coding’and’noncoding’sequence’...

Coding and noncoding sequence variants across cancer types

Ekta Khurana, Ph.D. Assistant Professor

Weill Cornell Medical College New York, NY

Email: [email protected]

@ekta_khurana

1

Outline

•  Coding & Noncoding sequence variants in cancer •  How to idenMfy drivers: coding variants •  Complex modes of acMon of noncoding variants •  ComputaMonal approach to idenMfy noncoding drivers: FunSeq

•  ICGC/TCGA consorMum for ‘Pan Cancer Analysis of Whole Genomes’

•  Precision Medicine

3

First cancer WGS,

Ley, Mardis, et al., Nature, 2008

~500 cancer WGS Alexandrov, et al., Nature,

2013

> 2000 WGS from ICGC/TCGA

2014

Seq Universe

TCGA (The Cancer Genome Atlas) endpoint: ~2.5

Petabytes

SRA >1 petabyte

Star forma)on 100K Genomes England ADSP

16 TB

220 TB

46 TB

31 TB

19 TB

TCGA 910Terabytes

in CGHub

NHLBI ESP

32 TB

NHGRI LSSP

ARRA AuMsm

Clinical

miRNA

9 TB

[slide courtesy of Heidi Sofia, NHGRI] 4

Genomic variants idenMfied from sequencing

5

ATGAACTGCAATTTCCAGAAGCATGCACCCTTGGAAG - - - TCTA"

ATGAACTGCAAATTCCAGAAGCATG - - - - CTTGGAAGAGTTCTA"SNP

Human Ref.

DeleMon InserMon Small Indels < 50 bp

InserMons

DeleMons

DuplicaMons Inversion

Large structural variants

Human Ref.

An average human genome contains ~3 million inherited variants.

6

Cancer cells contain thousands

of mutaZons relaZve to the normal cells

Where is Waldo ? What are the key mutaMons with funcMonal impact ?

7

Most frequently mutated genes in different cancer types

8

Source: InternaMonal Cancer Genome ConsorMum (dcc.icgc.org)

ComputaMonal methods to predict funcMonal impact of missense mutaMons

Gnad et al, BMC Genomics, 2013

Outline

•  Coding & Noncoding sequence variants in cancer •  How to idenMfy drivers: coding variants •  Complex modes of acZon of noncoding variants •  ComputaMonal approach to idenMfy noncoding drivers: FunSeq



Noncoding elements in the genome

11 Ecker, Nature – News and Views, 2012 Enhancers,

Insulators, Silencers

Cross-‐species sequence alignments

Compile all connecMons to build regulatory and enhancer-‐promoter

networks

ConnecMng regulatory elements to target coding

genes

Open chromaMn regions (DHS peaks)/

Histone modificaMons or TF

binding (ChIP-‐Seq peaks)

TF binding moMf

ubiquitous Brain-‐sp.

Heart-‐sp.

Liver-‐sp.

Brain

Heart

Liver

TGAATA

TGATTA

TGATTA

TGATTG

TGATTG

TTATTG

CGCCAT

CGCCAT

CGCCTT

CGCTTT

CGCTTT

CGCTTT

TGAGGG

TGATTA

TGATGG

Human

Chimp

Mouse

Regulatory elements idenMfied

Target gene

Very low or no signal

from funcMonal genomics assays

from evoluMonary conservaMon only

TF TF TF TF TF

Identifying cis-regulatory

regions

Khurana et al, Nature Rev Genet, under review

13

MB: medulloblastoma DLBC: B cell lymphoma STAD: gastric BRCA: breast PAAD: pancreaMc PRAD: prostate LIHC: liver PA: pilocyMc Astrocytoma LUAD: Lung adenocarcinoma

Khurana et al, Nature Rev Genet, under review

Somatic variants from WGS: most are in noncoding

regions

14

CGGAGG

CGGAAG mRNA

Gain-of-motif TF

TF 0.0

1.0

2.0WT

Mutated

Loss-of-motif

Loss-of-motif

TATCTAT X

TF

TATTTAT TF

WT

Mutated

T0.0

1.0

2.0

CGT

ATG

TGCA5G

CTA

CTG

ATT10AG

TGA

G

ACT

CGAT

ACT15GTA

C

Altered binding effects 2.0

Promoter Gene

X

•  MYB moZf created & drives TAL1 overexpression in T-‐ALL (Mansour et

al, Science, 2014) •  TERT promoter

Modes of action of noncoding variants: transcription factor binding disruption

Killela et al, PNAS, 2013 Horn et al, Science, 2013 Huang et al, Science, 2013

TERT promoter is frequently mutated in many different cancer types"

Modes of action of noncoding variants: Enhancer hijacking by genomic rearrangements in medulloblastoma (Northcott et al, Nature, 2014)

16

Modes of action of noncoding variants: ncRNAs & their binding sites

17

Figure'3D'

miRNA&binding&sites& miRNAs&

5’& 3’&

Muta4ons&in&miRNA&

binding&sites&

X

Increased&gene&expression&

CDS&mRNA&

XX

X

XX

Transla4onal&repression&

Figure'3E'

