ASHG sequencing workshop

Dr. Mike Evans — Chief Executive

A unique targeted sequencing service providing meaningful results, not insurmountable data

Outline of presentation

• Delivering a unique next generation sequencing service —Dr Mike Evans, CEO

• Optimised bait design for targeted sequencing — Dr Volker Brenner, Head of Computational Biology

• Adding value through analysis — Dr Volker Brenner, Head of Computational Biology

• Summary• Q&A

OGT - provides advanced clinical genetics solutions - develops innovative molecular diagnostics

• Founded by Ed Southern in 1995• 64 people

OGT Begbroke: Corporate offices and high-throughput labs

OGT Southern Centre: Biomarker discovery

IP Licensing40 licence relationships

TechnologiesFor Molecular

Medicine

Clinical and Genomic SolutionsCytogenetics products and genomic services

Diagnostic BiomarkersGenomic- and protein-based diagnostics

OGT’s key businesses

Clinical and Genomic Solutions

Addressing the challenges of high-throughput, high-resolution molecular technologies:

• High equipment and staff training costs• Short equipment lifespan• Complex study design and processes (e.g. platform evaluation &

selection)• Vast amounts of data

• Extensive computing infrastructure • Data analysis expertise and resource

The solution: Genefficiency Genomic Services

High-quality data & complete reassurance

• Experimental and array design expertise• High-throughput processing (>2000 samples / week)• Applications: aCGH-CNV, methylation, miRNA, gene expression

analysis• Comprehensive data analysis services • >40 QC checks on each sample to ensure high-quality data

Genefficiency™ — World’s leading aCGH service

Independent accreditations

• First Agilent High-Throughput Microarray Certified Service Provider

• ISO 9001:2008 — Quality management systems

• ISO 27001:2005 — Information security

• ISO 17025:2005 — aCGH Laboratory services

F S 5 6 1 1 5 6

I S 5 6 1 1 5 7

4 5 9 3

20,000 samples. 1,000 samples / week

“In order to characterise genetic variants, reproducible performance and reliable processing of the high resolution microarrays is essential. We were pleased with OGT’s responsive approach and attention to producing high quality data to tight deadlines”

Dr Matt Hurles, Wellcome Trust Sanger Institute.”

Customer satisfaction…

OGT collaborators and customers

A world-class team

Our expert team deliver:• Excellent project management and customer service

• >600 projects to date• >50,000 samples

• Unparalleled expertise in study and probe design• Advanced data analysis though a dedicated team of

bioinformaticians• Rapid turnaround times• A wealth of experience of clinical and translational

research projects

New Genefficiency Targeted Sequencing Services

Delivering discovery

Genefficiency Targeted Sequencing Services — designed to be different:

• Comprehensive — taking you from genomic DNA to filtered, qualified results• Rigorously designed — project and probe design expertise maximises your

likelihood of discovery• Expert support — experienced team of biologists and bioinformaticians• Dedication to quality — from sample to result, delivering reliable results

every time

Delivering an integrated, comprehensive service

27/10/2011 13

1. Selection of most appropriate genomic regions for enrichment

2. Capture, sample multiplexing and sequencing

3. Data analysis and advanced filtering of variants

Delivering expert project design

Step 1: Selection of most appropriate genomic regions for your project and budget

Whole exomePre-designed, validated whole exome capture probes

Coding regions are “most likely” candidates for many disorders

Custom genomic regionsExpert custom design of capture probes for your regions of interest

Flexibility to focus on regions of clinical significance or GWAS regions

Delivering class-leading technology

We have fully optimised the DNA capture and sequencing methodologies, so you don’t have to!

Step 2: Performing the capture, sample multiplexing, library preparation and sequencing

• Options for sample indexing and multiplexing to minimise sequencing cost

• Depth of sequencing coverage to suit your samples and project

• Paired-end sequencing on the industry-leading Illumina HiSeq 2000

OGT delivers discovery, not just data

Step 3: Data analysis and advanced filtering of variants

• OGT’s dedicated analysis pipeline brings you beyond data, to a filtered list of variants relevant to your study

SEQUENCE FILTER DISCOVER

Genefficiency Targeted Sequencing Services

The PLATFORM• Core sequencing platform: Illumina HiSeq 2000 • Core sequence capture technology: Agilent SureSelect

The PEOPLE• Team of highly skilled molecular biologists and bioinformaticians• Core expertise in probe design • Successful development of advanced analysis solutions





• Summary• Q&A

Agenda

• Important Definitions and Terminologies

• Introduction to Targeted Enrichment

• Custom Bait Design

Definitions and terminologies

• Read length — The number of bases sequenced in a fragment

• Capture efficiency

• Paired end sequencing

• Read depth — How many times has a base been sequenced?

