Implementa)on
of
computa)onal
pipelines

to
support
next
gen
applica)ons 



Winter
School
in
Mathema0cal
and
Computa0onal
Biology,

6th

–
10th

July
2009


Roberto
Barrero

(rbarrero@ccg.murdoch.edu.au)


miR-Seq: microRNA profiling ChIP-Seq: Chromatin modification Bi-Seq: DNA methylation analysis

*

*

*

Bioplatforms Australia

NCRIS 5.1 Evolving Biomolecular Platforms and Informatics

NCRIS 5.1

Australian Bioinformatics Facility

Genomics Australia

Embedded Activities

Proteomics Australia

Metabolomics Australia

Embedded Activities

Embedded Activities

Non-Embedded Activities

Non-Embedded Activities

Non-Embedded Activities

Project 1

Project 2

Project 3

Project 1

Project 2

Project 3

Project 1

Project 2

Project 3

Development of cross –omics Platform Projects Development of cross NCRIS Investment Projects

NCRIS 5.16 NeAT

BioNeAT pending

NCRIS 5.16 NCI

Specialised Facility in

Bioinformatics pending

•  Implementa0on
of
a
short
read
mapping
pipeline


–  Benchmarking
of
freely
available
aligners

•  miR‐Seq:
Profiling
of
miRNAs
and
miRNAs*
•  ChIP‐Seq:
Defining
genomic
regions
associated


with
histone
modifica0ons


•  Bi‐Seq:
Determining
genome‐wide
CpG,
CHG
and
CHH
methyla0on
marks


Overview


•  Implementa0on
of
a
short
read
mapping
pipeline


–  Benchmarking
of
freely
available
aligners

•  miR‐Seq:
Profiling
of
miRNAs
and
miRNAs*
•  ChIP‐Seq:
Defining
genomic
regions
associated


with
histone
modifica0ons


•  Bi‐Seq:
Determining
genome‐wide
methyla0on
marks
on
CpG,
CHG
and
CHH
marks


Overview


Tool Name Performance Sanger Capillary ILLUMINA SOLiD 454 Finds

Mismatches Finds Indels Uses Quality Information

Tested platforms

ELAND

Large-Scale Alignment of Nucleotide Databases

FAST N Y N N Y N Y Linux. OsX

GMAP Genomic Mapping and Alignment Program

FAST Y N N Y Y Y N Linux

MOSAIK Reference guided aligner/assembler SLOW Y Y N Y Y Y Y Linux, OSX

SHRiMP Maps short reads to a reference sequence

SLOW N Y Y N Y Y N Linux

MAQ Mapping and Assembly with Qualities

FAST N Y Y N Y Y Y All BSD Platforms/Linux/OSX

NOVOALIGN Genomic Mapping and SNP/indel finder

FAST N Y N N Y Y Y Linux-64/ OSX

SOAP Short Oligonucleotide Alignment Program

VARIABLE N Y N N Y Y N Linux-64/32 /OSX

SSAHA Sequence Search and Alignment by Hashing Algorithm

FAST Y Y N Y Y N N Linux

Initial list of available tools (as of April 2008)

Tool Name Performance Sanger Capillary ILLUMINA SOLiD 454 Finds

Mismatches Finds Indels Uses Quality Information

Tested platforms

ELAND

Large-Scale Alignment of Nucleotide Databases

FAST N Y N N Y N Y Linux. OsX

GMAP Genomic Mapping and Alignment Program

FAST Y N N Y Y Y N Linux

MOSAIK Reference guided aligner/assembler SLOW Y Y N Y Y Y Y Linux, OSX

SHRiMP Maps short reads to a reference sequence

SLOW N Y Y N Y Y N Linux

MAQ Mapping and Assembly with Qualities

FAST N Y Y N Y Y Y All BSD Platforms/Linux/OSX

NOVOALIGN Genomic Mapping and SNP/indel finder

FAST N Y N N Y Y Y Linux-64/ OSX

SOAP Short Oligonucleotide Alignment Program

VARIABLE N Y N N Y Y N Linux-64/32 /OSX

SSAHA Sequence Search and Alignment by Hashing Algorithm

FAST Y Y N Y Y N N Linux

Initial list of available tools (as of April 2008)

•  ELAND does ungapped alignment of SE/PE reads up to 32 nt in length and generate accurate mapping qualities. •  MAQ uses probability models to measure the alignment quality of each read using sequence quality information.

•  SHRiMP uses seeding and a Smith-Waterman algorithm for aligning short reads to a reference genome.

•  RMAP map reads taking into account base-call quality scores to determine important positions.

•  NOVOALIGN finds global optimum alignments using full Needleman-Wunsch algorithm with affinity gap penalties.

