Using Cufflinks For Rna-Seq Without Tophat/Bowtie.

Question

[Travis]

Hi,

I was wondering if anyone has successfully used Cufflinks in combination with an aligner from another source?

Since Bowtie doesn't make use of exon junction spanning reads I thought it would be a good idea to try an aligner or combination that considers spliced reads (e.g. RUM), but I am wondering if alignments generated in this way will be compatible with Cufflinks. Obviously the format will still be SAM/BAM but it is the nature of the spliced alignments whose compatibility I am questioning - not the file format.

EDIT: I guess I should ask specifically - does Cufflinks know how to interpret the junction spanning reads produced by the alternative aligner.

Thanks in advance.

Answer 1

[Gww]

The key thing you need to supplement your SAM/BAM file with is a special tag.

Note the use of the custom tag XS. This attribute, which must have a value of "+" or "-", indicates which strand the RNA that produced this read came from. While this tag can be applied to any alignment, including unspliced ones, it must be present for all spliced alignment records (those with a 'N' operation in the CIGAR string).string).

So basically you need to use splice site consensus sequences (ie. GT:AG) or known gene annotations to determine the strand for spliced transcripts. Some of the popular RNA-seq aligners already support this tag such as GSNAP and MapSplice.

Answer 2

[Leszek]

Hi Travis,

I have used GEM for mapping and converted results to SAM and then to sorted BAM. Then I feed those to successfully to Cufflinks.

Unfortunately, I was unable to code splice junction in the way TopHat does it. It codes splice junction in special way. So I lost splice junction information...

EDIT: You have to code junction information is special way in order Cufflinks can read those (contact the developers). I don't know any other mapper that provides that info except TopHat.

[brentp]

more accurate? how so? you can change the parameters to fewer mismatches or whatever but bowtie won't do anything you don't ask it to.

[Travis]

So in this case, what would you say the advantage of using GEM instead of Tophat/Bowtie was?

[Leszek]

GEM is faster and more accurate than bowtie.

Answer 3

[Sean Davis]

GSNAP works with cufflinks, it seems. Just output as SAM/BAM from GSNAP.

Answer 4

[Edge]

Hi flag,

You already try running RUM? How is the output result for RNA-seq analysis? I'm start running RUM for my RNA-seq Illumina data right not. Might need your advice to discuss how to interpret RUM output result.

edge

[Simon Cockell]

This is not an answer to the question. If you have your own questions about RUM, feel free to post them.