polyA containing reads in RNA-seq

[wenhuang]

Hi,

I am just curious if any of the existing softwares for RNA-seq analysis handles polyA containing reads. I am not aware of one that explicitly tries to map polyA containing reads.

Thanks!

[john_mu]

http://www.nature.com/nature/journal/v464/n7289/full/nature08872.html

This paper deals with detection of Polyadenylation sites from RNA-seq reads, but they do not release the software.

[epigen]

It depends on what you want to do with poly-A containing reads. The usual software does not look at your reads to classify them in any way, it just tries to map them. You can create something like an "artificial genome" containing sequences you want to exclude (or specifically want to extract), e.g. poly-A, repeats, rRNA etc., as a reference to which you align your reads. If this is what you're looking for, e.g. the SOLiD BioScope WT pipeline provides such sequences to be used as filters, but I'm sure you can also find them somewhere else.

[ryanmcg]

http://www.nature.com/nature/journal...ture08872.html

This paper deals with detection of Polyadenylation sites from RNA-seq reads, but they do not release the software.

I am interested in doing exactly what this paper describes, but their methods do not give me enough detail to do this (without already having a fair amount of programming knowledge).

Is there not a tool that is built to do this type of analysis? I imagine it would be popular.

[ryanmcg]

Here is the passage from their supplement. Any suggestions on how to implement this?

As described above, our mapping strategy allowed us to find putative novel polyadenylation sites.
We first identified all sequencing reads that did not initially map to the genome and either began or
ended with a run of at least four As or T. We then trimmed off the run of As or Ts and remapped
the reads to the genome using MAQ. At those reads that then mapped uniquely to the genome,
we inferred the precise base where cleavage occurred. To filter out cleavage sites possibly due
to sequencing errors, we removed putative polyadenylation sites where the downstream genomic
regions contained at least three As or Ts, reasoning that a sequencing error at the non-A or T
site might lead to mis-mismapping and spurious calling of a poly-A site.

[ryanmcg]

part of the way there: trimest

http://emboss.toulouse.inra.fr/cgi-bin/emboss/help/trimest

[Brian Bushnell]

You can filter or trim reads matching certain patterns, like poly-A, with BBDuk. For example:

bbduk.sh in=reads.fq out=trimmed.fq ktrim=r k=8 literal=AAAAAAAAAAAA mm=f rcomp=f

That will trim reads to the right, starting at the first poly-A of at least 8 in a row. If you only want to look at tails, you could use "restrictright=20" to only look for matches in the last 20bp of the read. Without the "ktrim=r" flag, it will run in filtering mode:

bbduk.sh in=reads.fq outm=matched.fq out=unmatched.fq k=8 literal=AAAAAAAAAAAA mm=f rcomp=f

"rcomp=f" means only look for a kmer and not its reverse complement (in this case, poly-T); you can turn that on or off.