On target Off targetOff target

Region of Interest

Region of Interest

Fragment 1

Fragment 2

Assuming no allelic bias the theoretical read depth required to detect heterozygous variation with given accuracy can be calculated using a binomial distribution

• Minimum capacity required = Region of interest (ROI) x required depth• Q30 variant detection for 15Kb ROI requires 210Kb sequencing capacity

Calculations based on variation being seen in at least 2 reads• Should not be just one read as this could be ‘noise’• Required observations could be a percentage of reads

Read depth required for mutation detection

Depth Required Het. Call Accuracy Probability of Error Quality11 99% 1:100 Q2014 99.9% 1:1000 Q3018 99.99% 1:10000 Q4025 99.999% 1:100000 Q50

Agenda




Why use targeted enrichment?

Flexibility in choice of genomic loci• Allows capture of specific regions of interest for SNP and Indel detection

Cost Effectiveness• Ideal for clinical applications

• Specific candidate genes are targeted • Fine mapping post-GWAS

• Cost Benefits• Enables multiplexing to fill capacity

Streamlined Data Analysis• Reduced noise due to targeted specificity

14x (Q30)

Targeted gene sequencing can lead to some targets without therequired depth of coverage

Example of design bias — Insufficient coverage

Inadequate Coverage

*data kindly provided by C. Mattocks National Genetics Reference Lab, Salisbury, UK

Option 1:• Increase coverage by

increasing depth of sequencing

• Coverage of all targets proportionally increased

• Increased cost of sequencing

• Some bases still missed

(Q30)

Solution: Intelligent design to improve coverage:

Option 2:• Intelligent design of capture probes

increases under-represented loci• More even coverage of entire region,

no loci missed (more likely to find mutations present)

• No need to increase sequence depth overall (more cost effective)

Agenda




Problems facing users

• Design tools not user friendly• Design tools only good for draft design• Potential sources of bias

• Regions of interest too short• Bait thermodynamic behaviour

• GC content• Melting Temperature

• Risk of Design Errors

• OGT’s extensive experience in designing probes for microarrays allows us to minimise bias and ensure evenness of coverage giving the best chance to identify mutations

OGT’s design pipeline — what we need from you

• Regions of Interest• Gene lists• Chromosomal locations

• Genome build version

• Data file format• Text, Excel, etc....• Consistent e.g. chr1: 2247628-2248537

3. Singletons2. Draft Design1. Data 4. Thermo-

dynamics 5. Report

• Assess the output:• Coverage• Bait distribution• Repeat masking

Region of Interest

Run draft design

3. Singleton Baits

2. Draft Design1. Data 4. Bait Thermo-

dynamics 5. Report

Repeat masking

OGT custom bait design gives increased read depth around edges of target regions.

Custom baits improve coverage at region boundaries

1KGOGT

• This ensures that small regions are captured as well as large regions

• Advantage — Improves evenness of capture across the design

Before After

• Review the draft design and identify any regions covered by a single bait• These regions span less than 120 bases

• Add additional singleton baits to the design

Correction for singleton baits

3. Singleton Baits


dynamics 5. Report

Custom approach ensures variant detection

OGT

1KG

Even at more than 50x coverage, whole exome sequencing does not accurately identify all SNPs.OGT custom baits design compared with 1000 Genomes whole exome capture data.

GC content • Calculate GC content for all baits• Identify those baits where GC

content is extreme (for instance >65% and <40%)

• Add additional copies of these baits

Region of Interest

GC extreme

Correction for bait thermodynamicsTm content • Calculate the Tm for all baits• Identify those baits where Tm is

extreme (e.g. > 75oC)

• Add additional copies of these baits

Tm extreme

3. Singleton Baits


dynamics 5. Report

In a region with 70% GC content OGT custom bait design achieved a maximum read depth of 50x. The Agilent SureSelect 50Mb capture kit does not capture any reads in this region.

OGT

SureSelect

OGT custom bait designs help overcome GC issues

Relative capture of targets within a single gene. Agilent coverage is 20x for the target with no GC content bias, and minimal for targets with a GC content of 65%. In contrast OGT custom baits perform excellently in this region.

