instructions for viewers - sciencemag.org slides_broad ngs... · brought to participating experts...

Instructions for Viewers

• To share webinar via social media:

• To see speaker biographies, click: View Bio under speaker name

• To ask a question, click the Ask A Question button under the slide window

• To share webinar via e‐mail:

Advancing the promise of genomic medicineMay 12, 2015

Webinar Series

Sponsored by:

Participating ExpertsBrought to you by the Science/AAAS Custom Publishing Office


Webinar Series

Sponsored by:

Daniel MacArthur, Ph.D.Massachusetts General HospitalBoston, MA

Eliezer Van Allen, M.D.Dana-Farber Cancer InstituteBoston, MA

Using big data to improve thediagnosis of rare disease

Daniel MacArthurBroad Institute of Harvard and MIT Massachusetts General Hospital

www.macarthurlab.org | Twitter: @dgmacarthur

Geography of my lab

Traditional vs genomic diagnosisStandard diagnosis• Muscle biopsy/pathology

• Sequencing: ACTA1, TPM2, TPM3

• Linkage analysis to exclude NEBmutations

Time: 9 years

Genomic diagnosis• Exome sequencing identifies compound het mutations in new disease gene (since confirmed causal)

Time: 7 weeksSuccess rate: 40‐60%

~50% of muscle disease patients lack genetic diagnosis

...ACCCGTGGACACTTCGGATACACC...~3 billion bases: ~$5,000

...ACCAGTGGACACTTCGGTAAGTCC...~30 million bases: ~$600

~$1,800

Making sense of one genome requires tens of thousands of genomes

vs

• Over 500K exomes, 50K genomes have been sequenced globally

• Challenge: these data are siloed and inconsistently processed

Exome Aggregation Consortium (ExAC): aggregating and calling 92,000 exomes

Subset of 60,706 “reference” samples

All data reprocessed

with BWA/Picard

Joint calling across all samples

with GATK 3 Haplotype

Caller

Daniel MacArthurDavid AltshulerDiego ArdissinoMichael BoehnkeMark DalyJohn DaneshRoberto ElosuaGad GetzChristina HultmanSekar KathiresanMarkku LaaksoSteven McCarroll

Mark McCarthyRuth McPhersonBenjamin NealeAarno PalotieShaun PurcellDanish SaleheenJeremiah ScharfPamela SklarPatrick SullivanJaakko TuomilehtoHugh Watkins

ExAC Principal Investigators1000 GenomesBulgarian TriosFinland-United States Investigation of NIDDM Genetics (FUSION)GoT2DInflammatory Bowel DiseaseMETabolic Syndrome In Men (METSIM)Myocardial Infarction Genetics Consortium:

• Italian Atherosclerosis, Thrombosis, and Vascular Biology Working Group

• Ottawa Genomics Heart Study• Pakistan Risk of Myocardial Infarction

Study (PROMIS)• Precocious Coronary Artery Disease

Study (PROCARDIS)• Registre Gironi del COR (REGICOR)

NHLBI-GO Exome Sequencing Project (ESP)National Institute of Mental Health (NIMH) ControlsSIGMA-T2DSequencing in Suomi (SISu)Swedish Schizophrenia & Bipolar StudiesT2D-GENESTaiwanese TriosThe Cancer Genome Atlas (TCGA)Tourette Syndrome Association International Consortium for Genomics (TSAICG)

Contributing cohorts

Monkol LekEric MinikelKaitlin SamochaMenachem FromerDoug RuderferPradeep NatarajanRon DoAndrew HillJames WareAdam Kiezun

AnalysisMonkol LekFengmei ZhaoRyan PoplinEric BanksTimothy Fennell

Production

Konrad KarczewskiBrett ThomasBen Weisburd

Website

Broad Genomics Platform

Scalable pipeline for joint calling

Monkol Lek

EuropeanLatinoSouth AsianEast AsianAfricanOther

Monkol LekPrincipal component analysis using 5,800 common SNPs

Laramie Duncan

010

000

2000

030

000

4000

050

000

6000

0

Indi

vidu

als

with

Exo

me

Sequ

ence

Dat

a East AsianSouth AsianEuropeanMiddle EasternAfricanNative American ancestryDiverse Other

