Next Generation Sequencing in Genetic Diagnostics Alan Pittman, PhD

Transcription

1 Next Generation Sequencing 1 Institute of Neurology Faculty of Brain Sciences University College London Outline of lecture Sanger sequencing Next generation sequencing technologies Target enrichment Analysis of NGS data Applications of NGS technology in genetics diagnostics 2 Sanger sequencing 3 1

2 Sanger sequencing (2) Invented by Fred Sanger in 1977 Cycle sequencing One reaction + one sequence Sequence 800 base pairs per reaction Accurate (99.999%) but very slow 4 Human genome project Human genome project ( ) 3 billion base pairs long All done by Sanger sequencing Unravelled human genome sequence to drive genetics research We can now achieve this amount of sequencing in as little time as one day 5 Next generation sequencing technologies 6 2

3 The next generation of DNA sequencing (NGS) Move from Sanger DNA sequencing to a more high-throughput approach Sequence DNA molecules in parallel rather than sequence one molecule at a time Driven by new technological advances 7 The next generation of DNA sequencing (NGS) (2) Technological advances leading to a decrease in the cost of DNA sequencing Since the end of 2007, the cost has dropped at a rate faster than that of Moor s law 8 The next generation of DNA sequencing (NGS) (3) Development of new NGS methods began 10 years ago with 454 Pyrosequencing Solexa/Illumina was developed in 2005 DNA sequencing throughput jumped 10 orders of magnitude A decades perspective on DNA sequencing technology; Elaine R. Mardis. Nature 470, (10 February 2011) 9 3

4 Illumina (Solexa) sequencing Illumina is the most common sequencer (>80%) so will use this technology for the examples in the remainder of the talk Understand the sequencing-by-synthesis (SBS) principle 10 Illumina (Solexa) sequencing (2) Step 1: DNA library construction In the lab prepare the DNA sample for sequencing Sheared Sample DNA fragments End - Repair A-tailing Adapter ligation 11 DNA library Illumina (Solexa) sequencing (3) Step 2: cluster generation Hybridise DNA library to flowcell Perform bridge PCR to generate clusters Now ready to preform sequencing Illumina SBS technology Reversible terminator chemistry foundation DNA 12 Illumina flowcell Sample preparation Cluster growth Sequencing 4

5 13 Illumina (Solexa) sequencing (4) Step 3: sequencing-by-synthesis Sequence nucleotides 1 cycle at a time 4 bases = 4 different dyes Modified nucleotides mean only one base can be incorporated at a time Camera images the coloured bases on the surface of the flowcell every cycle Each image is converted to a nucleotide base call Mitchel and Metzker, 2010 Incorporate all four nucleotides, each label with a different dye Wash, fourcolour imaging Cleave dye and terminating groups, wash Illumina/Solexa reversible terminators Repeat cycles Top: Bottom: CATCGT CCCCCC Illumina sequencers 14 For genome scale sequencing Illumina HiSeq Generates approximately 1Tb of data per machine run 1,600,000,000 short DNA reads For smaller scale sequencing Illumina MiSeq Generates approximately 5Gb per run Illumina HiSeq Illumina MiSeq Target enrichment 15 5

6 16 Why target enrichment? Target enrichment (2) The haploid human genome is 3 billion base pairs long There are ~20,000 genes in the human genome Often we are only interested in the gene protein coding exons or exome The coding portion of the genome represents 1-2% of the genome More efficient to only sequence the bits we are interested in, rather than the entire genome! Costs 1,000-5,000 for a genome, but only 500-1,000 for an exome 17 Capture target regions of interest with baits Potential to capture several Mb of genomic regions (typically Mb) Commonly used methods: Illumina TruSeq Agilent SureSelect NimbleGen EZ Target enrichment (3) Unbound fraction discarded Genomic sample (set of chromosomes) Hybridization NGS Kit Genomic sample (Prepped) Wash beads and digest RNA SureSelect HYB buffer Streptavidin coated magnetic beads Amplify SureSelect biotinylated RNA library baits Bead capture Illumina TruSeq Custom Amplicon Multiplex PCR Target enrichment (4) Up to 384 simultaneous PCR reactions in one tube For sequencing 1-30 genes at a time Useful for small, targeted applications such as small gene sequencing projects and genetic diagnostics Prepare DNA Samples in the lab Use DesignStudio to create custom oligo capture probes flanking each region of interest Custom Probe 1 Region of interest Custom Probe 2 CAT (custom amplicon tube) CAT probes hybridize to flanking regions of interest in unfragmented gdna Custom Probe 1 Custom Probe 2 Extension/Ligation between Custom Probes across regions of interest PCR adds indices and sequencing primers P7 Index 1 Index 2 P Uniquely tagged amplicon library ready for cluster generation and sequencing P7 Index 1 Index 2 P5 6

