Next-generation sequencing: informatics & software aspects

Next-generation sequencing:informatics & software

aspects

Gabor T. MarthBoston College Biology Department

Next-gen data

Read length

read length [bp]0 100 200 300

~200-450 (variable)

25-70 (fixed)

25-50 (fixed)

20-60 (variable)

400

Paired fragment-end reads

• fragment amplification: fragment length 100 - 600 bp• fragment length limited by amplification efficiency

Korbel et al. Science 2007

• paired-end read can improve read mapping accuracy (if unique map positions are required for both ends) or efficiency (if fragment length constraint is used to rescue non-uniquely mapping ends)• instrumental for structural variation discovery

• circularization: 500bp - 10kb (sweet spot ~3kb)• fragment length limited by library complexity

Representational biases

• this affects genome resequencing (deeper starting read coverage is needed)• will have major impact is on counting applications

“dispersed” coverage distribution

Amplification errors

many reads from clonal copies of a single fragment

• early PCR errors in “clonal” read copies lead to false positive allele calls

early amplification error gets propagated into every clonal copy

Read quality

Error rate (Solexa)

Error rate (454)

Per-read errors (Solexa)

Per read errors (454)

Applications

Genome resequencing for variation discovery

SNPs

short INDELs

structural variations

• the most immediate application area

Genome resequencing for mutational profiling

Organismal reference sequence

• likely to change “classical genetics” and mutational analysis

De novo genome sequencing

Lander et al. Nature 2001

• difficult problem with short reads

• promising, especially as reads get longer

Identification of protein-bound DNA

Chromatin structure (CHIP-SEQ)(Mikkelsen et al. Nature 2007)

Transcription binding sites. (Robertson et al. Nature Methods, 2007)

DNA methylation. (Meissner et al. Nature 2008)

• natural applications for next-gen. sequencers

Transcriptome sequencing: transcript discovery

Mortazavi et al. Nature Methods 2008

Ruby et al. Cell, 2006

• high-throughput, but short reads pose challenges

Transcriptome sequencing: expression profiling

Jones-Rhoads et al. PLoS Genetics, 2007

Cloonan et al. Nature Methods, 2008

• high-throughput, short-read sequencing should make a major impact, and potentially replace expression microarrays

Analysis software(resequencing)

Individual resequencing

(iii) read assembly

REF

(ii) read mapping

IND

(i) base calling

IND(iv) SNP and short INDEL calling

(vi) data validation, hypothesis generation

(v) SV calling

The variation discovery “toolbox”

• base callers

• read mappers

• SNP callers

• SV callers

• assembly viewers

GigaBayesGigaBayes

1. Base calling

base sequence

base quality (Q-value) sequence

diverse chemistry & sequencing error profiles

454 pyrosequencer error profile

• multiple bases in a homo-polymeric run are incorporated in a single incorporation test the number of bases must be determined from a single scalar signal the majority of errors are INDELs

454 base quality values

• the native 454 base caller assigns too low base quality values

PYROBAYES: determine base number

PYROBAYES: Performance

• better correlation between assigned and measured quality values

• higher fraction of high-quality bases

Base quality value calibration

RawIllumina reads(1000G data)

Recalibrated base quality values (Illumina)

RecalicratedIllumina reads(1000G data)

… and they give you the picture on the box

2. Read mapping

Read mapping is like doing a jigsaw puzzle…

…you get the pieces…

Unique pieces are easier to place than others…

Non-uniqueness of reads confounds mapping

• Reads from repeats cannot be uniquely mapped back to their true region of origin

• RepeatMasker does not capture all micro-repeats, i.e. repeats at the scale of the read length

Strategies to deal with non-unique mapping

• Non-unique read mapping: optionally either only report uniquely mapped reads or report all map locations for each read (mapping quality values for all mapped reads are being implemented)

0.8 0.19 0.01

read

• mapping to multiple loci requires the assignment of alignment probabilities (mapping qualities)

Longer reads are easier to map

454 FLX(1000G data)

Paired-end reads help unique read placement

• fragment amplification: fragment length 100 - 600 bp• fragment length limited by amplification efficiency

