Applications of Human Genome Sequencing: Disease and

Personal Genomics

Conflicts: Personalis, Genapsys, AxioMx, Novartis

Michael Snyder

February 3, 2015

Outline •  Quick review on human variation •  Disease genome sequencing

–  Cancer, Mystery Diseases

•  Personal genomics

Genetic Variation Among People: Three

Types

3.7 Million/person

2) Short Indels (Insertions/Deletions 1-100 bp)

GATTTAGATCGCGATAGAG GATTTAGATCTCGATAGAG

1) Single nucleotide variants (SNVs)

GATTTAGATCGCGATAGAG GATTTAGA------TAGAG

300-600K/person

Structural Variation (SV): Large Blocks (1000bp) of DNA that are Deleted, Inserted or Inverted

-" People Have >3,000 differences (>2.0kb) Relative to the Reference Human Genome Sequence

-" Likely responsible for much human differences and disease

Heterogeneity in Olfactory Receptor Genes (Examined 851 OR Loci)

CNVs affect: 93 Genes 151 ψgenes

The Cost of DNA Sequencing is Dropping

Human Genome Cost <$2Khttp://www.genome.gov/

Examples of People Who Have had Their Genomes Sequenced

From: www.genciencia.com

Jim Watson Craig Ventor Ozzy Osbourne

sciencewithmoxie.blogspot.com.au/2010_11_01_archive.html

•  Understand and Treat Disease –  Cancer–  Mystery diseases–  Prenatal diagnostics

•  Managing Health Care in Healthy Individuals?

Impact of Genomics on Medicine

Cancer Genome Sequencing 1)  Cancer is a genetic disease: both inherited and

somatic

2) TCGA-like projects have sequenced >10,000 cancer –Many types Ovarian, colon, breast, bladder, par=ncreatic ….

3) Each cancer contain many mutations

4) 10-20 “driver” mutations

Results of cancer genomics

10-100,000 SNVs for many cancer types + Structural variants (large insertions, deletions, fusions)

~254 cancer genes have been identified

Most common genes are already known e.g. p53, RB1, KRAS, BRCA1/2, etc

B Vogelstein et al. Science 2013

Cancer Genome Sequencing

Vogelstein et al., March Science, 2013

5) Each cancer is unique

6) Often fall into common pathways

TCGA-identified Pathways in Ovarian Cancer

12

RB and PI3K/AKT Signaling Homologous Recombination

Notch Signaling

12

RB and PI3K/AKT Signaling Homologous Recombination

Notch Signaling

Cancer genome sequencing Sequence genomes (cancer tissue and normal) find genetic changes and suggest possible therapies

Normal (~30X)

Cancer (>60X)

VS

Image: Bruce Blaus Image: National Cancer Institute

GATTTAGATCGCGATAGAG GATTTAGATCACGATAGAG

Initially; Now >200X)

Copy number alterations from metastatic Colon tumor

Chromosome 7: Two amplification regions

Chr 7p arm Chr 7q arm Log 2 ratio -

genomic copy

number

Chromosome 7 coordinates (Mb) NCBI 37

•" Two regional amplifications with complex genomic structure.

•" Both loci > 10 copy number showing statistical

significance.

CEN

With Hanlee Ji

CML (Chronic Myeloid Leukemia) - Gleevac

1)  Chromosome Translocation involved Abl Protein Kinase and bcr

2)  Gleevac—binds active site of Abl and inhibits it; causes remission

3)  Gleevac binds other kinases (Kit) that are mutated in other cancer: Gleevac can be used to treat them.

Pharmacogenomics—Matching Drugs to Disease !

Her2 Mutated in 25% of

Breast Cancer

Iressa®: –" Used to treat nonsmall

cell lung carcinoma–" 10% of patients

respond –" Works only if patients

tumors have very specific mutations in EGFR

Matching drugs to cancer types - BRAF BRAF: Oncogene mutated in

~65% of melanoma, and more ~15-20% of colon cancers. Vast majority of BRAF mutations are at V600 (most commonly V600E).

Vemurafenib®, PLX4032®:

E!ective BRAF-inhibitors specific for V600 mutations. (Few side-e!ects) Resistance frequently

arises.

