Next Frontiers in Diagnostic Sequencing

Emily Winn-Deen, Ph.D. November 8, 2012

San Diego Section

Genomes

10k

1M

100k

100

1,000

The Tipping Point for Human Genome Sequencing Costs has Passed

Cos

t per

Hum

an G

enom

e

Venter

Watson

Costs

$1M

$10k

$100M

$10M

$100k

2012

Year

2008 2007 2009 2010 2011

African, Asian, Cancer pair

169 in Genbank

Individual Genome Sequencing

~$3500 today

Clinical Sequencing Costs Today

•  HIV drug resistance mutations –  CPT code 87901 - $359.69

•  BRCA1and BRCA2 (sequencing + interpretation) –  Myriad Genetics - Medicare $3340, Medicaid $3100

•  Whole Exome (including interpretation) –  Ambry Genetics - $7900 per sample (June 2012) –  Gene Dx - $5000 for proband, $9500 for trio (August 2012) –  Baylor College of Medicine - $9000 per sample (November

2011) •  Whole Genome (sequencing only)

–  Illumina - RapidTrack WGS, $9,500 per sample (June 2012)

SEQUENCING TECHNOLOGY

Disruptive Sequencing Technology

Genera&ons  are  defined  by  the  customers  

Main  Frame     Mini  Computer     Personal  Computer  

Sanger  Sequencing   Next-­‐Gen  Sequencing   Semiconductor  Sequencing  

Massively parallel sequencing

Sequencing Method Comparison

variant

gene

sequence

PCR amplicon

A G

A. Bi-directional Sanger Approach B. Illumina SBS Approach

A G

>30x

AAAACCAGAGTCTAGCACCTTCTCATCAGGAGCAGAAACCAGAGTCTAGCACCTTCTCATCAGGAGCAAC AACCAGAGTCTAGCACCTTCTCATCAGGAGCAACGACCAGAGTCTAGCACCTTCTCATCAGCAGCAACGTACCAGAGTCTAGCACCTTCTCATCAGGAGCAGCGTCCAGAGTCTAGCACCTTCTCATCAGGAGCAACGTC

GAGTCTAGCACCTTCTCATCAGGAGCAACGTCTGCCTAGCACCTTCTCATCAGGAGCAGCGTCTGCCTTCTAGCACCTTCTCATCAGAAGCAACGTCTGCCTTCGAGCACCTTCTCATCAGGAGCAACGTCTGCCTTCGC

CCCTTCTCATCAGGAGCAGCGTCTGCCTTCGCTAGACCTTCTCATCAGTAGCAACGTCTGCCTTCGCTAGCTTCTCATCAGGAGCAACGTCTGCCTTCGCTAGGC

ATCAGGAGCAGCGTCTGCCTTCGCTAGGCTGACATATCAGGAGCAACGTCTGCCTTCGCTAGGCTGACATTCAGGAGCAGCGTCTGCCTTCGCTAGGCTGACATC

GAGCAACGTCTGCCTTCGCTAGGCTGACATCGCGGGAGCAACGTCTGCCTTCGCTAGGCTGACATCGCGG

AACGTCTGCCTTCGCTAGGCTGACATCGCGGGACCAACGTCTGCCTTCGCTAGGCTGACATCGCGGGACC

AAAACCAGAGTCTAGCACCTTCTCATCAGGAGCAGCGTCTGCCTTCGCTAGGCTGACATCGCGGGACC

Sanger vs. SBS Sequencing Characteristic Sanger SBS

Through-put One gene segment is sequenced on one direction in each capillary

Many gene segments are simultaneously sequenced in both directions using paired end reads

Signal Purity Signal is averaged across a population of molecules

Signal is generated for each molecule separately

Heterozygote Detection

Decrease in base signal intensity and appearance of a second base signal.

Accumulation of sequence reads from each allele

Limit of Minor Allele Detection

15-20% 1-5%, minor allele detection limit is dependent on the number of clusters sequenced (depth of coverage)

Quality Scoring Qualitative measure of peak height relative to surrounding sequence

Quantitative measure per base relative to signal/noise, sequence alignment and number of sequences reads

Illumina Sequencing by Synthesis

Image acquisition

1 2 3 7 8 9 4 5 6

Base calling

T G C T A C G A T …

DNA (<1 ug)

Single molecule array Sample

preparation Cluster growth (0.1 – 0.5 billion)

5’

5’ 3’

G

T

C

A

G

T

C

A

G

T

C

A

C

A

G

T C

A

T

C

A

C

C

T A G

C G

T A

G T

Sequencing (2 x 35-100 bases)