OGT

SureSelect

OGT custom bait designs help overcome GC issues

3. SingletonBaits


dynamics 5. Report

• Design Parameters

• Depth of Coverage• On target / Off target• Regions not covered – and why not

• Bait Details• Singletons• GC distribution• Tm distribution

• Library Design• Baits generated

Customer report

• Custom design of regions for targeted sequencing offers significant flexibility for many applications

• Expert probe design will ensure:

• Better ‘evenness’ of coverage helps ensure no regions are missed and maximises the likelihood of variant detection

• Improvement of overall capture efficiency and on-target performance equals cost effective sequencing downstream

• Increase capture efficiency of SNPs and Indels equals an increase in the likelihood of detection

• Reduction of risk and better performance

Summary

Adding value through analysis

• Introduction• NGS data analysis

• Primary analysis• Mapping and assembly• Q score re-calibration• NGS sequencing QC• NGS alignment QC

• Secondary analysis• SNP and Indel calling• Annotation and evaluation pipeline• SIFT and PolyPhen

• Deliverables• Case study• Summary

The analysis challenge

NGS Raw data Mapping Annotation Filtering Reporting

SequencerHard drive

with ~4Gb per exome

Publication

Mapping Annotation Filtering Reporting

Raw data: FASTQ(standard text representation of short reads)

FASTQ uses four lines per sequence.

• Line 1: '@' followed by a sequence identifier

• Line 2: raw sequence letters

• Line 3: '+' (and optional sequence identifier)

• Line 4: quality values for the sequence in Line 2. Must contain the same number of symbols as letters in the sequence. (The letters encode Phred Quality Scores from 0 to 93 using ASCII 33 to 126)

Example

@SEQ_IDGATTTGGGGTTCAAAGCAGTATCGATCAAATAGTAAATCCATTTGTTCAACTCACAGTTT+!''*((((***+))%%%++)(%%%%).1***-+*''))**55CCF>>>>>>CCCCCCC65

Phred quality scores

• Phred is an accurate base-caller used for capillary traces (Ewing et al Genome Research 1998)

• Each called base is given a quality score Q• Quality based on simple metrics (such as peak spacing) calibrated against a

database of hand-edited data• QPhred = -10 * log10(estimated probability call is wrong)

Q30 often used as a threshold for useful sequence data

Phred Quality Score Probability of incorrect base call Base call accuracy

10 1 in 10 90 %

20 1 in 100 99 %

30 1 in 1000 99.9 %

40 1 in 10000 99.99 %

Adding value through analysis

• Introduction• NGS data analysis

• Primary analysis• Mapping and assembly• Q score re-calibration• NGS sequencing QC• NGS alignment QC

• Secondary analysis• SNP and Indel calling• Annotation and evaluation pipeline• SIFT and PolyPhen

• Deliverables• Case study• Summary

Primary analysis — Mapping and alignment

Raw Sequence

Files

FASTQ Format

Mapping

BWA/Bowtie

Raw Alignment

Files

SAM/BAM Format

Local Realignment(around InDels)

GATK

Duplicate marking

Analysis-ready

Alignment

Picard SAM/BAM Format

Quality score re-

calibration

Picard

Why mark duplicates and realignment around indels?

3 incorrect calls within 40bp!

Primary analysis — Mapping and alignment

Raw Sequence

Files

FASTQ Format

Mapping

BWA/Bowtie

Raw Alignment

Files

SAM/BAM Format


GATK

Duplicate marking

Analysis-ready

Alignment


Quality score re-

calibration

Picard

NGS variant calling methods

Option 1 - Hard filteringExample: SNP can only be called if

• read depth >10 • >35% of reads carry SNP

Effective filtering Transparent to user– Simplistic approach– Will miss high quality calls that don’t pass threshold

Option 2 - Statistical analysisBased on quality scores of individual basepairs, the alignment and statistical probability models

Robust Optimum balance of sensitivity and specificity due to the use of statistical models Fewer false positive and false negative SNP calls– Requires correctly pre-processed data with reliable quality scores

Base quality score re-calibration

Source: The Broad Institutehttp://www.broadinstitute.org/files/shared/mpg/nextgen2010/nextgen_poplin.pdf

Before Recalibration After Recalibration

Primary analysis — Raw data and assembly QC

Raw Sequence

Files

FASTQ Format

Mapping

BWA/Bowtie

Raw Alignment

Files

SAM/BAM Format


GATK

Duplicate marking

Analysis-ready

Alignment


Quality score re-

calibration

Picard

Sequence QC check

Raw data QC Report

FastQC AlignmentQC Report

Alignment QC check

Picard

Secondary analysis SNP and Indel calling, annotation and filtering

GATK

Unified Genotyper

Analysis-ready

alignment

SNPs

InDels

VCF Format

Variant Evaluation

• Known variant?