Bollag et al. Nature 2010

Challenges associated with cancer genome sequencing

Identification of the “Driver mutations is difficult; thousands of variants; only a few “drive” cancer cell growth

- Focuses on known genes - Focuses on protein coding regions

Tumor heterogeneity Drug resistance usually appears Drug combinations

Other omes: RNA-Seq •  Identifies the expressed genes i.e. potential drivers

Example: Esophageal cancer Deep genome sequencing >20,000 SNVs & Indels >1000 Structural Variants

1/3 of synonymous variants expressed as RNA

Other omes: Circulating Tumor DNA

•  CAPP-Seq Targeted search for

somatic mutations

•  Biotinylated oligos 139 recurrent mutated genes in NSCLC

•  96% specificity for mutant allele fractions down to ∼0.02%

(Newman et al 2014) Nat. Med.

What About Cancer Predisposition?

Identifying “heritable” causes of cancer

From Jim Ford

APC FANCE PMS2

ATM FANCF PRSS1

BLM FANCG PTCH1

BMPR1A FANCI PTEN

BRCA1 FANCL RAD51C

BRCA2 LIG4 RET

BRIP1 MEN1 SLX4

CDH1 MET SMAD4

CDK4 MLH1 SPINK1

CDKN2A MLH2 STK11

EPCAM MSH6 TP53

FANCA MUTYH VHL

FANCB NBN

FANCC PALB2

FANCD2 PALLD

42 Gene Panel Breast Cancer Study

From Jim Ford

Multiple Gene Panel Test Results

•  36 of 361 women without B1/2 (10%) carried a potentially pathogenic mutation –  ATM, BLM1, CDH1, CHEK2, MLH1, MSH2, MSH6, PMS2,

MUTYH, NBN, PRSS1, SLX4, RAD51C, PALB2, BRIP1 •  Participants notified of significant results

–  11/14 from initial 200 followed-up, confirmed, counseled •  Variants of Uncertain Significance (VUS) were common

–  40% of patients (0.7 per patient)

Kurian, Ford et al. Journal of Clinical Oncology 2014 Ford, Montreal BRCA Conference 2014

Mendelian Conditions: Undiagnosed Mystery Diseases •  0.4% of live births •  8% of adults have genetic disorder

recognized by adulthood •  25 M US Citizens •  $5M/individual/lifetime

Ng et al., 2010 Nat. Genetics

Family with Charcot-Marie Disease

•  Neuropathy –  Heterogenous disease—many different genes mapped

•  Sequence genome to 30X coverage •  3.4 M SNPs: (561,719 novel)

–  2,255,102 in intergenic –  1,165,204 in genes, introns etc.

•  174 nonsynonymous SNPs in region of interest •  54 related to Neuropathies •  Ultimately zoomed in on SH3TC2 gene:

Full blown disease has two mutations: Y169H (missense), R954X (nonsense)* Single heterozygotes have some phenotypes

*Implicated previously Lupski et al 2010 NEJM

Solving Mystery Diseases: Dizygotic Twins:

Dopamine Responsive Dystonia

•  Constantly sick, colicky, failed to meet milestones “floppy”; MRI showed some abnormalities

•  Children diagnosed with dystonia

•  Trial of L-DOPA showed dramatic improvement in 2 days

•  Sequenced genomes-found mutation in SPR Gene

•  Administered dopamine + seratonin precursor

From Richard Gibbs, Baylor

X

Solving Mystery Diseases: Child With Variety of Conditions

Developmentally Delayed, Significant Health Issues

F M

A1

Mother SNVs: 3,125,880 Private: 581,754 Indels: 723,379

Father SNVs: 3,119,588 Private: 596,691 Indels: 750,522

Child SNVs: 3,118,638 Private: 33,158 Indels: 673,809

SNVs: Single nucleotide variants Indels: = Insertions/deletions (~<100bp)

Candidates: TCP10L2, SUPV3L1, PIEZO1 DNAH2, NGLY1, FANCA, WFS1

Lessons Learned

1)  Overall success rate for identifyng causative mutations is low 25%

2)  Information not always directly actionable but still valuable.