7 Analysis of NGS data 19 NGS data analysis How is NGS data analysed? NGS data analysis represents the real challenge There are 4 main steps: 1. Primary analysis (process the machine raw data) 2. Secondary analysis (align the short read sequences to the genome) 3. Tertiary analysis (call variants in the sequence data) 4. Quaternary analysis (interpret the genetic data) 20 Process raw sequence data (primary analysis) Standard fastq file Four lines per read: 1. Sequence ID 2. Nucleotide sequence 3. Strand 4. Per base quality score Machine DNA base calls 21 Generate sequence files 7

8 Alignment to the genome (secondary analysis) Align short sequence reads (the fastq files) to the reference genome Specialist alignment programs; BWA, Novoalign Generates a.bam alignment file Short-read sequences (FastQ format) Generate overlapping sequence contig Align to genome reference 22 GenomeBrowse/ Variant detection (tertiary analysis) Look for differences in the aligned data compared to that of the reference genome Specialist programs: SAMTOOLS, GATK DNA Variant End result of variant detection is a table of variants found in the sequenced DNA sample Standard variant call (.vcf) format Making sense of the variant data (quaternary analysis) We need to interpret the variant data Typically there are 3 million variants detected in any given human genome We detect 23,000 variants in an exome What does all this genetic variation mean? GENE information Variant frequency information Publically available databases 24 Variant data Annotated variant (.vcf file) data for interpretation Annovar annotation software 8

9 Applications of NGS technology Neurogenetics Many adult onset neurodegenerative conditions, such as Alzheimer's disease, are currently untreatable Many neurodegenerative disease have a genetic cause Symptoms of some can sometimes be alleviated Research - new gene discovery/refinement gives new insights into the molecular pathogenesis Diagnostics - diagnosing those at risk may allow for early therapeutic intervention Need to identify carriers/individuals at risk Neurogenetics (2) Research - new gene discovery/refinement gives new insights into the molecular pathogenesis Families with an unknown genetic basis Whole exome sequencing Identify shared variants between distantly related relatives Identify the causal mutation 27 Johnson et al. Nature Neuroscience

10 Neurogenetic diagnostics Can we use new NGS approaches to genetic diagnostics of neurological disease? Diagnostics - diagnosing those at risk may allow for early therapeutic intervention Need to identify carriers/individuals at risk I will now present two examples of the use of NGS technology in a genetic diagnostic setting 28 Applications of NGS technology in genetic diagnostics I Design and implementation of NGS diagnostic panel for dementia using the Illumina MiSeq 29 Dementia 1 in 6 people will go on to get some form of dementia by the age of 80 By 2021 there will be 1 million people in the UK living with dementia Worldwide, there are currently 36 million people living with dementia The estimated cost of dementia worldwide is 380 billion Alzheimer s disease = 62% - most common Vascular dementia = 17% Mixed dementia = 10% Dementia with Lewy bodies = 4% Frontotemporal dementia = 2% Parkinson s dementia = 2% 30 Other = 3% 10

11 31 Genetic causes of dementia Alzheimer's disease - 1-5% of cases PSEN1, PSEN2, APP Frontotemporal Dementia - >10% of cases MAPT, GRN, VCP, TARDBP and CHMP2b, c9orf72 CJD - 15% of cases PRNP Familial British dementia - 100% of cases ITM2B Other: Rare genetic diseases that present clinically as dementia Design of a NGS dementia diagnostic panel Illumina TruSeq custom amplicon assay + MiSeq platform 15 genes Exons only 50Kb in size (<0.001% of genome) 99% was designable 425 base pair fragments 2x250 reads v2.0 chemistry Careful design of PCR probes APOE APP CHMP2B CSF1R FUS GRN ITM2B MAPT PRNP PSEN1 PSEN2 TARDBP TREM2 TYROBP VCP P7 Use DesignStudio to create custom oligo capture probes flanking each region of interest Custom Probe 1 Region of interest Custom Probe 2 CAT (custom amplicon tube) CAT probes hybridize to flanking regions of interest in unfragmented gdna Custom Custom Probe 1 Probe 2 Extension/Ligation between Custom Probes across regions of interest PCR adds indices Index 1 and sequencing primers Index 2 P Uniquely tagged amplicon library ready for cluster generation and sequencing P7 Index 1 Index 2 P5 Design of a NGS dementia diagnostic panel (2) Sequence gene exons only and introns of known splice site mutations e.g. PSEN1 gene Targeted amplicons

