Functional Genomics with Next-Generation Sequencing Jen Taylor Bioinformatics Team CSIRO Plant Industry

Functional Genomics with Next-Generation Sequencing

Jen Taylor

Bioinformatics Team

CSIRO Plant Industry

CSIRO. INI Meeting July 2010 - Tutorial - Applications

Capacity and Resolution

• Next generation sequencing• Increasing capacity leads to increased resolution

Eric Lander, Broad Institute


How a Genome Works?

Parts Description• Function?• Interconnectedness?

Comparisons• Population - level• Between genomes


Application domains

Reference genome

No Reference Genome

Partially sequenced

UNsequenced

“PUN Genomes”


Impact of a Reference Genome

Sequence Data

Alignment

Read Density

Characterisation

Genome Assembly

Contigs


Applications of Next Generation Sequencing

• Profiling of Variation• Genetic variation• Transcript variation• Epigenetic variation• Metagenomic variation

• Discovery• Novel genomes• Novel genes• Novel transcripts• Small / long non-coding RNA

RNA Sequencing (RNASeq)

• Coding and non-coding transcript profiling

• Dynamic and Context dependent

Epigenomics

• Genome-wide protein-DNA interactions, DNA modifications

• Heritable and reversible regulation of gene expression

Today


RNASeq

• Qualitative – transcript diversity• Quantitative – transcript abundance

• Impact of NGS• Observation of transcript complexity

• Transcript discovery

• Small / long non-coding RNA

• Analytical challenges• Transcript complexity

• Compositional properties


RNASeq

Library Construction

Sample

Total RNA

PolyA RNA

Small RNA

Sequencing

Base calling & QC

Mapping to Genome

Assembly to Contigs

Digital “Counts”

Reads per kilobase per million (RPKM)

Transcript structure

Secondary structure

Targets or Products

Reference

PUN

Analysis


RNASeq – Transcript Complexity

Mapping :

• Reads with multiple locations

•Conserved domains ?

•Sequencing error ?

• Reads Spanning Exons

• Gapped alignments ?

• Sequencing error ?

Erange Pipeline : Mortazavi et al., Nature Methods VOL.5 NO.7 JULY 2008


RNASeq – Compositional properties

Depth of Sequence• Sequence count ≈ Transcript Abundance

• Majority of the data can be dominated by a small number of highly abundant transcripts

• Ability to observe transcripts of smaller abundance is dependent upon sequence depth


RNASeq – Compositional properties

Composition• Sequence counts are a composition

of a fixed number of total sequence reads

• Therefore they are sum-constrained and not independent

• Large variations in component numbers and sizes can produce artefacts

True Reads

RPKM


RNASeq - Correspondence

• Good correspondence with :• Expression Arrays• Tiling Arrays• qRT-PCR

• Range of up to 5 orders of magnitude

• Better detection of low abundance transcripts

• Greater power to detect• Transcript sequence polymorphism• Novel trans-splicing• Paralogous genes• Individual cell type expression


Reference Genome - RNASeq


Reference Genome - RNASeq

Human Exome

Number of exons targeted: ~180,000 (CCDS database)plus700+ miRNA(Sanger v13)300+ ncRNA


Epigenome

• Protein-DNA interactions [ChIPSeq]• Nucleosome positioning• Histone modification• Transcription factor interactions

• Methylation [MethylSeq]

• Impact of NextGen• Whole genome profiling• Resolution

• Analytical challenges• Systematic bias• Unambiguous mapping• Robust event calling