Korbel et al. Science 2007

• circularization: 500bp - 10kb (sweet spot ~3kb)• fragment length limited by library complexity

PE

MP

• PE reads are now the standard for genome resequencing

MOSAIK

INDEL alleles/errors – gapped alignments

454

Aligning multiple read types together

ABI/capillary

454 FLX

454 GS20

Illumina

• Alignment and co-assembly of multiple reads types permits simultaneous analysis of data from multiple sources and error characteristics

Aligner speed

3. Polymorphism / mutation detection

sequencing error

polymorphism

Allele calling in “trad” sequences

capillary sequences:• either clonal• or diploid traces

Allele calling in next-gen data

SNP

INS

New technologies are perfectly suitable for accurate SNP calling, and some also for short-INDEL detection

Human genome polymorphism projects

common SNPs

Human genome polymorphism discovery

The 1000 Genomes Project

New challenges for SNP calling

• deep alignments of 100s / 1000s of individuals • trio sequences

Rare alleles in 100s / 1,000s of samples

Allele discovery is a multi-step sampling process

Population Samples Reads Allele detection

Capturing the allele in the sample

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1E-0

4

2E-0

4

5E-0

40.

001

0.00

20.

005

0.01

0.02

0.05 0.

10.

20.

5

Population AF

Pro

b(a

llele

cap

ture

d in

sam

ple

)

n=100

n=200

n=400

n=800

n=1600

Allele calling in deep sequence data

aatgtagtaAgtacctacaatgtagtaAgtacctacaatgtagtaAgtacctacaatgtagtaAgtacctacaatgtagtaCgtacctacaatgtagtaCgtacctac

aatgtagtaAgtacctacaatgtagtaAgtacctacaatgtagtaAgtacctacaatgtagtaAgtacctac

aatgtagtaAgtacctacaatgtagtaAgtacctacaatgtagtaAgtacctacaatgtagtaAgtacctac

Q30 Q40 Q50 Q60

1 0.01 0.01 0.1 0.5

2 0.82 1.0 1.0 1.0

3 1.0 1.0 1.0 1.0

Allele calling in the reads

1 2

1 21

1

1 2

Pr | Pr | Pr , , ,

Pr | Pr | Pr , , ,

Pr , , , |i

kT

ii n

l kT

nk ki i i n

i

nk k l l l li i

iG

n

B T T G G G G

B T T G G G G

G G G B

base call

sample size

GigaBayesGigaBayes

individual read coverage

base quality

More samples or deeper coverage / sample?

Shallower read coverage from more individuals …

…or deeper coverage from fewer samples?

simulation analysis by Aaron

Quinlan

Analysis indicates a balance

SNP calling in trios

2

2

2 22 2

2

2

2

2 2

2

11 12 22

1 111: 1 1

2 2 11: 111: 11 1

11 12 : 2 1 12 : 2 1 1 12 : 12 2

22 : 22 : 11 122 : 1

2 2

1 1 111: 1 1 11:

2 2 4Pr | , 1 1

12 12 : 2 1 12 2

1 122 : 1

2 2

M M M

F

C M F

F

G G G

G

G G GG

2 2 2

2 22 2

2 22

2

2 22 2

1 1 1 11 1 11: 1

2 4 2 21 1 1 1 1

12 : 2 1 1 2 1 12 : 1 2 14 2 4 2 2

1 1 1 1 122 : 1 1 22 : 1 1

4 2 4 2 2

1 111: 1

2 211: 11 1

22 12 : 1 12 : 12

22 : 1FG

2

2

2

11:

2 1 12 : 2 12

22 : 11 122 : 1 1

2 2

• the child inherits one chromosome from each parent• there is a small probability for a mutation in the child