12 Design of a NGS dementia diagnostic panel (3) VCP - Some partial areas in exons NOT designable 34 Just have to keep it in mind and get on with it Traditional Sanger sequencing to fill in the gaps Not ideal really as we want to do it all with next generation sequencing Dementia diagnostic panel: blind study Determine feasibility of technology in clinical setting Validation of the technology 85 positive control dementia samples and 10 unaffected controls Call mutation (or not) in each sample MUST prove technology works??? 35 Which DNA sample carries which disease causing mutation??????? Dementia diagnostic panel: are the genes covered? >95% covered at least once >90% between x coverage If less than 20x coverage, amplicon judged to be a fail Last exon of PRNP Not covered in the middle

13 Dementia diagnostic panel: blind study results Are the blind mutations identified??? For point mutations: Sensitivity = 98% Specificity = 100% APP duplication missed Partial deletion of GRN missed PSEN1 Delta Dementia diagnostic panel: summary A targeted sequencing panel on the Illumina MiSeq has been developed for dementia Sequences all currently known genes that can cause dementia 98% effective Not 100%... Research tool to study the genetics of dementia Can diagnose the genetic basis of dementia All findings still need to be Sanger confirmed 38 Applications of NGS technology in genetic diagnostics II Design and implementation of NGS diagnostic panel for Parkinson's disease using the Illumina MiSeq 39 13

14 Parkinson s disease Progressive and degenerative movement disorder of CNS Characterized by tremor, slowness of movements and rigidity 1% people will go on to get Parkinson's disease 10 Million people worldwide Cost of $25 billion in the US alone William Richard Gowers, A Manual of Diseases of the Nervous System (1886) 40 Parkinson s disease genetics 15% of people with Parkinson s disease have a positive family history Parkinson s disease previously considered a non-inherited condition Both dominant and recessive 10 years research = 10 genes discovered Genes in other overlapping syndromes 41 Sun Ju Chung, MD. Genetics in Parkinson s Disease. J Korean Med Assoc Jan; 54(1): Design of a Parkinson s disease NGS diagnostic panel Illumina TruSeq custom amplicon + MiSeq 15 genes Exons 98% designable 425 bp amplicon size 2x250 reads v2.0 chemistry 262 amplicons; 241 targets Avoid SNP where possible PINK1 LRRK2 PARK2 PARK7 SNCA VPS35 POLG DCTN1 FBOX07 PLA2G6 CCH1 TH SPR ATP13A2 GBA Typical causal genes Atypical causal genes Risk Factor (5-fold) 42 14

15 43 Validating Parkinson s disease diagnostic panel: blind study Positive control samples = 49 Variants in genes = 62 6 exonic rearrangements 56 point mutations No exonic gene rearrangement could be detected still requires additional screening technique Sensitivity for point mutations = 89% Discovered 113 rare variants, of these 105 had coverage of >20 reads The remainder need to be Sanger sequencing confirmed Parkinson's disease diagnostic panel: blind study results Reasons for failing to detect mutations: PINK1 exon 1 failed amplification PARK2 deletion reads carrying this 40bp deletion NOT aligned consequence, mutation in trans was called homozygous rather than heterozygous 44 Deletion 45 Parkinson s disease NGS diagnostic panel: summary We have developed a targeted sequencing panel on the Illumina MiSeq Screens all currently known genes that can cause Parkinson's disease 89% effective needs a few design tweaks New analysis software to better pick up large insertions/deletions Research tool to study the genetics of Parkinson's disease Can diagnose the genetic basis of Parkinson's disease 15

16 Thank you!