Illumina MiSeq

SBS reversible terminator chemistry Data compatibility with other Illumina platforms Data quality with fast cycle times Integrated paired end read capability Sequences through homopolymers

Workflow On-board cluster generation and data analysis Fully integrated system and preloaded reagents RFID tagging

Performance Throughput up to 1-1.5 Gb Scalable run time vs output: 36bp to 2x150 bp

Footprint 52.3 cm high x 2 feet wide x 2 feet deep

MiSeq Workflow

PCR or Selective Pull-

out 1.5 hours

(15-30 min hands on)

Cluster Generation and

Paired-End Sequencing 22.6 hours* (20 minutes hands on)

Alignment and Variant Calling

2 hours (fully automated)

*2 x 100 bp run

DNA Isolation (From Blood or

FFPE) 1.5 hours

(15-30 min hands on)

Illumina MiSeq Path to CE-IVD and 510(k)

MiSeq Dx Menu

–  Cystic fibrosis screening assay •  Targeted sequencing of 162 variants in the CFTR gene

–  Cystic fibrosis diagnosis assay •  Targeted sequencing of all medically relevant areas of the CFTR gene

– HIV sequencing assay •  Partnership with Siemens announced in November 2011

Core Sequencer Completion RUO sequencer launched in October 2011

IVD Assay Development and Regulatory Filings

product lock down

2011 2012

Third-Gen Sequencing - The Chip is the Machine

Scalability Simplicity Speed

Ion Torrent Technology Scalable Semiconductor Technology plus Simple Chemistry

Wafer Semiconductor Manufacturing

Chip Semiconductor Packaging

Millions of Sensors Semiconductor Design

Sensor Plate

Silicon Substrate Drain Source Bulk

∆ pH

∆ V

Sensing Layer

H+

Single Sensor Chemical to Digital Sequence

TCGTACC…

ION Torrent - Fast Direct Detection

–  Natural nucleotides flow sequentially over the semiconductor chip –  Direct detection of H+ ions released during natural DNA extension –  Only a few seconds per incorporation

Sensor Plate

Silicon Substrate Drain Source Bulk

dNTP

To column receiver

∆ pH

∆ Q

∆ V

Sensing Layer

H+

Rothberg  J.M.    et  al  Nature  doi:10.1038/nature10242  

Natural Nucleotide Chemistry

Ion Torrent Assay Workflow

Ion PGM™ Path to CE-IVD and 510(k)

Ion Dx technology

–  Accessible to all clinical labs

–  2-4 hour sequencing runs

–  Low cost sequencer and kits

–  Millions of 200+ base pair long reads

–  Field proven robustness (1000s of systems placed)

Core Product Completion Low cost, starting at < $299 per run

Regulatory filings Builds on Life’s IVD experience: 7500 Fast Dx & 3500Dx

product lock down

2011 2012

CLINICAL SEQUENCING APPLICATIONS

Human Microbiome •  Many publications this year from the Human

Microbiome Project –  242 healthy young adults (20s) from the Houston and

St. Louis areas –  18 body sites –  Sequenced using Roche 454 platform –  Human sequences were filtered out with bioinformatics

•  Normal gut microbiome –  700-900 taxa found at >0.1% –  100 taxa dominate –  >99% bacterial

Gut Microbiome •  Many conditions show abnormal gut microbiomes

–  Celiac disease –  C. difficile colitis –  Inflammatory bowel syndrome –  Crohn’s disease –  Short bowel syndrome –  Type 2 diabetes

•  Is this a cause or an effect? •  Could these diseases be treated by restoring a normal

microbiome population? •  Will 16S rRNA sequencing replace GI panels for diagnosis

of GI infection?

Pediatric Cancer Genome Project

•  Joint program at St. Jude’s and Washington University –  Announced February 2012 –  St. Jude’s provides cases and pediatric cancer expertise –  Washington University provides genomic sequencing and

bioinformatics expertise

•  Privately funded with $65 million •  Project will sequence 600 tumor/normal genome pairs

–  >300 cases completed to date –  Sequencing various cancer types to 30x coverage –  Will perform WGS, RNAseq, and methylation analysis

BASIC3 Project •  MD Anderson program

–  Focused on pediatric oncology –  Started in August 2012

•  Project will exome sequence 250-300 tumor/normal pairs –  All sequenced in a CLIA lab environment –  Sequencing all solid tumors, including brain tumors –  Also request blood samples from parents to determine inheritance –  Will measure impact on treatment decisions

•  Need to decide on what germline mutation information from the affected child will be communicated to parents –  Cancer susceptibility –  Pharmacogenetic markers –  Other findings

Targeted Next-Gen Sequencing •  Multiple methods for target

sequence enrichment. " Amplification (multiplexed PCR,

multiplexed target amplification with universal primers, emulsion PCR)

