sequence assembly and next generation sequencing...

Sequence Assembly and Next Generation Sequencing Informatics

CBPS7711Oct 11 2011Oct 11, 2011

Sonia Leach, PhDSonia Leach, PhDAssistant Professor

Center for Genes, Environment, and Health

National Jewish Health

sonia.leach@gmail.com

Center for Genes, Environment, and Health

Next Generation Sequencing

• Increasing PresenceIncreasing Presence– ISMB 2010 Tutorial on Next Gen Sequencing

– ECCB 2010 Tutorial on Next Gen Sequencingq g

– RECOMB 2011 focus on Next Gen Sequencing

• Current Trends/Discussions– ChIP-seq, RNA-seq (mRNA, miRNA), Methyl-Seq

– Whole-genome sequencing

– Single molecule sequencing technologies

2Center for Genes, Environment, and Health

Introduction• Sequencing technologies

– 1st Gen: Gel based sequencing and Capillary electrophoresis

Introduction

– 2nd Gen: Next-Generation Sequencers:• Sequencing by Synthesis: Pyrosequencing with ePCR (Roche 454), Reversible Dye-termination with

cluster/bridge amplification (Illumina Solexa)

• Sequencing by Ligation: ePCR and diColor system (ABI SOLiD)

– 3rd Generation Sequencers: • Single molecule sequencing (ABI SMS, PacBio SMRT, Helicos), nanopore sequencing, …

• De novo assembly versus mapping to reference sequence– Human Genome Project (Hierarchical versus Shotgun Sequencing)– Human Genome Project (Hierarchical versus Shotgun Sequencing)

• Contig assembly and ordering

– Mapping/Alignment Programs• Repetitive regions (interspersed repeats (SINEs, LINEs, LTR elements, and DNA transposons),

satellite sequences, and low-complexity sequences)satellite sequences, and low complexity sequences)

• Polymorphisms and structural variants (insertions, deletions, substitutions, inversions, translocation)

• Sequencing applications– ChIP Seq – transcription factor binding and chromatin modification

– RNA Seq – transcript detection (boundaries, novel exons), quantification

– Whole Genome/Exome Sequencing - sequence variant detection

Early Sequencing Technologies (1977)• Maxam-Gilbert Sequencing

– double-stranded DNA segment ends labeled with 32P, divided into 4 samples, treated with chemical specific to 1-2 of 4 bases to nick DNA at few sites, treated with piperidine to destroy nicked base, creates fragments of length from 32P to nicked base, run fragments in 4 lanes on acrylamide gel to separate by size, , g y g p y ,auto-radiograph, read sequence off gel

– fallen out of favor due to technical complexity prohibiting its use in standard molecular biology kits, extensive use of hazardous chemicals, and difficulties with scale-upwith scale up

• Sanger Sequencing (chain termination method)– single-stranded DNA template and short primer for one end, add polymerase

with normal bases (dNTPs) and special dideoxynucleotides (ddNTPs) that cannot ( ) p y ( )extend chain at certain ratio to ensure variable length fragments, each ending with ddNTP is incorporated. Create one sample with each ddNTP flavor, run in 4 lanes on gel,, auto-radiograph and read sequence off gel

• Gilbert and Sanger shared 1980 Nobel Prize in Chemistry for • Gilbert and Sanger shared 1980 Nobel Prize in Chemistry for sequencing

Maxam-Gilbert Sequencing Sanger Sequencing

5From Recombinant DNA, 2nd Ed, Watson et al

Evolution of Sequencing

Gel Sequencing and

radioactive tags

Capillary Sequencing

and fluorescent ttags

Adapted from D. Pollock

Base call Quality• Used to assess sequence

quality and determine quality and determine accurate consensus sequence (originally from Phred program 1998).

Q alit scores Q • Quality scores Q logarithmically linked to base-calling error probabilities P:

Q = 10 log PQ = -10 log10 P

• Thus 10 means 1 in 10 probability base called incorrectly, 20 means 1 in 100 ( 99% b ll 100 (or 99% base call accuracy), 30 means 1 in 1000 (or 99.9%). Typical Q20 or Q30 for ‘high

li ’ b (

Roseisis 21:57, 22 February 2007 (UTC), image created by authors,quality’ bases (max out at 35-40, <10 unusable)

Rate of sequencing over past 30 years

8Center for Genes, Environment, and HealthMR Stratton et al. Nature 458, 719-724 (2009)

Evolution of SequencingRoche 454 (2005)

Gel Sequencing and

radioactive tags

Capillary Sequencing

and fluorescent ttags

Next GenerationSequencing

Illumina GAII (2006)

Sequencing

ABI SOLiD (2007)

• Adapter-ligated DNA fragments (400-500 bp) fixed

Roche 454

to small DNA-capture beads in a water-in-oil emulsion for emulsion-based PCR in PicoTiterPlate, a fiber optic

i i i http://www dkfz de/gpcf/886 htmlchip, mixed with Enzyme Beads containing also ATP sulfurylase + luciferase.