3)  Best success with a)  Specific phenotypes b)  Large families

4)  Need large database to share information: Recurrence is key. ClinVar

Fetal DNA Sequencing 1)  Cell free fetal DNA can be detected in maternal

blood as early a 4-5 weeks gestation

2)  4-13% circulating DNA is fetal à increases with pregnancy

3)  Targeted detection of mutations

4)  Whole genome sequencing routinely used to detect trisomies: Down’s (Chr. 21), Chromosome 18 and Chromosome 13. 99% sensitivity

5)  Taking over from aminocentesis

Fetal DNA Sequencing

Srinivasan A, Bianchi DW, Huang H, et al. Am J Hum Genet 2013; 1–10.

Personal Genome Sequencing:

Can genome sequencing of a healthy person be useful in health care?

•  Pharmacogenomics–  Drug doses and side effects

•  Predict disease risk

•  Catch disease early

Impact of Genomics on Medicine

Genome Interpretation Pipeline

Sequence Genome (Illumina, Complete Genomics)

Call SNVs, Indels, SVs Phase Variants

Disease Gene (HGMD)

Risk Variants

Disease Gene Complex/Common

Disease Risk Genotypes

Specific Disease Variants

Pharmaco-genomics

Complex/Complex/Complex/Common Pharmaco

Specific

Genome

Transcriptome (mRNA, miRNA, isoforms, edits)

Proteome

Metabolome

Personal Omics Profile

Autoantibody-ome

Personal “Omics” Profiling (POP)

Cytokines

Epigenome

Initially 40K

Molecules/Measure-

ments

Now Billions! Microbiome (Gut, Urine,

Nasal, Tongue, Skin)

Personal Omics Profile 58 months; 92 Timepoints; 7 Viral Infections

/

/

Chen et al., Cell 2012

Accurate Genome Sequencing

3.3 M Hi conf. SNVs, 217K Indels and 3K SVs 2 or more Platforms

(Plus low confidence)

Whole Genome Sequencing •" Complete Genomics: 35 b paired ends (150X) •" Illumina: 100 b paired ends (120X)

Exome Sequencing •" Nimblegen •" Illumina •" Aglilent

3.30M 89%

100K 2%

345K 9%

CG Illumina

Coding Non-Coding

miRNA Splice UTR

miRNA targets

Seed sequence SIFT PP2

OMIM/Curated Mendelian disease

(51)

Nonsynonymous (1320)

Synonymous

mRNA stability

tRNA rate

Approach I: Known Disease Risk Pipeline

Rick Dewey & Euan Ashley

Damaging (234)

All variants ~3.5M

Rare/novel variants (<5%)

(1)

Curated List of Rare Variants (SNVs, All heterozygous)

Missense •  ALAD, ABCC2, ACADVL, ADAMTS13, AGRN, BAAT, CDS1,

CHD7, COL4A3, CTSD, DGCR2, DLD, DYSF, EPCAM, FGFR1OP, FKRP, GAA, GNAI2, HSPB1, IGKC, ITPR1, MED12, MKS1, NTRK1, PCM1, PKD1, PLEKHG5, PMS2, PRSS1, PTCH2, SERPINA1, SETX, SYNE1, TERT, TTN, VWF, ZFPM2, PNPLA2.

Bolded Genes expressed in PBMC (RNA).

Nonsense •  PRAMEF2, PLCXD2, NUP54, RP1L1, PIK3C2G,

NDE1, GGN, CYP2A7, IGKC

Not Rare But Important •  KCNJ11 , KLF14, GCKR …

Missense •" ALAD, ABCC2, ACADVL, ADAMTS13, AGRN, BAAT, CDS1,

CHD7, COL4A3, CTSD, DGCR2, DLD, DYSF, EPCAM, FGFR1OP, FKRP, GAA, GNAI2, HSPB1, IGKC, ITPR1, MED12, MKS1, NTRK1, PCM1, PKD1, PLEKHG5, PMS2, PRSS1, PTCH2, SERPINA1, SETX, SYNE1, TERT, TTN, VWF, ZFPM2, PNPLA2.

Nonsense •" PRAMEF2, PLCXD2, NUP54, RP1L1, PIK3C2G,

NDE1, GGN, CYP2A7, IGKC

Not Rare But Important •" KCNJ11 , KLF4, GCKR ! Not Rare But Important

KCNJ11 , KLF4, GCKR

Diabetes

Not Rare But Important !