" Targeted pull-down (hybridization on arrays, capture in solution)

" Advantages " Increased sequence coverage is

better for heterogeneous samples

" Simpler sample processing

" Disadvantages " Not useful for discovery

" Ignores potentially critical genomic information contained in non-targeted regions

Mamanova et al., Nature Methods 7:111-118 Feb 2010

Oncology Sequencing Panels

•  University of Pennsylvania – Hematologic cancer - 35 genes – Solid tumors – 48 genes

•  MD Anderson – Lymphoid tumors – 13 genes

•  Baylor College of Medicine – Leukemia - 48 genes

•  Washington University, St. Louis – Solid and hematologic cancer – 27 genes

Ion AmpliSeq™ Cancer RUO Panel 190 Amplicons, One Tube, One Day

Single tube amplification for hundreds of amplicons 3.5 hour library construction Starting with 10ng of FFPE DNA Comprehensive coverage of 739 known mutations

Compatible with Ion OneTouch or Ion Xpress Template Kits 4 hour amplification

1.5 hour sequencing run Compatible with all Ion Chips

Single day workflow from DNA to Variants- 1 hour analysis Ion AmpliSeq Variant Caller Software Detection of allele frequencies at 5% with 99% confidence

For Research Use Only. Not intended for any animal or human therapeutic or diagnostic use.

RUO Cancer Panel Comparison AmpliSeq Panel v.1 (Life Technologies)

TruSeq Panel (Illumina)

Number of tubes 1 1 Number of genes 46 48 Number of amplicons 190 212

Amplicon length (bp) ~119 170-190 Panel size (kb) 12 35 Input DNA (ng) 10 250 Sample multiplexing 16 96 Sequencing Unidirectional Paired ends (2x150)

Time to results 2 days 3 days

Reproducibility Better Good

RUO Cancer Panel Comparison

•  Assay Limitations: –  Both panels cover mutation hot-spots and not the

entire coding sequence of the genes in the panel •  Illumina Advantages:

–  Includes 2 more genes that ION (GNA11, GNAQ) –  Can multiplex up to 96 samples in a run (vs 8 for ION)

•  Ion Torrent Advantages –  Only needs 10 ng DNA (vs. 250 ng for ILMN) –  End-to-end workflow in 1 day (vs 48 hours for ILMN)

Genetics Sequencing Panels

•  Emory University – Epilepsy panel – 173 genes – X-linked intellectual disability panel – 107

genes – Congenital disorders of glycosylation panel –

38 genes – Congenital muscular dystrophy panel – 13

genes – Autism panel – 52 genes – Short stature panel – 11 genes

Genetics Sequencing Panels

•  University of Minnesota – Cardiac panel – 4 genes/4 diseases – Metabolism panel – 14 genes/10 diseases – Hearing loss panel – 80 genes/12 diseases – Eye disease panel - 104 genes/13 diseases – Familial cancer panel – 34 genes/15 diseases – Neurological disease panel – 223 genes/31

diseases

Principals of Fetal Aneuploidy Testing ~13% of DNA fragments in a pregnant woman’s blood are from the fetus ( ) and ~87% are from the mother ( ).

Free Fetal DNA in Maternal Circulation The number of fetal fragments vs. maternal fragments for each chromosome is proportional to the size of the chromosome, which is relatively consistent from sample to sample, and patient to patient.

Chromosome 1 Chromosome 21

Sequencing of Free Fetal DNA Sequencing is used to assign each fragment to its corresponding chromosome

Commercial Non-Invasive Prenatal Testing

•  Prenatal testing for trisomy 13, 18, and 21 based on massively parallel sequencing of free fetal DNA in maternal circulation – Sequenom – MaterniT21 Plus

•  CLIA lab (San Diego, CA)

– Verinata – Verifi test •  CLIA lab (Redwood City, CA)

– Ariosa – Harmony test •  Available through CLIA labs at Integrated Genetics

and LabCorp

Clinical Whole Genome/Exome Sequencing •  Indications for Use:

–  Diagnosis refractory cases •  Particularly helpful in pediatrics

–  Cancer/tumor characterization •  Particularly when tumor does not respond to standard of care

•  Issues: –  Counseling is critical

•  Takes 6-10 hours per person

–  Interpretation can be time consuming •  Still debating how much information should be reported

–  Yes: medically actionable results –  No: variants of unknown significance (VUS) –  It depends: Late onset/untreatable disorders

Clinical Whole Genome/Exome Sequencing

•  Advantages of WES – Lower cost – Can afford to use deeper coverage – Easier to interpret

