Starting out with mRNA-seq analysis

[James]

So I'm a newbie to this so have a few questions on what programs to use (just to get me started) that are good for the beginner, I've had a good look through the list of programs but I'd like some peoples opinions. For an average pipeline for illumina mRNA-seq data.

Align to my cds/genome: I have used bowtie to do this easy to use.

Count alignments in cds and create RPKM values: How can I do this?

Compare RPKM values from different data sets: I guess you could use scatter graphs to compare, any other good ways?

Visualization of the reads on the genome: The UCSC browser looks superb, however it doesn't have the genome of the critter I work on. What else is good for this?

Sorry for the awful newbie questions. Thanks, J

[Siva]

So I'm a newbie to this so have a few questions on what programs to use (just to get me started) that are good for the beginner, I've had a good look through the list of programs but I'd like some peoples opinions. For an average pipeline for illumina mRNA-seq data.

Align to my cds/genome: I have used bowtie to do this easy to use.

Count alignments in cds and create RPKM values: How can I do this?

Compare RPKM values from different data sets: I guess you could use scatter graphs to compare, any other good ways?

Visualization of the reads on the genome: The UCSC browser looks superb, however it doesn't have the genome of the critter I work on. What else is good for this?

Sorry for the awful newbie questions. Thanks, J

Hi
To count alignments and create RPKM you could use Cufflinks it does transcript assembly: check out: http://cufflinks.cbcb.umd.edu/

To get read counts per feature : You could use a script htseq-count (within http://www-huber.embl.de/users/ander.../overview.html). It takes a SAM output and compares it to a reference annotation (GFF and GTF) and assigns read counts per feature (exon by default).

If your lab is willing to pay you could use a proprietary aligner/assembler like http://www.softgenetics.com/NextGENe.html

and this forum will have many great suggestions so stay tuned

Siva

[shurjo]

Hi James,

My two cents worth:

I would use TopHat rather than Bowtie for the alignment so as to include reads mapping to splice junctions

Cufflinks and ERANGE both calculate RPKMs (or FPKMs in the case of Cufflinks). For getting raw counts of reads mapping to transcripts, in addition to htseq-count, BedTools has an utility called coverageBed which I use.

Comparing RPKMs may be as simple as a scatter plot, or if you need to do statistical tests, there are multiple R/Bioconductor packages designed for this purpose (DEGSeq/DEseq/edgeR/Bayseq).

Best of luck,

Shurjo

[James]

Thanks guys RPKM/FPKM is where I'm stumbling currently. Will try some of those out.

Thanks, J

[Siva]

For getting raw counts of reads mapping to transcripts, in addition to htseq-count, BedTools has an utility called coverageBed which I use.

Shurjo

Hi Shurjo
In BEDTools, can I use coverageBed using a SAM file and GFF file as inputs or should I convert the SAM file to a bed file? What kind of output does this produce?

thanks
Siva

[shurjo]

Hi Siva,

You need to convert both your reads and the gene annotation file to BED format (the latter can be easily downloaded from UCSC). BEDTools has an utility called bam2bed that will convert bam to BED, and if you have samtools you can easily convert from sam to bam. The output format is explained in section 5.9 of the BEDTools manual.

Regards,

Shurjo

[ScottC]

You might also find this review useful:

Computation for ChIP-seq and RNA-seq studies
Shirley Pepke et al.

http://pubget.com/paper/19844228

[alaincoletta]

At InSilico DB (https://insilicodb.org), we use the tophat-cufflinks-cummeRbund pipeline (www.nature.com/protocolexchange/protocols/2327)

There is an example of some of the results here: https://insilicodb.org/differential-...ng-cummerbund/

Align to my cds/genome: I have used bowtie to do this easy to use.

We use Tophat with Bowtie2

Count alignments in cds and create RPKM values: How can I do this?

We use Cufflinks

Compare RPKM values from different data sets: I guess you could use scatter graphs to compare, any other good ways?

We use Cuffdiff for deifferential gene expression

Visualization of the reads on the genome: The UCSC browser looks superb, however it doesn't have the genome of the critter I work on. What else is good for this?

You can send your BAM file to GenomeSpace and visualie them with IGV (http://genomespace.org)


Hope this helps