• Impact on gene expression?

• Splicing affected?

• Non-synonymous or frameshiftmutation?

• Impact on protein function?

• How confident are we in the call?

• Zygosity?

Comprehensiveinteractive OGT

Report

AlignmentQC Report

Sequence QC Report

SAM/BAM Format OGT

SNP/Indel classification(standard analysis)

We check and annotate every single detected SNP and Indel against all human Ensembl genes and transcripts and dbSNP

dbSNP annotation:• Is the variant known?• Obtain allele frequency

Does it affect any of the following• Promoter region• UTR• Splice sites or intronic region• CDS

• Synonymous mutation• Non synonymous mutation• Frameshift mutation• Stop codon (truncated/elongated protein sequence)• Overlap with protein domain• Consequence on protein function predicted (SIFT & PolyPhen)

OGT Processing Overview

Gather All detected SNP/Indels

Not Described in dbSNP

Mapped to Promoter Regions

Perform pairwisegenome analysis

Filter out variants present in “baseline” genome (e.g. somatic tissue, healthy sibling)

Additional Filtering and Analysis

Mapped to Exons, Splice sites or UTRs

and Protein domains

Non-synonymous Coding Variations




Variations with Serious Consequences to the

Protein Sequence (SIFT)




Described in dbSNP Rare RS ID Variations




Individual Genome Analysis

(Standard Level)

Multi Genome Analysis, Data Gathering and Comparison

(Advanced Level)

Tailored analysis based on client’s

individual requirements

(Expert Level)


Filter out variants

present in any “baseline”

exome (e.g. somatic tissue, healthy sibling)

AND not all “case” exomes

Study specific additional in-depth filtering and analysis

DataInformation

NGS data delivery

Hard drive(or FTP)

ship data

Double click!

Comprehensive HTML analysis report

File location& share results

Analysis report: Summary section

Analysis report: QC section — Read QC

Analysis report: QC section — Read QC

Analysis report: QC section — Alignment QC

Analysis report: QC section — Alignment QC

Analysis section — Overview

The Variant Table View

Data display

Data export

The Variant Table View — External links

The Detailed Variant View

Predicted consequences on protein function

Alignment View of selected variant in IGV

OGT data processing ensures detection of insertions

Detection of an 31bp insertion

Detection of an 84bp deletion

OGT data processing ensures detection of deletions: Example1

Detection of homozygous and heterozygous deletions

Heterozygous deletion

Homozygous deletion

No deletion (reference sequence)

Interactive data filtering

Customer data: Analysis of consanguineous samples

1

1

2

2II

I

Data courtesy of Dr. Bernd Wollnik, Institute of Human Genetics, University Hospital of Cologne

HACE1Exon11c.994C>TR332X(CGA -> TGA)

ANK1 ANK1 HECT69-161 168-258 602-909

R332XControl

Mother

Father

Patient1

Patient2

H V F R I G PX

Data courtesy of Dr. Bernd Wollnik, Institute of Human Genetics, University Hospital of Cologne

Confirmation by Sanger sequencing

Customer feedback...

Analysis of Consanguineous Samples

“Just wanted to let you know that we have probably identified the

causative gene and mutation in the patient sample.

The mutation is located in the middle of an 18 Mb homozygous

stretch and is a homozygous nonsense mutation!!!

Wow, its going so nicely with your data!!!”

Dr. Bernd Wollnik, Institute of Human Genetics, University Hospital of Cologne

SummaryOGT offers fast, accurate & powerful NGS analysis

Standard Analysis

• Robust statistical data analysis

• Comprehensive variant annotation

• Interactive filtering and prioritisation of data based on• chromosomal region

• allele frequency / novelty

• zygosity

• confidence score and read depth

• severity of mutation

Advanced Analysis• Multi-genome comparison

Bespoke analysis • Tailored to your specific requirements





• Summary• Q&A

Speak to one of our team or visit booth 713 to:

• Book a demonstration of our interactive analysis report — Hurry limited availability

• Discuss your specific project requirements

• Take part in our short survey and have your chance to win an Amazon Kindle

75

Thank youwww.ogt.co.uk

ASHG sequencing workshop

Education

high quality data

sequencing options

end sequencing

analysis dr volker brenner

insurmountable data

probe design expertise

genefficiency genomic

high equipment