•  Advantages of WGS – Hypothesis-free approach – No bias from selection method – Recent work from the ENCODE project

indicates intragenic regions have significance

$1,000 genome, but $100,000 analysis? “Having recently attended the Personal Genomes meeting at Cold Spring Harbor Laboratories (I was an organizer this year), I was struck by the number of talks that described the use of whole-genome sequencing and analysis to reveal the genetic basis of disease in patients. These patients included a child with irritable bowel disease, a child with severe combined immunodeficiency, two siblings affected with Miller syndrome, and several with cancers of different types. Although each presenter emphasized the rapidity with which these data can now be generated using next-generation sequencing instruments, they also listed the large number of people involved in the analysis of these datasets. The required expertise to 'solve' each case included molecular and computational biologists, geneticists, pathologists and physicians with exquisite knowledge of the disease and of treatment modalities, research nurses, genetic counselors, and IT and systems support specialists, among others. While much of the attendant effort was focused on the absolute importance of obtaining the correct diagnosis, the large number of specialists was critical for the completion of the data analysis, the annotation of variants, the interpretive 'filtering' necessary to deduce the causative or 'actionable' variants, the clinical verification of these variants, and the communication of results and their ramifications to the treating physician, and ultimately to the patient. “

Elaine Mardis, Washington University, St. Louis Genome Med. 2010; 2(11): 84

Interpretation of Sequencing Data

•  Sequence variation is previously reported and is a recognized cause of the disorder. –  Reported when there exists credible documentation of

a sequence variation and a clinical outcome. •  Sequence variation is previously unreported and

is of the type which is expected to cause the disorder. –  Insertions, deletions, frame shifts, and changes at the

invariant splice site AG/GT nucleotides can disrupt normal protein synthesis or regulation of cellular processes such as transcription and translation.

American College of Medical Genetics Guideline, 2008

Interpretation of Sequencing Data

•  Sequence variation is previously unreported and is of the type which may or may not be causative of the disorder. –  Missense changes, in-frame deletions or insertions, or

variants in a splice consensus sequence or that may produce cryptic splice sites, affect regulatory processes, interrupt exonic splicing enhancers and suppressor sites, or participate in other mechanisms that may be associated with disease. Further analysis may be warranted to clarify the clinical significance of such changes.

American College of Medical Genetics Guideline, 2008

Interpretation of Sequencing Data •  Sequence variation is previously unreported and

is probably not causative of disease. –  These variations represent base changes that do not

change the coding sequence or effect known processing or regulatory pathways, or are found in addition to a known pathologic change (in autosomal dominant disorders).

•  Sequence variation is previously reported and is a recognized neutral variant. –  Reported when evidence is available that the

sequence variation has been consistently observed in a normal population without association to disease or predisposition.

American College of Medical Genetics Guideline, 2008

Interpretation of Sequencing Data •  Sequence variation is not known or expected to

be causative of disease, but is found to be associated with a clinical presentation. –  Variants in this category have been suggested to

contribute to disease as low-penetrance mutations, which alone or in combination may or may not predispose an individual to disease or modify the severity of a clinical presentation in complex disorders. Variants in this category should be reported with appropriate caveats that the variants are not definitive mutations and medical management decisions should not be made on the presence of the variants alone.

American College of Medical Genetics Guideline, 2008

Clinical Whole Genome/Exome Sequencing •  Interpretation Challenges

–  There is no “reference” genome to map individuals against

•  All healthy people have benign single nucleotide and larger copy number variants

–  Using current annotation tools, a single whole genome can take up to 200 person-hours to interpret

–  Curated variant databases are still under development •  VariMed database will be based on results from the 1000

Genomes Project (healthy variants) •  Regulome database will be based on known transcription factor

binding sites and other regulatory elements •  Disease related variant databases vary greatly in quality and

this information is not consolidated in a single location

Human Genome Sequencing in 2012

•  Whole human genome sequencing time and cost has dropped significantly

•  Whole exome sequencing is replacing targeted gene sequencing panels for multigenic diagnostic applications

•  Bioinformatic analysis and clinical grade data interpretation costs now exceed sequencing costs

Who Should be Sequenced?

AMP Poll – October 2012 •  Would you want your genome sequenced?

– Audience response •  Yes: 76% •  No: 24%

Predicting the Future – 2014 to 2016

•  Full human genome sequencing cost drops to ~$1000 due to a combination of better sequencing technologies and automated bioinformatic analysis pipelines –  Neonatal screening by sequencing begins to replace current

methods –  Individual whole genome sequencing replaces many genotyping

applications

•  Human microbiome sequencing replaces panel testing for gastrointestinal diseases

•  RNA sequencing replaces arrays and real time PCR panel testing for expression panel analysis