Not Rare But Important KCNJ11 , KLF4, GCKR KCNJ11 , KLF4, GCKR KCNJ11 , KLF4, GCKR

High Cholesterol !!KCNJ11 , KLF4, GCKR High Cholesterol

PRSS1, PTCH2, SERPINA1, SETX, SYNE1, TERT, TTN, MKS1, NTRK1, PCM1, PKD1, PLEKHG5, PMS2,

PRSS1, PTCH2, SERPINA1, SETX, SYNE1, TERT, TTN,

Aplastic Anemia

Rare Variants in Disease Genes (51 Total)

A B

C

D

E

F

Figure 2. Medical Findings(A) High interest disease- and drug-related variants in the subject’s genome.

(B) RiskGraph of the top 20 diseases with the highest posttest probabilities. For each disease, the arrow represents the pretest probability according to the

subject’s age, gender, and ethnicity. The line represents the posttest probability after incorporating the subject’s genome sequence. Listed to the right are the

numbers of independent disease-associated SNVs used to calculate the subject’s posttest probability.

(C) RiskOGram of type 2 diabetes. The RiskOGram illustrates how the subject’s posttest probability of T2D was calculated using 28 independent SNVs. The

middle graph displays the posttest probability. The left side shows the associated genes, SNVs, and the subject’s genotypes. The right side shows the likelihood

ratio (LR), number of studies, cohort sizes, and the posttest probability.

(D) Blood glucose trend. Measurements were taken from samples analyzed at either nonfasted or fasted states; the nonfasted states (all but days 186,

322, 329, and 369 and after day 400) were at a fixed time after a constant meal. Data was presented as moving average with a window of 15 days. Red

1298 Cell 148, 1293–1307, March 16, 2012 ª2012 Elsevier Inc.

Rong Chen and Atul Butte 0% 100%

* *

*

* Known *

Approach II: Complex Disease Risk Profile Using VariMed

VariMed: Highly curated Database of genetic associations with complex disease

Sum over likelihood ratios for disease risk

Disease risk profile T2D

HNF1B rs4430796 GG

MTNR1B rs10830963 CC

rs17036101 GG

rs4607103 CC

THADA rs7578597 TT

rs1153188 TA

rs12779790 AG

rs5945326 AA

TP53INP1 rs896854 TC

EPO rs1617640 AA

WFS1 rs10010131 GG

rs9300039 CC

RBMS1 rs1020731 GA

JAZF1 rs864745 TC

ARAP1 rs1552224 AA

KCNQ1 rs231362 GG

rs4457053 GG

rs7578326 AA

PPARGC1A rs2970847 TC

rs1111875 TT

KCNJ11 rs5219 TT

IGF2BP2 rs4402960 GT

SLC30A8 rs13266634 CT

CDKAL1 rs7754840 GG

FTO rs8050136 CC

KCNQ1 rs2237892 CT

rs10811661 TC

TCF7L2 rs7903146 CT

PrevalenceGenotype Test

1.13 1 11320 46%

0.94 1 16061 43%

1.02 1 89920 44%

1.04 1 89920 44%

1.03 1 89920 43%

0.96 1 89920 42%

1.06 1 89920 43%

1.09 1 94337 42%

1.01 1 94337 40%

1.48 2 4011 39%

1.07 2 30248 31%

1.05 2 42170 29%

0.95 2 84605 28%

1.00 2 89920 29%

1.03 2 94337 29%

1.07 2 94337 29%

1.12 3 94337 27%

1.07 3 94337 25%

1.31 4 5558 24%

0.88 6 93188 19%

1.15 7 87066 21%

1.06 8 104401 19%

0.94 9 145718 18%

0.91 10 51327 19%

0.87 10 63470 20%

0.80 13 6570 23%

0.85 18 154141 27%

1.18 49 140717 30%

27% LR Studies Samples Probability

Type 2 diabetes

10% 50% 100% Rong Chen and Atul Butte

!"#$ %&& %'& (&& ('& )&& )'& '&& ''& *&& *'&& "#$

!"#$%&'()*+,-./(

012(3#*4'5()6'"1789'(7%(:&7(;<='$78%</(

+&,&

-&&--&-%&-(&-)&-'&-*&

.-'&

!"#$%&'()*+,%

GLUCOSE LEVELS

HRV INFECTION (DAY 0-21)