• 4 DNA nucleotides added

http://www.dkfz.de/gpcf/886.html

sequentially in fixed order, extending DNA strand, 'sequencing-by-synthesis’ (pyrosequencing) whereby pyrophosphate released upon nucleotide incorporation. Signal strength proportional to number of nucleotides -linear up to 8 homopolymers(3.3% error A5,~50% for A8)

11Center for Genes, Environment, and HealthRoche / 454 FLX

Roche 454

http://mammoth.psu.edu/howToSeqMammoth.html

Roche 454Roche 454

http://mammoth.psu.edu/howToSeqMammoth.html

Are homopolymers a problem, really?TCR alpha and Ig Light Chain

V’s CJ’sGermline DNA

Somatic DNA

• T-cell receptor alpha chain (John Kappler NJH)

• Gene undergoes somatic mutation in T-cells so every cell has different VJ combination from original germlinehas different VJ combination from original germlinetemplate

• Each mRNA sequenced has one V and one J region, must q J g ,map each separately to ‘reference’

Are homopolymers a problem, really?

• Representative Representative sequences for J

• Note high sequence g qsimilarity among different J and many homopolymerstretchesstretches

Ill i GAIIIllumina GAII• 'PCR on a chip' - primer-ligated fragments (1 on each end) hybridize to

slide which has complement of both primers, DNA extended, original molecule denatured and washed away. Extended molecule bends and hybrides to 2nd PCR primer (bridge amplification). Molecule extended on bridge, denatured, rebridged, extended, denatured, etc for 35 cycles, slide contains 'cluster' of double-stranded bridges - denatured

– For excellent graphical explanation of bridge amplication: http://seq.molbiol.ru/sch_clon_ampl.html

• Sequencing by Synthesis, with reversible termination so have competitive incorporation of all 4 bases per cycle, compile images across cycles to get sequence

– Technical issues: dephasing, error increases with length, polymerase problems with transversions/transitions

Illumina / Solexa

Genetic Analyzer

http://www.illumina.com/documents/products/techspotlights/techspotlight_sequencing.pdf

ABI SOLiD

Note: unlike 454/Illumina paired-ends, mate-pairs sequenced in same direction

Applied BiosystemsSOLiD

ABI SOLiD• Sequencing by

ligation with ligation with fluorescently-labeled di-base probes compete (2 bases, 3 (degenerate N, 3 universal Z)

• Successive cycles h di b means each di-base

seen twice.

ABI SOLiD• Di-base encoding allows accurate and fast detection/disambiguation

f i / lid l hiof sequencing errors/valid polymorphisms

Paired End/Mate Pair Reads• Illumina: paired end is both ends

Known Distancep

of ~300 bp fragment (shorter than a 454 read, shorter than most TEs)

454 i d d 3 8 20 Kb

Read 1 Read 2

• 454: paired ends span ~3,8,20 Kb

• SOLiD: paired ends span ~1, 2, 3, 5, 10 Kb5, 0 b

http://obi-linux.lbl.gov/alexa-seq-lbl.org/alexa_seq/BCCL/MCF7.htm

Paired read often map uniquely

Single reads can map to multiple positions Adapted from D. Pollock

Evolution of Sequencing3rd Generation

Sequencing(Single Molecule)

3rd-Generation Sequencing• Real time Single Molecule/Nanopore:

polymerase and single DNA molecule fixed to f E h f 4 DNA b l b ll d i h surface. Each of 4 DNA bases labelled with

fluorescent dye and when incorporated by polymerase, generates flash of light. System records time and color series of flashes (Pacific Biosystems, Helicos, ABI, Illumina)y , , , )

Example Single Molecule• Currently plagued by high

t (d h t error rates (dye perhaps not incorporated on dNTP, reaction too slow or fast for camera, etc) causes errors and phasing

bl i 2nd ti problems, as in 2nd generation technologies

– PacBio approaches problem by creating circular construct with DNA for repeated re-reads of same pmolecule, Helicos 2-pass sequencing

• Because real-time capture of incorporation, can distinquish nucleotide type by lag-time between incorporation (PacBio reports exciting implications for detecting methylated C, or A ( l ) ) l b A (plants) …), also strobe sequencing - PacBio