5’&

5’&

3’&

PGENE&competes&for&miRNA&binding&

PGENE&dele4on&

CDS&mRNA&

PGENE&3’&

CDS&mRNA&degrada4on&

OR'

Ribosome&

miR31 binding sites mutated, overexpression of androgen receptor

in prostate cancer"

PTEN pseudogene deletion leads to downregulation of PTEN "

Lin et al, Cancer Res, 2013 Poliseno et al, Nature, 2010 Khurana et al, Nature Rev Genet, under review

Outline

•  Coding & Noncoding sequence variants in cancer

•  How to identify drivers: coding variants •  Complex modes of action of noncoding

variants •  Computational approach to identify

noncoding drivers: FunSeq •  ICGC/TCGA consortium for ‘Pan Cancer

Analysis of Whole Genomes’ •  Precision Medicine

19

EvoluZonary conservaZon -‐  Typically defined by comparison across species

ConservaZon among humans -‐ DepleMon of common variants/Enrichment of rare variants

1

2

4

3

Estimating negative selection

Common variant

Rare variants

Enrichment of rare SNPs as a metric for negative selection

20

0.55

0.60

0.65

0.70

0.4

0.5

0.6

0.7

0.8

0.9

A

Frac

tion

of ra

re S

NPs

(non

syn)

All C

odin

g

LOF-

tol.

Rece

ssive

GW

AS

Dom

inan

t

Esse

ntia

l

Canc

er

B

Frac

tion

of ra

re S

NPs

DHS Coding

Bone

Urot

heliu

mEm

br s

tem

cel

lBr

east

Bloo

dPa

ncr d

uct

Panc

reas

Skin

Eye

Live

rM

uscle

Bloo

d ve

ssel

Hear

tLu

ngBl

astu

laM

yom

etriu

mEp

ithel

ium

Pros

tate

Brai

nEm

br lu

ngG

ingi

val

Kidn

eySp

inal

cor

dFo

resk

inCo

nnec

tive

Adip

ose

Live

rCe

rvix

Trac

hea

Sple

enPr

osta

teEs

opha

gus

Lung

Test

isPl

acen

taBl

adde

rCo

lon

Hear

tTh

yroi

dTh

ymus

Kidn

eyBr

ain

Ova

ry

(rare=derived allele freq < 0.5%)

•  Depletion of common polymorphisms in regions under selection

Negative selection restricts the allele frequency of deleterious mutations.

•  Results for coding genes consistent with known phenotypic impacts

•  Other metrics for selection •  EvoluMonary conservaMon

(e.g. GERP) •  SNP density

(confounded by mutaMon rate)

LOF-‐tol (Loss-‐of-‐funcZon tolerant): least negaZve selecZon Cancer: most selecZon Khurana et al., Science, 2013

Differential selective constraints among sub-categories of noncoding elements

21 Khurana et al., Science, 2013

Negative selection and tissue-specificity of coding and noncoding regions

22

q Ubiquitously expressed genes and bound regions show stronger selection

q Differences in constraints amongst tissues q Constraints in coding genes and regulatory genes are

correlated across tissues

Specific categories with statistically significant enrichment of rare SNPs: randomization and multiple

hypothesis correction •  Generate null for each category by sliding elements across the

genome 1000 times

•  Compare the actual fraction of rare variants in the category to null distribution and call it under significant selection accepting a low type 1 error rate (α=0.2%)

•  Total number of categories discovered falsely under selection (False discovery rate, FDR)

E(R0), Expectation of false positives amongst all categories tested = α*RA

RA = Number of categories tested R = Number of categories identified to be under selection

At α=0.2%, FDR = 1.3% •  Using these criteria, 102 categories are under significantly strong

selection 23

Category 1 Category 2 Category 3 Category 1’ Category 2’ Category 3’

FDR = E(R0 )

R

Which noncoding categories are under very strong “coding-like” selection ?

q  Top categories among 700 categories

q  Binding peaks of some general TFs (eg FAM48A)

q  Core moMfs of some TF families (eg JUN, GATA)

q  DHS sites in spinal cord and connecMve Mssue

~0.4% genomic coverage (~ top 25)

~0.02% genomic coverage (top 5)

~400-‐fold

~40-‐fold

Enrichment of know disease-‐causing mutaMons from Human Gene MutaMon

database

Khurana et al., Science, 2013 24

Building human regulatory network from linear noncoding annotations

Peak Calling (ChIP-Seq) Assigning TF binding sites to targets Filtering high confidence edges

~28K proximal edges

PotenMal Distal Edge

Strong Proximal Edge

TF TF

Nodes 119 TFs and ~9000

target genes Edges

28,000 interacMons

Using correlaMon with expression data Gerstein¶…..Khurana¶…., Nature, 2012 (¶ co-first authors) Yip et al, Genome Res, 2012 Fu et al, Genome Biol, 2014 25

Distal edges for ~17K genes

26

IdenZficaZon of noncoding candidate drivers among somaZc variants: FunSeq

Khurana et al., Science, 2013

•  Core score –  Each feature gets a weight

FunSeq2: weighted scoring scheme

27 Fu et al., Genome Biology, 2014

•  Feature weight -‐ Weighted with mutaMon paperns in natural polymorphisms

(features frequently observed weighed less) -‐ entropy based method

!! = 1+ !!!"#!!! + 1− !! !"#! 1− !! !!!