RSV INFECTION (DAY 289-311)

LIFESTYLE CHANGE (DAY 380-

CURRENT) 44

HbA1c (%): 6.4 6.7 4.9 5.4 5.3 4.7 (Day Number) (329) (369) (476) (532) (546) (602)

Genome Analysis of 12 Healthy PeopleDewey, Grove, Pan, Ashley, Quertermous et al JAMA 2012

Ethnicity:7 Asians5 Europeans

Sequence genomes with Illumina (all 12; Mean depth: 50X (38-62)

9 also sequenced with Complete Genomics

Inherited Disease Risk and Carrier Status

# Variants Per Subject Median (Range)

Candidate Variants Manually Curated - Previously reported or potential pathogenic variants in ACMG genes

108 (90-127) 3 (1-7)

Reportable variants associated with disease risk (HGMD) - Reported disease-associated variants - Rare expected pathogenic variants - Genetic variants of unknown significance

5 (2-6)

0 (0-2) 0 (0-1) 3 (1-6)

Reportable variants associated with carrier status - Reported disease-associated variants - Rare expected pathogenic variants - Genetic variants of unknown significance

13 (8-18)

2 (0-4) 2 (1-4) 9 (4-12)

Study of 12 Healthy PeopleDewey, Grove, Pan, Ashley, Quertermous et al

-  3 followup diagnostic tests (range 0-10)-  Cost ~$400-$1400 per individual (median $663-

$773

-  54 minutes per variants

-  One individual had a BRCA1 nonsense mutation—no known family history

Gene SNP Patient genotype Drug(s) affected

CDKN2A/2B rs10811661 C/T Troglitazone (increased beta-

cell function)

CYP2C19 rs12248560 C/T Clopidogrel (increased activation)

LPIN1 rs10192566 G/G Rosiglitazone (increased effect)

SLC22A1 rs622342 A/A Metformin (increased effect)

VKORC rs9923231 C/T Warfarin (lower dose required)

High interest drug response–related variants: PharmaGKB

Drug Side Effects

•  One teenager had her genome sequenced; discovered genetic susceptibility for blood clots (factor V mutation) (3-8% of European have this mutations)

à affects her birth control decisions as hormonal therapies increase blood clots

Many Unaddressed Challenges1)  Accuracy and coverage

2)  Interpreting non protein coding regions

3) Other information

Sequencing Accuracy Sequencing the Same Genome Twice

Personalis

146,100 SNPs (3.7%)

All genesMendelian disease genes in ClinVarACMG reportable genes

Median percentage of gene covered by ten or more reads

Perc

enta

ge o

f gen

es0

2040

6080

100

<70 70−74 75−79 80−84 85−89 90−94 95−98 99−1000 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0

6 7 7

92 92 93

All genesMendelian disease genes in ClinVarACMG reportable genes

Minimum percentage of gene covered by ten or more reads

Perc

enta

ge o

f gen

es0

2040

6080

100

4 4 2 1 2 0 2 2 2 3 4 26 7

211 13

18

3136

4540

32 30

<70 70−74 75−79 80−84 85−89 90−94 95−98 99−100

Coverage of Mendelian disease genes is high but incomplete

All genesMendelian disease genes in ClinVarACMG reportable genes

Median percentage of gene covered by ten or more reads

Perc

enta

ge o

f gen

es0

2040

6080

100

<70 70−74 75−79 80−84 85−89 90−94 95−98 99−1000 0 0 0 0 0 0 0 0 0 0 0 1 1 0 2 2 2

