lecture 15 regulatory variation and eqtls

Lecture 15Regulatory variation and eQTLs

Chris [email protected]

6.047/6.878/HST.507Computational Biology: Genomes, Networks, Evolution

Module 4: Population / Evolution / Phylogeny

• L15/16: Association mapping for disease and molecular traits– Statistical genetics: disease mapping in populations (Mark Daly)– Quantitative traits and molecular variation: eQTLs, cQTLs

• L17/18: Phylogenetics / Phylogenomics– Phylogenetics: Evolutionary models, Tree building, Phylo inference– Phylogenomics: gene/species trees, coalescent models, populations

• L19/20: Human history, Missing heritability– Measuring natural selection in human populations– The missing heritability in genome-wide associations

• And done! Last pset Nov 11 (no lab), In-class quiz on Nov 20– No lab 4! Then entire focus shifts to projects, Thanksgiving, Frontiers

Today: Regulatory variation and eQTLs1. Quantitative Trait Loci (QTLs), Regulatory Variation

– Molecular phenotypes as QTs: expression, chromatin…– Discretization: a GWAS for each gene. Cis-/Trans-eQTLs– Underlying regulatory variation: eQTLs, GWAS, cis-eQTL

2. Finding trans-eQTLs (distal from gene that varies)– Challenges: Power, structure, sample size– Cross-phenotype analysis: trans QTLs affect many genes

3. Identifying underlying regulatory mechanisms– Cis-eQTLs: TSS-distance, cell type specificity– eQTLs vs. GWAS: Expression as intermediate trait

4. Population differences, emerging efforts– Shared associations, SNP-gene pairs, allelic direction– Confound: environment, preparation, batch, ancestry

Quantitative traits- weight, height- anything measurable- today: gene expression

QTLs (QT Loci)- The loci that control

quantitative traits

Regulatory variation

• What do trait-associated variants do?• Genetic changes to:

– Coding sequence **– Gene expression levels– Splice isomer levels– Methylation patterns– Chromatin accessibility– Transcription factor binding kinetics– Cell signaling– Protein-protein interactions

Regulatory

BASIC CONCEPTSHistory, eQTL, mQTL, others

Within a population

• Damerval et al 1994• 42/72 protein levels differ in maize• 2D electrophoresis, eyeball spot quantitation• Problems:

– genome coverage– quantitation– post-translational modifications

• Solution: use expression levels instead!

Usual mapping tools available

• Discretization approach

gene 3

Whole-genome eQTL analysis is an independent GWAS for expression of each

gene

gene 2

gene N

gene 5

gene 4

gene 1

• cis-eQTL– The position of the eQTL maps near

the physical position of the gene.– Promoter polymorphism?– Insertion/Deletion?– Methylation, chromatin

conformation?

• trans-eQTL– The position of the eQTL does not

map near the physical position of the gene.

– Regulator?– Direct or indirect?

Modified from Cheung and Spielman 2009 Nat Gen

Genetics of gene expression (eQTL)

eQTL – THE ARRAY ERAyeast, mouse, maize, human

Yeast

• Brem et al Science 2002• Linkage in 40 offspring of lab x wild strain

cross • 1528/6215 DE between parents• 570 map in cross

– multiple QTLs– 32% of 570 have cis linkage

• 262 not DE in parents also map

trans hotspots

Brem et al Science 2002

Yvert et al Nat Genet 2003

Mammals I

• F2 mice on atherogenic diet• Expression arrays; WG linkage

Schadt et alNature 2003

Mammals II

Chesler et al Nat Genet 2005

10% !!

Mammals III

• No major trans loci in humans– Cheung et al Nature 2003– Monks et al AJHG 2004– Stranger et al PLoS Genet 2005, Science 2007

WHERE ARE THE TRANS eQTLS?Open question

gene 3

Whole-genome eQTL analysis is an independent GWAS for expression of each

gene

gene 2

gene N

gene 5

gene 4

gene 1

Issues with trans mapping

• Power– Genome-wide significance is 5e-8

– Multiple testing on ~20K genes– Sample sizes clearly inadequate

• Data structure– Bias corrections deflate variance– Non-normal distributions

• Sample sizes– Far too small

But…

• Assume that trans eQTLs affect many genes…

• …and you can use cross-trait methods!

Association data

Z1,1 Z1,2 … … Z1,p

Z2,1

::

Zs,1 Zs,p

Cross-phenotype meta-analysis

SCPMA ~L(data | λ≠1)

L(data | λ=1)

Cotsapas et al, PLoS Genetics

CPMA detects trans mixtures

Open research questions

• Do trans effects exist?– Yes – heritability estimates suggest so.– Can we detect them?

• Larger cohorts?– Most eQTL studies ~50-500 individuals– See later, GTEx Project

• Better methods?– Collapsing data?– PCA, summary statistics, modeling?

CAN WE LEARN REGULATORY VARIATION FROM eQTL?

First, let’s define the question

• Can we use genetic perturbations as a way to understand how genes are regulated?

• In what groups, in which tissues? • To what stimuli/signaling events? • Do cis eQTLs perturb promoter elements?• Do trans perturb TFs? Signaling cascades?