!!"#$%! = ! !!!!!!!"!!"#$%&$'!!"#$%&"'!

28

HOT region

SensiMve region

Polymorphisms

Genome

p = probability of the feature overlapping natural polymorphisms

Feature weight:

For a variant:

wdp

Identification of noncoding candidate drivers among somatic variants: Example"

Validation of a candidate driver identified in prostate cancer sample in WDR74 gene promoter!!q  Sanger sequencing in 19 additional samples confirms the

recurrence "

""""""""""""""""""""

q  WDR74 shows increased expression in tumor samples"

benign" PCs"

29

IdenZfied candidate drivers with high FunSeq scores in 607 whole-‐genomes

Type Sample Coding (all samples) Noncoding (all samples)

ALL 1 17/87 3/7653

AML 7 13/87 10/3325

Breast 119 1430/6495 3128/638835

CLL 28 41/338 124/51406

Liver 88 1464/6257 3188/843489

Lung_Adeno 24 2255/9479 5291/1428263

Lymphoma_B 24 241/1212 529/126186

Medulloblastoma 100 282/1461 226/123387

Pancreas 15 129/965 179/111044

PilocyMc_A 101 13/103 15/10453

Stomach 100 5739/20374 10829/1891465

30

•  Can be further prioriMzed, e.g. gene expression, luciferase reporter assays, etc.

Fu et al, Genome Biology, 2014 Data from Alexandrov et al, Nature, 2013

Outline


•  ICGC/TCGA consorZum for ‘Pan Cancer Analysis of Whole Genomes’


ICGC/TCGA collaboraZve effort called ‘PCAWG’

•  Pan Cancer Analysis of ~2500 whole genomes (tumor-‐normal pairs)

•  ~1500 samples have RNA-‐Seq data •  ~1400 samples have methylaMon data •  ~25 cancer types •  All the processed data and variant calls will be available to public as a resource once the project is completed

Working groups of PCAWG •  SomaMc mutaMon calling, structural variant calling •  MutaMons (drivers) in noncoding regions, integrate networks &

pathways •  IntegraMon of transcriptome & genome •  IntegraMon of epigenome & genome •  MutaMonal signatures & processes •  Germline cancer genome •  EvoluMon & heterogeneity •  Molecular subtypes & classificaMon •  Mitochondrial •  Pathogens •  Portals & visualizaMon •  TranslaMon to the clinic

Outline




What is precision medicine ?

•  Drugs for specific mutaMons in paMent’s cancer

•  Example: BRAF gene is frequently mutated in melanoma (V600E). Vemurafenib is a BRAF inhibitor and received FDA approval in 2011.

Source: "3OG7" by Boghog2 -‐ Own work. Licensed under Public Domain via Wikimedia Commons

35

Flaherty KT et al. N Engl J Med 2010;363:809-‐819.

Current challenges of precision medicine

•  No drugs available for many mutated proteins •  Resistance to drugs

37

38

Acknowledgements Yale

Yao Fu, Xinmeng Mu (Broad), Jieming Chen, Lucas Lochovsky, Arif Harmanci, Alexej Abyzov

(Mayo), Suganthi Balasubramanian, Cristina Sisu, Declan Clarke, Mike Wilson, Yong Kong, Mark

Gerstein

Sanger Vincenza Colonna, Yuan Chen, Yali Xue, Chris

Tyler-Smith

Cornell Steven Lipkin, Jishnu Das, Robert Fragoza,

Xiaomu Wei, Haiyuan Yu

WCMC

Andrea Sboner, Dimple Chakravarty, Naoki Kitabayashi, Vaja Liluashvili,

Zeynep H. Gümüş, Mark A. Rubin

In FunSeq2 Zhu Liu, Shaoke Lou, Jason Bedford, Kevin Yip

U of Michigan Hyun Min Kang U of Geneva

Tuuli Lappalainen (NYGC), Emmanouil T. Dermitzakis

Baylor Daniel Challis, Uday Evani, Donna

Muzny, Fuli Yu, Richard Gibbs EBI

Kathryn Beal, Laura Clarke, Fiona Cunningham, Paul Flicek, Javier Herrero, Graham R. S. Ritchie

Boston College Erik Garrison, Gabor Marth

Mass Gen Hospital Kasper Lage, Daniel G. MacArthur,

Tune H. Pers Rutgers

Jeffrey A. Rosenfeld

FuncZonal InterpretaZon

Group ~50 people

Khurana lab •  Priyanka Dhingra

(Postdoctoral associate) •  RecruiZng more

postdocs and students

coding’and’noncoding’sequence’...

Documents