11 12 14

85 85 84

All genesMendelian disease genes in ClinVarACMG reportable genes

Minimum percentage of gene covered by ten or more reads

Perc

enta

ge o

f gen

es0

2040

6080

100

1 0 0 0 0 0 0 0 0 1 1 0 2 2 05 5 2

18 1923

74 73 75

<70 70−74 75−79 80−84 85−89 90−94 95−98 99−100

Illumina Complete Genomics

Coverage of ACMG reportable genes is high but incomplete

ACTA2

ACTC1

APC

BRCA1

BRCA2

CACNA1S

COL3A1

DSP

GLA

LDLR

MLH1

MSH2

MSH6

MUTYH

MYBPC3

MYH11

MYH7

MYL2

MYL3

NF2

NTRK1

PCSK9

PMS2

RB1

SCN5A

SDHAF2

SDHB

SDHC

SDHD

TGFBR2

TMEM

43TNNT2

TP53

TSC1

TSC2

VHL

DSC2

APOB

FBN1

RYR2

LMNA

DSG2

SMAD3

MYLK

MEN1

PRKAG2

TGFBR1

RET

TPM1

RYR1

PTEN

PKP2

WT1

KCNQ1

KCNH2

STK11

02040608090

92

94

96

98

100

Gene

IlluminaComplete Genomics

% o

f exo

nic

base

s co

vere

d by

10

read

s

Mapping Regulatory Variation to Personal Genomes

Two approaches:

1)  Mapping transcription factor binding in different people.

2)  RegulomeDB: Assembling regulatory information from the ENCODE Project and other sources.

54

X

RegulomeDB

Alan Boyle

Data Type Types Features Genomic Coverage (bp)

Transcription Factor ChIP-Seq (ENCODE) 495 conditions / cell lines 7,721,822 230,795,743

Transcription Factor ChIP-Seq (non-ENCODE) 32 conditions / cell lines 397,534 140,534,725

Transcription Factor ChIP-exo 1 condition 35,161 2,604,066

Histone Modifications 284 conditions / cell lines/ marks 23,055,241 2,805,205,184

DNase I Hypersensitive Sites 114 conditions / cell lines 20,710,098 614,973,579

FAIRE Sites 25 conditions / cell lines 4,816,196 476,386,909

DNase I Footprints 50 cell lines 128,266,803 178,722,370

Predicted Binding (PWMs) 1,158 motifs 239,713,973 1,151,732,122

eQTLs 142,945 SNPs 142,945 142,945

dsQTLs 6,069 SNPs 6,069 6,069

Manual Annotations 6 Genomic Regions 282 11,607

VISTA Enhancers 1,448 Enhancers 1,325 1,658,146

Validated SNPs affecting binding 855 SNPs 855 855

Alan P Boyle RegulomeDB

Damaging Variation in an Individual

Protein Coding Non-coding

Damaging Variation in an Individual

Gene Regulatory region

Protein Coding Non-coding

Gene

and

Alan P Boyle RegulomeDB

CAPN1 Compound Heterozygote

CAPN1 Regulatory region

Maternal Chromosome

Paternal Chromosome and

Calcium-sensitive cysteine protease in brain synapses Protective against Alzheimer’s Disease (Trinchese et al. 2008)

What About Other Omes?

Health Is a Product of Genome & Exposome

Food Health Disease

Genome

Pathogens

Exercise

Stress

Epigenomics Definition: From Wikipedia

•  Epigenomics is the study of the complete set of epigenetic modifications on the genetic material of a cell, known as the epigenome. Epigenetic modifications are reversible modifications on a cell’s DNA or histones that affect gene expression without altering the DNA sequence (Russell 2010 p. 475).

•  Two of the most characterized epigenetic modifications are DNA methylation and histone modification.

Key Points

Many Opportunities for Modification Throughout Life

DNA Methylation A Sixth Base: 5Hydroxymethylation

http://www.ks.uiuc.edu/Research/methylation/

DNA Methylation in Promoters is Traditionally Associated with Gene Silencing

•  Transfection experiments – 1980s – Methylated promoterà gene off – Unmethylation promoterà Gene on

•  Correlates globally with gene expression (or lack thereof)

Whole genome profiling of DNA methylation

bisu

lfite

sequ

encin

g5m

C

5mCG 5mCG 5mCHG

Deep sequencing

Bisulfite converted library

Fragmented gDNA

Bisulfite conversion

MethylC-seq Bisulfite sequencing 2007...

Lister et al. et al. 2008

Illumina - SE or PE

Map reads against computationally BS converted + / - strands

Post-processing & stack

Call mC modifications

mC

Lister et al. 2009 In stem cells 62M mC (6% of Cs) Fibroblast 45M (4.3%) Stem cells 75% mCpG & 25% nonCpGs Fibroblasts: 99% are in CpGs Correlate with Gene Expression

Gene bodies are often methylated

DNA Methylation Associated with Many Diseases and Traits

Aging Nutrition Cancer Asthma and Allergies (Nadeau)