Image : ClearScience


ChIPSeq

MNaseLinker Digest

Sequence &Align

RemoveNucleosomes


ChIPSeq

MNaseDigest

Sequence &Align

RemoveNucleosomes


ChipSeq methods

Pepke et al., 2009

CisGenome

ERANGE

FindPeaks

F-Seq

GLITR

MACS

PeakSeq

QuEST


MethylSeq using Bisulfite conversion

Cytosine Uracil

Bisulfite conversion

Thymine

PCR

5-methylcytosine 5-methylcytosine Cytosine

Bisulfite conversion PCR


Limited publications from BS-Seq

• Mammals

• Methylation predominant occurs at CpG site

• Several publications in human

• One publications in mouse

• Plants

• Methylation occurs at CG, CHH, CHG sites

• Two publications in arabidopsis

H = A, G, T


Problems of mapping BS-seq reads

• Reduced sequence complexity

Cm methylated

C Un-methylated

Watson >>A Cm G T T C T C C A G T C>>


>>A Cm G T T T T T T A G T T>>

>>A C G T T T T T T A G T T>>


Problems of mapping BS-seq reads

• Increased search space

Watson >> A Cm G T T C T C C A G T C >> Crick << T G Cm A A G A G G T C A G <<

BSW >> ACmGTTTTTTAGTT >> BSC << TGCmAAGAGGTTAG <<


BSW >> ACmGTTTTTTAGTT >>BSWR << TG CAAAAAATCAA >>

BSCR >> ACG TTCTCCAAGA >> BSC << TGCmAAGAGGTTAG <<

PCR


ELAND

• Mapping reads to genome sequences

• Mapping reads to two converted genome sequences

• Cross match for reads mapping to multiple positions in converted genomes

• Mapping results were combined to generate methylation information

• Eland only allows 2 mismatches.

Lister et al. Cell (2008)


BSMAP

• Based on HASH table seeding algorithm

Xi and Li BMC Bioinformatics (2009)


Re-mapping of Lister’s data using BSMAP

Raw Reads MethodsUniquely

Mapped Reads

Unique and Nonclonal

Reads

Unique and nonclonal reads%

144,704,372

Eland 55,805,931 39,113,599 27.03%

BSMAP 67,975,425 48,498,687 35.52%

Lister et al. Cell (2008)


Methylation pattern throughout chromosomes

CHG

Crick

Watson

Position

Arabidopsis Chromosome 3

CG Watson

Crick

CHH Watson

Crick

Met

hyl

atio

n L

evel

/ 50

Kb

1.0

0.80

0.20


Partially / Unsequenced Genomes

Options for dealing with partial or unsequenced genomes• Wait for or generate the genome sequence

• ‘Borrow’ a reference genome from a phylogenetic neighbour

• Take a deep breath and ‘do denovo’

• Denovo Genome

• Denovo Transcriptome

DNA or RNA Sequence Data

Partial Sequence Database

Partial Assembly

Gene Annotation

Genetic Variation

Non-coding RNA

Transcript Variation


Plant Genomes – Haploid Size

Human

Arabidopsis

Rice

Potato

Sugarcane

Cotton

Barley

Wheat

Diameter proportional to genome haploid genome size


Plant Genomes – Total Size

Human Cotton Barley Sugarcane

Wheat


Denovo RNA Seq

• Why transcriptome ?• Large genome sizes with high repeat content are difficult to assemble

• Transcriptomes more constant size• Enriched for functional content

• Aims :• Transcript discovery• Small /long non-coding RNA profiling

• Analytical challenges• Assembly – ABySS, Velvet, Euler-SR• Comparisons between non-discrete, overlapping transcripts• Annotation• Ploidy


Summary – Impacts and Challenges

• RNASeq• Increased resolution

• Increased power for transcript complexity and variation

• Analytical challenges – transcript complexity, compositional bias

• Large gains in small and long non-coding RNA profiling

• Epigenomics• ChipSeq and MethylSeq

• Genome-wide with resolution

• Robust event calling is challenging

• Denovo transcriptomics• Attractive option for large, repeat rich genomes


Acknowledgements

CSIRO PI Bioinformatics Team

Andrew Spriggs

Stuart Stephen

Emily Ying

Jose Robles

Michael James

CSIRO Biostatistics

David Lovell

Documents

Functional Genomics with Next-Generation Sequencing Jen Taylor Bioinformatics Team CSIRO Plant Industry