23Center for Genes, Environment, and Healthhttp://www.helicosbio.com/Technology/TrueSingleMoleculeSequencing/tabid/64/Default.aspx

C i f N G T h l iComparison of Next-Gen TechnologiesSanger Roche 454

GS FLXIllumina GAII

ABI SOLiD Illumina HiSeq

Helicos tSMS

S l i 1 f / 1 f / i d 2 f / 5 20 2 f lSample requirement 1 µg frag/ 5µg paired

<1 µg frag/paired 2 µg frag/ 5‐20 µg paired

<2 µg frag only

Time for library prep

- 3-4 days 2 days 2-4.5 days <6 hours 1 day

Amplification - bead-based Bridge Bead-based Bridge n/a (single method emulsion PCR amplification on

flow cellemulsion PCR amplification on

flow cellmolecule sequencing)

Time for amplification

- 2 days 4 hours 2 days

Sequencing method Dye pyrosequencing Reversible dye Sequence-by- Reversible dye Virtual Sequencing method Dye terminators

pyrosequencing Reversible dye terminators SBS

Sequence byligation (dibase)

Reversible dye terminators

Virtual terminator SBS

Read length 800 bp 400 bases (800 by 12/2010)

36-100 bases 36/ 2x50/ 2x100 bases

45-50 bases

Time for sequencer 3 hours 10 hours 2.5 days 6-10 days 1.5 -8 days 8-9 days

Total bases per run 800 bp 400-500 Mb 3-6 Gb 10-20 Gb 26-200 Gb 21-28 Gb

Total bases per day 6400 bp ~1 Gb 1.5 Gb 1.7-2 Gb 25 Gb (2x100) 2.5 Gb

Cost per run $4-5 for ~750bp

~$8k ~$9k $17k < $10k(2 human, 30x)

Cost per Mb ~$84.39 ~$5.97 ~$5.80

Data per run 15 Gb 1 TB 15 TB

Compiled from Turner et al, Mamm Genome. 2009; Voelkerding et al, Clin Chem 2009; Illumina.com 2010; wikipedia 2008

PR Space versus Science Space• Alignment problems from length - 97% E coli uniquely aligned

with 18bp reads, only 90% of humna with 30bp reads – see Voelkerding, 2009

• Flow and phasing - skip basecall in cycle

• Data quality and Error rateData quality and Error rate– Variation along sequence (higher error at 3’ end as reagents begin to fail, error for next

base in elongation depends on identity of previous base, homomers)

– Quality scores (Equivalence? – can use GATK for calibration)

• Length distribution versus average– mate pair/paired end insert size

• Coverage versus GC content Log (cov wind/cov chrom)GC

Coverage versus GC content

• Raw versus recovered sequence - Uncallable bases waste reads -How much coverage with different methods? What about multi-mapped reads?

T i (b d ) d l i l i

Log (cov wind/cov chrom)

ABI BioScope Manual June 2010

25Center for Genes, Environment, and HealthAdapted from D. Pollock

• Tagging (barcodes) and multiplexing - Variation in coverage for each individual in multiplex

Comparison and Usage• Short Reads (ABI/Illumina) • Long Reads (454/single mol)

– Resequencing b/c high throughput runs (coverage)

– SNP detection (coverage)Mi RNA (23 b )

– Many fewer reads (lower coverage)

– Much longer (implications for – Micro RNA (23 bp)– Counting (e.g.,

transcriptome profiling, ChIPSeq)

assembly, indel detection)

– Better for de novo sequencing (can bridge scaffolds)

B tt i bilit t • Adjustable dynamic range ($$)

– Hard to place near repetitive elements

– Better in ability to span repetitive regions (find unique seq)

– Amplicon and tagging– Harder to assemble de novo– 75-100 bp reads intermediate

Amplicon and tagging• longer reads detect varying copy

numbers btwn individuals

– Meta-genomics 16S RNA d 500b d t • 16S RNA, need 500bp reads to see small variant region among species

RNA SRNA-Seq• cDNA made from transcript, sequenced, mapped to genome

– Seq more sensitive than micro-arrays since do not have to design array probes or decide chip content, especially important for microRNA (~22nt total)

– Seq can be used to distinguish isoforms, allelic expression, sequence variants

– Seq more expensive than microarrays and suffer from 5’/3’ biases, high abundance transcripts (housekeeping ribosomal) which make up the majority abundance transcripts (housekeeping, ribosomal) which make up the majority of sequencing data (e.g. in some tissues 5% of the genes represent up to 75% of the reads sequenced).