Most significant SNP per gene 0.001 permutation threshold

Significant associations are symmetrically distributed around TSS

Stranger et al., PLoS Gen 2012

268 271 262

73 85 82

86 86 86

Cell type-specific and cell type-shared gene associations(0.001 permutation threshold)

cell type

No.

of c

ell t

ypes

with

gen

e as

soci

ation

69-80% of cis associations are cell type-specific

• cis association sharing increases slightly when significance thresholds are relaxed• Cell type specificity verified experimentally for subset of eQTLs

Dimas et al Science 2009

Dimas et al Science 2009Slide courtesy Antigone Dimas

Open research questions

• Do cis eQTLs perturb functional elements?– Given each is independent, how can we know?

• Do tissue-specific effects correlate with the expression of a gene across tissues? Or a regulator?– Perhaps a gene is expressed, but in response to different

regulators across tissues?• If we ever find trans eQTLs…

– Common regulators of coregulated genes?– Tissue specificity?– Mechanisms?

APPLICATION TO GWASCandidate genes, perturbations underlying organismal phenotypes

eQTLs as intermediate traits

Schadt et al Nat Genet 2005

Modified from Nica and Dermitzakis Hum Mol Genet 2008

Exploring eQTLs in the relevant cell type is important for disease association studies

cell type not relevant for diseaserelevant cell type for disease

Importance of cataloguing regulatory variation in multiple cell types

Slide courtesy Antigone Dimas

Barrett et al 2008de Jager et al 2007

Franke et al 2010Anderson et al 2011

POPULATION DIFFERENCES

Shared association in 8 HapMap populations

APOH: apolipoprotein H Stranger et al., PLoS Gen 2012

Number of genes with cis-eQTL associations8 extended HapMap populations

Spearman Rank CorrelationEnsGene

0.01 0.001 0.0001# genes FDR # genes FDR # genes FDR

CEU 2869 0.06 657 0.03 313 0.01CHB 2832 0.06 774 0.02 378 0.00GIH 2959 0.06 698 0.03 300 0.01JPT 2900 0.06 795 0.02 386 0.00

LWK 3818 0.05 773 0.02 311 0.01MEX 2609 0.07 472 0.04 165 0.01MKK 4222 0.04 947 0.02 411 0.00YRI 3961 0.05 799 0.02 328 0.01

non-redundant 12494 3130 1132>2 pops 6889 0.55 1074 0.34 547 0.488 pops 151 0.01 63 0.02 28 0.02

4 PCA + 4 non-PCAGenes are Ensembl genes.Generated on July 22, 2009, modified July 28th because of MKK error.These numbers differ from v2 because all assocs where dist to true TSS > 1Mb were removed.This removed some genes. I redid the MostSigSnpEnsGene after removing those SNPs. This is in column MostSigSnpEnsGene_1

SRC: permutation threshold


Direction of allelic effectsame SNP-gene combination across populations

AGREEMENT

OPPOSITE

Population 1

rs40915

THAP5

TTCTCC

8.00

7.75

7.50

7.25

7.00

6.75

6.50

Population 2

rs40915

THAP5

TTCTCC

8.00

7.75

7.50

7.25

7.00

6.75

6.50

log 2 e

xpre

ssio

n

log 2 e

xpre

ssio

n

rs40915

THAP5

TTCTCC

8.00

7.75

7.50

7.25

7.00

6.75

6.50

log 2 e

xpre

ssio

n

rs40915

THAP5

TTCTCC

8.00

7.75

7.50

7.25

7.00

6.75

6.50

log 2 e

xpre

ssio

n


Slide courtesy Alkes Price

Population differences could have non-genetic basis

• Differences due to environment? (Idaghdour et al. 2008)

• Differences in cell line preparation? (Stranger et al. 2007)

• Differences due to batch effects? (Akey et al. 2007)

(Reviewed in Gilad et al. 2008)


Gene expression experiment

Does gene expression in 60 CEU + 60 YRI vary with ancestry?

Does gene expression in 89 AA vary with % Eur ancestry?

60 CEU + 60 YRI from HapMap, 89 AA from Coriell HD100AAGene expression measurements at 4,197 genes obtained using Affymetrix Focus array

c


Gene expression differences in African Americans validate CEU-YRI differences

c = 0.43 (± 0.02)(P-value < 10-25)

12% ± 3%

in cis


EMERGING EFFORTSRNAseq, GTEx

RNAseq questions

• Standard eQTLs – Montgomery et al, Pickrell et al Nature 2010

• Isoform eQTLs– Depth of sequence!

• Long genes are preferentially sequenced• Abundant genes/isoforms ditto• Power!?• Mapping biases due to SNPs

Strategies for transcript assembly

Garber et al. Nat Methods 8:469 (2011)

GTEx – Genotype-Tissue EXpressionAn NIH common fund project

Current: 35 tissues from 50 donors

Scale up: 20K tissues from 900 donors.

Novel methods groups: 5 current + RFA

RNAseq combined with other techs

• Regulons: TF gene sets via CHiP/seq– Look for trans effects

• Open chromatin states (Dnase I; methylation)– Find active genes– Changes in epigenetic marks correlated to RNA– Genetic effects

• RNA/DNA comparisons – Simultaneous SNP detection/genotyping– RNA editing ???

lecture 15 regulatory variation and eqtls

Documents

gene gene

gene n gene

gene expression qtls

snpgene pairs

yvert et

chesler et

humanscheung et

regulatory mechanismsciseqtls