RNA SEQ

IMGS 2012Bioinformatics Workshop:RNA Seq using Galaxy

Typical RNA_Seq Project Work Flow

Sequencing

Tissue Sample

Cufflinks

TopHat

FASTQ file

Gene/Transcript/ExonExpression

Visualization

Total RNA

mRNA

cDNA

StatisticalAnalysis

JAX Computational Sciences Service

Focus for Today

RNASeq Tasks, Tools and File Formats

Quality Control

Alignment

Summarization

FastQ,SangerFastQ

Cufflinks

TopHat

FastQC

SAM/BAM

GTF

Differential GeneExpression

Cuffdiff,Edge,DESeq, baySeq

Task

Tool

File Format

IGV

Tools

History

Dialog/Parameter Selection

ftp://ftp.ncbi.nlm.nih.gov/pub/church/GenomeAnalysis/h1-hESC_Sample_Dataset.fastq

Data upload review

Our data are H1 human embryonic stem cellRNA Seq data from the CalTech encodeproject. Single end reads from Illumina.

Typical RNA_Seq Project Work Flow

Sequencing

Tissue Sample

Cufflinks

TopHat

FASTQ file

Gene/Transcript/ExonExpression

Visualization

Total RNA

mRNA

cDNA

StatisticalAnalysis

JAX Computational Sciences Service

Prior to alignment, perform some quality control(QC) assessments of the data.

Here we use FastQC **.

**http://www.bioinformatics.babraham.ac.uk/projects/fastqc/

FastQC provides a wide range of QC checks. Herewe will only look at “Per base sequence quality”

Sequence quality per base position

The central red line is the median value

The yellow box represents the inter-quartilerange (25-75%)

The upper and lower whiskers represent the10% and 90% points

The blue line represents the mean quality

Good data

Consistent

High Quality Along the reads

Bad data

High Variance

Quality Decrease with Length

Quality Score

Position along sequencing read

Ourdata…

Galaxy has several tools for trimming sequences,removing adapters, etc. prior to alignment.

Using the information from FastQC, let’strim our input sequences so that theaggregate quality score is 15.

Typical RNA_Seq Project Work Flow

Sequencing

Tissue Sample

Cufflinks

TopHat

FASTQ file

Gene/Transcript/ExonExpression

Visualization

Total RNA

mRNA

cDNA

StatisticalAnalysis

JAX Computational Sciences Service

TopHat

Trapnell et al. (2009). Bioinformatics 25:1105-1111.

http://tophat.cbcb.umd.edu/

Figure from: Trapnell et al. (2010). Nature Biotechnology 28:511-515.

TopHat is a good tool foraligning RNA Seq datacompared to other aligners(Maq, BWA) because it takessplicing into account duringthe alignment process.

Setting parameters for TopHat in Galaxy

Be sure to use thequality trimmedsequences!

Does it seem like your Galaxy jobs never finish?!

Galaxy is increasingly popular so it can taketime for some of these computationallyexpensive processes to run…don’t restart yourjob or you will go to the end of the line!

Your job will continue to run on the Galaxyservers even if you shut down your computer.

$C:\Documents and Settings\cjb\Local Settings\Temporary Internet Files\Content.IE5\EYWCBOSJ\MP900178843[1].jpg$

For now we have pre-computed data toillustrate the mainpoints!

Visualizing alignments in Galaxy

When TopHat finishesthe alignments areavailable in BAM format.

You can look at thealignments in a variety ofbrowsers….

Which browser you choose is amatter of personal preference.

chr19:2,373,346-2,398,357

UCSC Browser…the track and thetitle of the track are madeautomatically for you from Galaxy.

UCSC also has controls to let youdisplay many other kinds ofannotations as tracks.

Click on an elementin the TopHat trackto see the details ofthe alignment…all ofthis information isstored in that verycompact BAM file!!!

Launch IGV (IntegratedGenome Viewer)

Typical RNA_Seq Project Work Flow

Sequencing

Tissue Sample

Cufflinks

TopHat

FASTQ file

Gene/Transcript/ExonExpression

Visualization

Total RNA

mRNA

cDNA

StatisticalAnalysis

JAX Computational Sciences Service

Cufflinks

Trapnell et al. (2010). Nature Biotechnology 28:511-515.

http://cufflinks.cbcb.umd.edu/

• Assembles transcripts,

• Estimates theirabundances, and

•Tests for differentialexpression and regulationin RNA-Seq samples

There are several ways to generate annotationfiles for Cufflinks to use.

Here we will create an annotation file using theUCSC genome browser tool in Galaxy.

A.There are many options for the features toinclude in the annotation file.

B.Cufflinks expects a GTF file format

Once you have selected your annotations…youcan send them directly to your history inGalaxy.

Setting parametersfor Cufflinks

Use the referenceannotations you justdownloaded…

http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSM521256

Example of an RNA Seq data set inNCBI’s Gene Expression Omnibus(GEO)…you don’t always need theraw sequences to do RNA Seq, youcan start with a SAM or BAM file.

SAM files need to beconverted into BAMformat in order torun Cufflinks….

There’s a tool inGalaxy for that!!

Cufflinks output canbe downloaded andviewed in Excel.

RPKM vs FPKM

•Reads Per Kilobase of transcript per Million mapped reads(RPKM)

–Used for single end sequencing reads

–Count # of uniquely mappable reads to a set of exons that constitute agene prediction/model.

•Fragments Per Kilobase of exon per Million fragmentsmapped (FPKM)

–Used for paired-end sequence data

•FPKM is an estimate of the number of reads per transcript

–TopHat aligns reads to the genome

–Cufflinks assembles reads into transcript models/fragments

–Cufflinks counts the number of reads per fragment to estimateFPKM

–FPKM is used as an indication of expression level for a gene

Quantification of gene expressionusing RNA Seq can be complicated byreads that don’t map uniquely to thegenome. RNA Seq by ExpectationMaximization (RSEM) takes mappinguncertainty into account whenestimating expression levels.