• Mapping to reference genome means recognizingover known/novel splice junctions, determining exon boundaries, potentiallydiscovering novel exons (TopHat, Cufflinks)

• Quantification done at exon or transcriptlevel (what about when exons shared?)

• Differential RNA expression between samples is measured as differences incoverage

29Center for Genes, Environment, and HealthImage from wikipedia (user:Rgocs)

ChIP S ChIP-Seq (Chromatin Immunoprecipitation + Sequencing)

• Requires read alignment and

k fi dipeak finding

30Center for Genes, Environment, and HealthImage from wikipedia Data from Dr Barbara Wold

Sequence AssemblySequence Assembly

Center for Genes, Environment, and Health 31

Short Read AlignmentShort Read Alignment

Center for Genes, Environment, and Health 38

44Center for Genes, Environment, and Health(Bioinformatics (2009) 25 (14): 1754-1760)

Example of BWT

Let W=‘go’ be a query substring of X. Then the SA interval of all occurrences of W is [1,2]. The suffix array values for this interval are 3 and 0 which give allpositions of the occurrences of ‘go’. Procedure called ‘backward search’ usesproperties of W and B to efficiently calculate. See Li and Durbin (Bioinformatics (2009) 25 (14): 1754-1760)

(2009) 25 (14): 1754-1760)

Sequence Variant Detection

W Lee et al. Nature 465, 473-477 (2010)

Sequence Variant Detection• Single Nucleotide Polymorphismsg y p

– Compute genotype from distribution of 4 nucleotides for reads mapping to same reference location

• Libraries can contain duplicate reads, inflates coverage, use Libraries can contain duplicate reads, inflates coverage, use unique

– Use coverage, read position to determine accuracy• SOLiD accurate dueSOLiD accurate due

to dibase encoding andrules for valid adjacent

– Homozygous SNPs Likely sequencing

Likely SNPs

more accurate thanheterozygous SNPs

• Typically cut-off for

sequencing errorNon-called

bases or indels?

Phasing?minimum number of readscalling each variant of het

Voelkerding et al, Clin Chem 2009

• Small Insertion-Deletions can be detected with Small Insertion Deletions can be detected with gapped-alignment of reads

• Large Insertion-Deletions suggested by g gg ydeviations from expected insert size for paired reads

• Transversions suggested by correct spacing Transversions suggested by correct spacing but unexpected orientation of paired ends, also suggested by drop in coverage for normal pairs

• Translocations suggested by unexpected, mapping of paired ends to different chromosomes

• In all examples, note importance of good quality reference genome, long length of insert for paired reads (mix of insert size also beneficial)

• Copy Number VariationCopy Number Variation• Can use difference from expected coverage to

indicate CNVs (e.g. see CNV-seq by Xie and Tammi, Bioinformatics, 2009)

• SOLiD algorithm: compute log-ratio compute log-ratio of coverage to expected coverage (GC t d) (GC corrected), use HMM to estimate copy number, take

pyViterbi sequence

Additional Resources• Voelkerding et al, Clinical Chemistry 55(4):641 (2009) g , y ( ) ( )

“Next-Generation Sequencing: from basic research to diagnostics”

M t l S i 287 2196 (2000) “A h l • Myers et al Science 287:2196 (2000) “A whole-genome assembly of Drosophila”

• Medvedev et al, Nat Methods. 2009 Nov;6(11 Medvedev et a , Nat Met ods 009 Nov;6(Suppl):S13-20. “Computational methods for discovering structural variation with next-generation sequencing”

D id P ll k’ lid f l t ( i ll th • David Pollock’s slides from last year (especially math for detecting coverage needed to detect sequence errors))

sequence assembly and next generation sequencing...

Documents

applications of next generation sequencing

next generation sequencing...

next-generation sequencing next-generation sequencing...

next generation sequencing presentation

overview of next generation sequencing platform …...

next-generation sequencing data analysis · next-generation...

next-generation sequencing next-generation sequencing ... ·...

next generation sequencing gene panels for targeted ... ·...

tumor sequencing and next-generation sequencing

transcriptomics and next generation sequencing · next...

next-generation sequencing

next generation sequencing (ngs): procedure ed...

an introduction to next-generation sequencing for … of...

cameras for microscopy and next-generation sequencing 2019...

next generation sequencing (ngs)

sharif a principles of next generation sequencing ... of...

next generation sequencing - pbgworks

next generation sequencing - visualization -

introduction to next generation sequencing

sequencing: the next generation