETIC STUDY DESIGN
Family or cohort study?
Genetic data
SNP
Ind1
Ind2
Ind3
Ind4
1
AA
AG
AA
GG
2
TC
CC
CC
CC
3
GT
GG
TT
GG
4
AC
CC
AC
AC
5
AT
AA
TT
AA
Linkage analysis and TDT
Transmission
Disequilibrium
Test
Case/control association
• H0: Frequency of ‘A1’ is independent of case/control
status.
A1
A2
Cases
w
x
Controls
y
z
c2 = (O-E)2/E
[Pearson’s chi-Square]
Odds Ratio (OR): Odds of Allele
occurring in cases to the odds of
Allele occurring in controls:
w/x
y/z
=
wz
xy
Power
Small sample size
Large sample size
Freq
Freq
Cases
Controls
Cases
Controls
Regression analysis
• Analysis of the relationship between a dependent or outcome
variable (phenotype) with one or more independent or
predictor variables (SNP genotype)
Logistic Regression
ln( pi )
( 1 - pi )
= b 0 + b 1 Xi + e i
Continuous Trait Value
Linear Regression
Yi = b0 + b1Xi + ei
Slope: b1
b0
0
Pro tip Z2 = χ2
1
Number of A1 Alleles
2
QQ plot
P < 5e-8
Manhattan plot
META-ANALYSIS
Combining data – meta-analysis
Replication cohort
GWA
S
GWAS
GWAS
Combine statistics
• Fisher’s method:
P <-> Z
• Sum of χ2
(or Z2)
k = number of studies
Efficient meta-analysis
POWER
wi is each test weight
wt is the sum of the weights
WARNING: must be same phenotype and scale (e.g. height in cm in all studies)
Regression coefficients
Bi = study beta; σ2 = variance of each individual betas
OBSERVATIONS
http://www.ebi.ac.uk/fgpt/gwas/images/timeseries/gwas-latest.png
Effect sizes – T1D
Petretto, Liu and Aitman NG 2007
Bulik-Sullivan et al NG 2015
Miki et al NG 2010
Fine mapping strategies
• If a SNP is causal, then r2
should predict association of
other SNPs in the area:
Kichaev et al. PLoS Genet. 2014
LD score
Bulik-Sullivan et al. NG 2015
Maurano et al Science 2012
GWAS signals
are enriched in
tissue-specific
gene regulatory
sequence
Gusev et al ASHG 2014
DRILLING DOWN
MS GWAS risk effect: NFKB1 locus
97 MS risk loci;
IMSGC, Nat Genet 2013
MS patients show altered NFκB
signaling in CD4+ T cells
Figure 1. Naïve CD4 cells from patients with MS
exhibit increased phospho-p65 NFκB. Flow
cytometry of PBMCs from age-matched healthy
+ T cells show
control
(HC) CD4
and relapsing-remitting
MS (RRMS)
ex vivo
higher
patients stained for CD4, CD45RA, CD45RO, and
p-p65
STM
2015)
pS529
p65 (Housley
NFκB. MFI of et
p65al,
results
are shown
gated on naïve CD4+CD45RA+CD45RO- T-cells.
CD4+ T cells from MS patients
proliferate more rapidly after
stimulus (Kofler et al JCI 2014)
MS risk effect near NFKB1 alters
signaling in CD4+ cells
Nuclear localization
rs228614
p= 0.037
p50 NFkB
30
20
10
0
Housley, STM 2015
GG
AA
p= 0.05
30
GG
AA
20
10
0
0
15
30
Minutes
GWAS loci harbor many NFκB
genes
Will Housley, David Hafler
Model: NFκB signaling variation
p50
External
stimulus
P-p50
p65
*NFκB
Activation
Proliferation
Survival
*NFκB
Activation
Proliferation
Survival
Broader
phenotype?
p50
P-p50
GV in NFκB pathway
New gene activation
patterns by NFκB
GV in NFκB TFBS
p65
IκB
p50
p65
Resting state
Stimulus
Phosphorylation
Nuclear translocation
P
rs228614-A
carriers
Low cytoplasmic NFκB
Gene
activation
P
Low levels
P
Moderate transcription
P
rs228614-G
carriers
P
High cytoplasmic NFκB
P
P
P
High levels
High activation threshold
Moderate proliferation
Distinct CD4+ subset fates
P
High transcription
P
Cell
phenotype
P
Other target activation
Low activation threshold
High proliferation
Plastic CD4+ subset fates