Mapping Tools

Genome coverage range by distinct next gen applications

Bi-Seq ChIP-Seq

Small RNAs

Genome

Generation of Simulated Short Reads (1)

Tool: MAQ-simulate DNA template: Human chromosome 22 Read length: 36 bases Mutation rates: 0.1% up to 16% Number of reads: 70,000 x 3 per mutation rate Number of SNPs: 220~3,500 Indels: 10% probability of SNPs

Single-end mapping performance at various mutation rates

•  70,000 reads •  Triplicate

Pair-end mapping and SNP calling

Mapping performance of real data

HapMap project NA18507 10.2 million SE reads

5.1 million PE reads

Generation of Simulated Short Reads (2)

Dataset Simulated Set 1 Simulated

Set 2 Mutation Rate 0.1% 1.0% Number of single-end reads(1) 8,453,489 8,453,489 Number of paired-end reads(2) 16,906,978 16,906,978 Number of insertions 1,512 15,131 Number of deletions 1,514 15,166 Total number of indels 3,026 30,297 Number of Heterozygous SNPs 18,024 182,055 Number of Homozygous SNPs 9,034 90,985 Total number of SNPs 27,058 273,040 Total number of SNPs+indels 30,084 303,337

(1) Total number of single-end (SE) reads utilized in the comparisons (2) Total number of paired-end (PE) reads utilized in the comparisons

36bp-long reads datasets were generated using MAQ-simulate

Selected tools: NOVOALIGN, MAQ, BOWTIE, BWA

Arabidopsis thaliana (chr 5)

Mapping

(Colin et al. Submitted)

SNPs

(Colin et al. Submitted)

Indels

(Colin et al. Submitted)

Run Time

(Colin et al. Submitted)

Benchmarking Conclusion

NOVOALIGN is the best overall aligner for mapping both SE and PE reads as well as SNP calling and indel detection.

•  Implementa0on
of
a
short
read
mapping
pipeline


–  Benchmarking
of
freely
available
aligners

•  miR‐Seq:
Profiling
of
miRNAs
and
miRNAs*
•  ChIP‐Seq:
Defining
genomic
regions
associated


with
histone
modifica0ons


•  Bi‐Seq:
Determining
genome‐wide
CpG,
CHG
and
CHH
methyla0on
marks


Overview


miRNA function in Drosophila: •  Cell proliferation/anti-apoptosis

(bantam)

•  Fat storage/anti-apoptosis (mir-14) •  Homeostasis/anti-apoptosis (mir-278) •  Anti-apoptosis (mir-2) •  Photoreceptor differentiation (mir-7) •  Neurogenesis/neurodegeneration

(mir-9) •  Muscle differentiation (mir-1) •  Homeotic transformation (iab-4) •  Energy metabolism/fat storage (mir-278) •  Metamorphosis (let-7, mir-100, 125, 34)

microRNA

pathway


2L 2R 3L 3R

X 4 U

mir-959,960,961,962,963,964 mir-275, 305

mir-1002, 968

mir-306, 79, 9b

mir-100, 125, let-7 mir-2a-2 2a-1, 2b-2

Drosophila melanogaster microRNA clusters

mir-281-2, 281-1

mir-6-3, 6-2, 6-2, 5, 4, 286, 3, 309

mir-310, 311, 312, 313, 991, 992

mir-277, 34 mir-994, 318

mir-13b-1, 13a, 2c

mir-998, 11

mir-982, 303, 983-1, 983-2, 984

mir-283, 304, 12

mir-972, 973, 974 mir-975, 976, 977, 978, 979

•  148 dme-miRNAs •  17 clusters (=

Method Type of method Resource

miRanda Complementarity http://www.microrna.org/

TargetScan Seed complementarity http://www.targetscan.org/

PicTar Thermodynamics http://pictar.bio.nyu.edu/

Canonical site

Dominant seed

Compensatory site

Prediction of miRNA targets

Chelicerata

Crustacea

Myriapoda

Insecta

Pasani et al. (2004) BMC Biology

A timescale of arthropod evolution

•  
Ixodes
scapularis
–  
ESTs:
183,834
–  
Genome
Project
(version:
IscaW1;
released
3Dec08)



•  Supercon0gs:
369,492
•  Annotated
genes:
24,925
•  Pep0des:

20,486



•  
Rhipicephalus
microplus
–  
ESTs:
13,643


Tick Genomic Resources

Genome

Precursor miRNA (Pre-miRNA)

miRNA miRNA*

5P 3P

pre-dme-miR-33

Drosophila melanogaster

Rhipicephalus microplus

microRNA

locus


Simplified
data

processing
pipeline


Unique
Seq
Reads
(USR)


USR
w/o
adaptors


Retain
clone
count


Map
onto
genome


• 
NOVOALIGN
• 
Up
to
3
mismatches
• 
Single‐locus
mapping


Mapped
reads


miRBase


miRNA
clusters


• 
Extract
coordinates



of
miRNAs



‐
mature
miRNAs

‐
pre‐miRNAs


Illumina
Short
reads


Adaptor
removal


OUTPUTS miRNA,
miRNA*,
Mul0ple
Alignments,
etc


1.  Collect Total RNA/small RNA fraction •  Eggs •  Larvae (frustrated larvae, larvae) •  Adult ticks (female, male)

2.  Construct small RNA libraries 3.  Illumina/Solexa sequencing

•  Eggs: 4,215,404 •  Larvae: 9,437,803 •  Adult ticks: 8,319,734

4.  Data Analysis Pipeline

21,972,942 short reads

LARVA

NYMPH

ADULT

EGGS

female male

microRNA
discovery


0.0010


0.0100


0.1000


1.0000


We
found
58
miRNAs
in
Rhipicephalus
microplus

expressed
at
various
life
cycle
stages
that
are
highly
conserved
in
Drosophila
melanogaster.