1000 Genomes

ESP ExAC World

Increase in size and diversity

LatinoAfricanEuropeanSouth AsianEast AsianOther

Catalogue of protein‐coding variation• Largest ever collection of human protein‐coding genetic variants– 3,296,522missense alleles: 99% with

frequency <1%, 52% seen only once!– 232,094 candidate loss‐of‐function alleles

affecting 16,007 protein‐coding genes– 21,204 previously reported severe disease‐

causing mutations from HGMD

The ExAC browser

KonradKarczewskiexac.broadinstitute.org

Frequency‐based filtering

Eric Minikel, James Ware

• Number of variants remaining in an exome after applying a 0.1% filter

• Best‐performing filter is popmax: variant must be <1% in all outbred populations

• Both size and ancestral diversity matter

East AsianSouth AsianLatinoAfricanEuropean

Identifying genes with significant depletion of variation

• Built a mutational model that allows us to predict the number of variants in a given functional class we should expect to see in each gene in a given number of people (Samocha et al. 2014 Nat Genet 46:944–950)

Obs

erve

d

Expected

Synonymous Missense Loss-of-functionR = 0.9778 R = 0.9482 R = 0.5866

Expected Expected

Empirical identification of genes subject to strong human constraint

• 4,728 genes significantly depleted for LoF– 1,450 have <10% of expected LoF

• enriched for disease genes, but most have no known LoF phenotype, many unknown function

0.97

0.33

0.03

1.02

0.55

0.01

Applying whole‐genome and transcriptome sequencing to exome‐undiagnosed muscle

disease families

Finding mutations in tricky cases

familyexome

success PCR-free WGSmuscle RNA-seq

(~500 families)

(~50 families)

PCR‐free WGS is amazing

DMD →

← DMD

Normal

Patient

A

A

B

B

Intergenictranscription

1.95Mb

Case 1: DMD disruption by a >2 Mb inversion

RNA sequencing of patient skeletal muscle confirms aberrant splicing:

Heterozygous exon skipping

Case 2: RNA‐seq detects exon‐skipping

• 14‐year‐old male with limb‐girdle muscular dystrophy presents with candidate splice mutation in novel candidate gene

Key messages• Exciting times in clinical genomics: new technologies and reference databases are transforming diagnosis

• Joint‐calling of tens of thousands of exomes is possible, and powerful

• When exomes fail, whole genome and transcriptome sequencing are strong follow‐up technologies

Thanks!

Nigel ClarkeKathryn NorthSandra CooperLeigh WaddellEmily OatesRoula GhaouiGina O’GradyMichaela Kreissl

Kate BushbyVolker StraubHans LochmullerAna TopfMonica EnsiniCathy TurnerElizabeth HarrisRachael ThompsonSteve Laval

Carsten BonnemannSandra DonkervoortNathan Bradley

Peter KangLou KunkelAlan BeggsElicia EstrellaHema K.R Muntra

Volker StraubMonica EnsiniAna Topf

Namrata GuptaLauren MargolinAlicia TangGenomics PlatformNHGRI Sequencing Grant

Monkol LekKaitlin SamochaEric MinikelAnne O’Donnell LuriaTaru TukiainenAndrew HillBeryl CummingsFengmei ZhaoBen WeisburdMark Daly

Exome Aggregation Consortiumexac.broadinstitute.org/about

Browser: Konrad Karczewski

We’re hiring:macarthurlab.org/jobs

Participating ExpertsBrought to you by the Science/AAAS Custom Publishing Office


Webinar Series

Sponsored by:



Expanding precision cancer medicine through clinical genomic

profilingEliezer (Eli) Van Allen, MD

Dana-Farber Cancer InstituteCenter for Cancer Precision Medicine

Broad Institute of MIT and HarvardHarvard Medical School

May 12, 2015

Disclosures

• Consultant– Syapse– Roche Ventana

Outline

• Rationale for clinical cancer sequencing• Analysis and interpretation• Knowledge sharing• Trials and regulation

Precision cancer medicine: A paradigm shift

Cancer Non-specific chemotherapy

Cancer Targeted therapy

Past Present

Present Future

Test for tumor-specific genetic targets that can

be “drugged”

EGFRBCR-ABLALKHER2…

erlotinibimatinibcrizotinibtrastuzumab…

Precision cancer medicine: A paradigm shift

Past

Present

Prospective clinical sequencing

PATIENTS TISSUE PROFILING

INTERPRETATION

COMMUNICATION DECISION-MAKING OUTCOME

MetastaticClinical trials(Neo)adjuvant

FrozenFFPE

CTCs/other

HotspotTargetedExome

The genomics clinical data challengeData points per patient

10

100

1,000

10,000

100,000

Source: NHGRI

History and PhysicalLabs, ImagingPathology…

Hotspot genotyping

WES, WGSTranscriptome

Targeted panels

A role for prospective clinical WES?