SNP calling in trios

aatgtagtaAgtacctacaatgtagtaAgtacctacaatgtagtaAgtacctacaatgtagtaAgtacctacaatgtagtaAgtacctacaatgtagtaCgtacctac

aatgtagtaAgtacctacaatgtagtaAgtacctacaatgtagtaAgtacctacaatgtagtaAgtacctacaatgtagtaAgtacctacaatgtagtaAgtacctac

aatgtagtaAgtacctacaatgtagtaAgtacctacaatgtagtaAgtacctacaatgtagtaAgtacctacaatgtagtaAgtacctacaatgtagtaCgtacctac

mother father

childP=0.79

P=0.86

Determining genotype directly from sequence

AACGTTAGCATAAACGTTAGCATAAACGTTCGCATAAACGTTCGCATA

AACGTTCGCATAAACGTTCGCATAAACGTTCGCATAAACGTTCGCATA

AACGTTAGCATAAACGTTAGCATA

individual 1

individual 3

individual 2

A/C

C/C

A/A

4. Structural variation discovery

SV events from PE read mapping patterns

Deletion

DNA reference

LM ~ LF+Ldel & depth: low

pattern

LMLF

Ldel

Tandemduplication

LM ~ LF-Ldup & depth: highLdup

Inversion LM ~ +Linv & ends flipped LM ~ -Linv depth: normalLinv

Translocation

LM ~ LF+LT1 LM ~ LF+LT2 & depth: normal LM ~ LF-LT1-LT2

LT2 LT1

LM LM

LM

InsertionLins

un-paired read clusters & depth normal

Chromosomaltranslocation

LT

LM ~LF+LT & depth: normal& cross-paired read clusters

Deletion: Aberrant positive mapping distance

Copy number estimation from depth of coverage

Spanner – a hybrid SV/CNV detection tool

Navigation bar

Fragment lengths in selected region

Depth of coverage in selected region

5. Data visualization

1. aid software development: integration of trace data viewing, fast navigation, zooming/panning

2. facilitate data validation (e.g. SNP validation): simultanous viewing of multiple read types, quality value displays

3. promote hypothesis generation: integration of annotation tracks

Data visualization

New analysis tools are needed

1. Tailoring existing tools for specialized applications (e.g. read mappers for transcriptome sequencing)

2. Analysis pipelines and viewers that focus on the essential results e.g. the few mutations in a mutant, or compare 1000 genome sequences (but hide most details)

3. Work-bench style tools to support downstream analysis

Data storage and data standards

What level of data to store?

images

traces

base quality values

base-called reads

Data standards

• Sequence Read Format, SRF (Asim Siddiqui, UBC)[email protected]

• Assembly format working grouphttp://assembly.bc.edu

• Genotype Likelihood Format (Richard Durbin, Sanger)

Summary

Conclusions: next-gen sequencing software

• Next-generation sequencing is a boon for mass-scale human resequencing, whole-genome mutational profiling, expression analysis and epigenetic studies

• Informatics tools already effective for basic applications

• There is a need both for “generic” analysis tools e.g. flexible read aligners and for specialized tools tailored to specific applications (e.g. expression profiling)

• Move toward tools that focus on biological analysis

• Most challenges are technical in nature (e.g. data storage, useful data formats, fast read mapping)

Software tools for next-gen data

http://bioinformatics.bc.edu/marthlab/Beta_Release

Roche / 454 system

• pyrosequencing technology• variable read-length• the only new technology with >100bp reads

Illumina / Solexa Genome Analyzer

• fixed-length short-read sequencer• very high throughput• read properties are very close to traditional capillary sequences • low INDEL error rate

AB / SOLiD system

A C G T

A

C

G

T

2nd Base

1st

Bas

e

0

0

0

0

1

1

1

1

2

2

2

2

3

3

3

3

• fixed-length short-reads• very high throughput• 2-base encoding system• color-space informatics

Helicos / Heliscope system

• short-read sequencer• single molecule sequencing• no amplification• variable read-length• error rate reduced with 2-pass template sequencing

Data characteristics

Data standards

• different data storage needs (archival, transfer, processing) often poses contradictory requirements (e.g. normalized vs. non-normalized storage of assembly, alignment, read, image data)

• even different analysis goals often call for different optimal storage / data access strategies (e.g. paired-end read analysis for SV detection vs. SNP calling) • requirements include binary formats, fast sequential and / or random access, and flexible indexing (e.g. an entire genome assembly can no longer reside in RAM)