Highly
conserved

0ck
miRNAs


Eggs (37)

26

Larvae (46)

Adults (44)

2 1

9 5

1

0

Fold

-incr

emen

t in

m

iRN

A ex

pres

sion

R

eads

Per

Mill

ion

0
20
40
60
80


100
120
140


Eggs
 Frus
Larvae


Larvae
Female
 Male


Eggs

Larvae

Adults

Pre-miRNA

Mature miRNA

miRNA:miRNA* co-expression

To
whom
it
may
concern:


Slides
containing
unpublished
data
were
removed.


We
appreciate
your
understanding.


RB.


mir-9a is conserved in the Ixodes scapularis genome

369,492 supercontigs

Finding I. scapularis miRNAs

BLAT onto D. melanogaster genome

Inspect known miRNA loci

Only mir9a was identified in the current I. scapularis supercontigs

•  Implementa0on
of
a
short
read
mapping
pipeline


–  Benchmarking
of
freely
available
aligners

•  miR‐Seq:
Profiling
of
miRNAs
and
miRNAs*
•  ChIP‐Seq:
Defining
genomic
regions
associated


with
histone
modifica0ons


•  Bi‐Seq:
Determining
genome‐wide
CpG,
CHG
and
CHH
methyla0on
marks


Overview


Active Gene Expression

Less Gene Expression

Acetylation Methylation

Implications of Chromatin Modifications

cisGenome


ChIP‐Seq
simplified

processing
pipeline


Map
onto
genome


Mapped
reads


FindPeaks


Illumina
Short
reads


OUTPUTS
Genomic
regions
associated
with
chroma0n
modifica0ons

NOVOALIGN

Ji et al. (2008) Nature Biotechnology 26: 1293-1300

One Sample Data Processing •  Scan genome with sliding windows and identifies regions with read counts greater than a user chosen cut off for bona fide binding regions.

•  False Discovery Rates (FDRs) are estimated by modeling the read count in nonbinding windows using a negative binomial distribution.

  Allows the background rate of occurrence of the reads to vary across the genome and to have a more flexible gamma distribution.

•  Use the directionality of reads to refine peak boundaries and filter out low-quality predictions.

cisGenome

Protein-DNA Interactions

Diverse genomic contexts for chromatin marks

Arabidopsis thaliana nucleosome

•  Implementa0on
of
a
short
read
mapping
pipeline


–  Benchmarking
of
freely
available
aligners

•  miR‐Seq:
Profiling
of
miRNAs
and
miRNAs*
•  ChIP‐Seq:
Defining
genomic
regions
associated


with
histone
modifica0ons


•  Bi‐Seq:
Determining
genome‐wide
CpG,
CHG
and
CHH
methyla0on
marks


Overview


Bisulfite Sequencing (Bi-Seq; BS-Seq)

Next
Gen
Sequencing


Genome‐wide
Methyla0on
Marks


Map
onto
genome


Mapped
reads


Check
Bisulfite
Conversion


Illumina
Short
reads


OUTPUTS
Bisulfite
conversion
report;
Genome‐wide
methyla0on
marks


MAQ

C T CpG CHG CHH

Sample C T Y Unconverted (Percentage) Converted

(Percentage)

run 1 10,806 10,577(97.88) 183(1.69) 14 (0.13) 10,760 (99.57) run 2 10,837 10,570(97.54) 219(2.02) 11 (0.10) 10,789 (99.56)

Sample Read

Sequences Unique

Alignments Gapped

Alignments Aligned run 1 11,653,511 133,712 4,129 275,944 run 2 11,540,171 132,690 4,251 273,806

Checking Bisulfite Conversion Efficiency

Aligned reads onto the Arabidopsis chloroplast genome

Bisulfite conversion efficiency of the chloroplast genome

Bisulfite conversion of the Arabidopsis thaliana chloroplast genome

Genome‐wide
Methyla0on
Marks


Coverage

CHG

CHH

CpG

chr1 chr2 chr3 chr4 chr5

hfp://ccg.murdoch.edu.au/yabi




Web
HPC
‐
Enabled


•  Zhang Bing •  Ala Lew-Tabor

Acknowledgements

Colin Hercus NCRIS 5.1

•  Zayed Albertyn •  Matthew Bellgard

An Australian Government Initiative

National Collaborative Research Infrastructure Strategy

Department of Primary Industries and Fisheries

Queensland Government

•  Frances Shannon •  Jun Fan

•  Liz Dennis •  Ian Greaves •  Sameer Tiwari