• Relevant territory already covered?

• Flexible for new discoveries

• Retrospective cohort analyses for discovery

WES rationaleChallenges

ERCC2 mutations and cisplatinsensitivity in bladder cancer

Van Allen, Mouw et al, Cancer Discovery 2014

A role for prospective clinical WES?

• Relevant territory already covered?

• Cost?

• Analytical considerations?

• Flexible for new discoveries

• Retrospective cohort analyses for discovery

• Approaching equivalence

• Germline, SCNAs, mutation rate, neoantigens, etc.

WES rationaleChallenges

“Unmatched” cancer panels and germline variants

Jones et al, Science Trans Med 2015

Neoantigens, passengers, and cancer immunotherapies

Rizvi, Hellmann, et al Science 2015

DCB: Durable clinical benefitNDB: No durable benefit

Outline


PHIAL: Precision Heuristics for Interpreting the Alteration Landscape

Van Allen, Wagle et al. Nature Medicine 2014

PHIAL applied to 14 patients

Germline WES interpretation algorithm

Cancer Gene

Census

PharmGKB

GET-PGP/HG

MD

Somatic Data

Cancer-Specific SNPs

Cancer-Pharm SNPs

Other Pharm SNPs

General SNPs

Somatic-Specific SNPs

Cancer-Specific SNPs

Cancer-Pharm SNPs

Other Pharm SNPs

General SNPs

Somatic-Specific SNPs

Population Frequency and Database Cross-Reference

Ranked Variants For Committee Review

General and Context-specific Variant Filtering

Filter Sort Review

WES germlineSNPs and indels

Incidental findings

• 56 “can’t miss” genes• Spans disease types (predominantly non-

cancer)• New challenges in the advanced setting

Germline and treatment decision-making

Tumor-onlyvariants

Germlinevariant

PHIAL

Metastatic CRPC

PARP inhibitor Platinum chemotherapy

PI3K inhibitorAKT inhibitor

mTOR inhibitor

BRCA2 K3326* nonsense

Responds to cisplatin

Van Allen, et al Prostate Cancer Prostatic Dis. 2014

PHIAL and CLIA cancer WES

DNAFFPE

Whole BloodCells

Tissue

Whole Exome LC & Sequencing

Picard Alignment

GATK Variant Calling

MuTect & Indelocator

Germline calling of Normal

Somatic calling

PHIAL Report

Outline


Clinically focused cancer variant knowledgebase

• MyCancerGenome: www.mycancergenome.org• Personalized Cancer Therapy: pct.mdanderson.org• Drug Gene Interaction: dgidb.genome.wustl.edu• Others

• Breadth and depth need for clinical WES• Challenge of scale, entry point, and rapidly evolving

clinical relevance

Crowdsourcing variant knowledge?

Gad GetzMike LawrenceTed LiefieldEila Arich-Landkof

www.tumorportal.org

www.tumorportal.org

Crowdsourcing variant knowledge?

Outline


A rapidly evolving landscape

• Stressing the current clinical trial infrastructure– Genotype-driven trials?– Basket trials?– Access to emerging therapies in alternative

indications?• Regulatory requirements for testing labs• Medico-legal implications

Center for Cancer Precision Medicine

Levi A. GarrawayJudy Garber

Gregory KryukovSteve JoffeStacy GrayPasi Janne

Nikhil WagleNelly Oliver

David BarbieKwok-Kin WongGlenn DranoffPhilip Kantoff

Mary-Ellen TaplinSteve Hodi

George DemetriNeal LindemanLynette ShollScott Rodig

Many others…

Broad CGA

Gad GetzAdam Keizun

Manaswi GuptaMany others…

Clinical computationaloncology team

Ali Amin-MansourAndrea Garofalo

William GibsonMichal Barziny-Rokni

Diana MiaoTravis ZackJasmine Mu

David Liu

Let’s work together!

vanallenlab.dana-farber.org@vanallenlab

Funding

Participating ExpertsBrought to you by the Science/AAAS Custom Publishing Office To submit your

questions, click theAsk a Question

button


Webinar Series

Sponsored by:



For related information on this webinar topic, go to:

Look out for more webinars in the series at:

webinar.sciencemag.org

To provide feedback on this webinar, please e‐mailyour comments to [email protected]

genomics.broadinstitute.org

Brought to you by the Science/AAAS Custom Publishing Office


Webinar Series

Sponsored by:

instructions for viewers - sciencemag.org slides_broad ngs... · brought to participating experts...

Documents