How to retrieve 3'-UTRs from RNA-seq Data?

[rondon]

Hi there.
first, I tried to look in those links for my anwser:

http://seqanswers.com/forums/showthread.php?t=27272

http://seqanswers.com/forums/showthread.php?t=5383

http://seqanswers.com/forums/showthread.php?t=27702

but it didn't help.

I have some RNAseq data from Leishmania braziliensis ando also the anotaded genome. The transcriptomes are made of cDNA from mRNA and runned in Illumina. I must retrieve the 3'-UTR from all genes.
First I tried to write a small script calling those sequences by reads mapping, but sometimes the 5'-UTR from the next gene is too close or inside the 3'-UTR, so it's difficult to automate.
Then I tried to assembly the transcriptomes and align the contigs against annotated CDS to retrieve the 3'-UTR. It seems to work, but most times the 3'-UTRs came without polyAs. Is it correct?
Are there some good softwares to retrieve 3'-UTR easily?

thankyou very much.
Rondon

[syfo]

I am not sure to see what you mean by "retrieve 3'UTRs" exactly. Is the goal to improve the existing gene annotation (by extending the annotated 3'UTRs for instance)?

Also,

most times the 3'-UTRs came without polyAs

Since polyAs are added on RNAs you will not find them in the genomic sequence, the annotated CDS nor in the mapped reads. That is normal.

[rondon]

I have the annotated genome, without 3utr information. I also have the transcriptome reads. I want to pick every single 3utr. If I map against the genome, sometimes is impossible to decide where a 3utr finish and the next 5utr starts.

My assembled transcripts doesnt have polyA. But I already know what happened. It's cos' the library construction cut them off.

I just aligned the transcripts against CDS and cut the transcripts after the stopcodon alignment. Just want to know if it's right or if there is a tool to retrieve my 3UTRs.

[rondon]

Also,

Since polyAs are added on RNAs you will not find them in the genomic sequence, the annotated CDS nor in the mapped reads. That is normal.

If you read again what I wrote, you'll see that I was talking about assembled transcripts.

[syfo]

I have the annotated genome, without 3utr information. I also have the transcriptome reads. I want to pick every single 3utr.
If I map against the genome, sometimes is impossible to decide where a 3utr finish and the next 5utr starts.

Yes, that is a limitation of most RNA-seq technologies. As long as RNA fragmentation is required you can never be sure that a given read comes from the transcript it maps to. It can come from another splicing isoform or even from another gene as you mention.

What I do not understand is how "de novo" transcript assembly would solve that problem, because ambiguities will remain. If a read pair maps to a genomic area that is both annotated as 3'UTR and 5'UTR the only way I can imagine to solve the assignation issue is to simultaneously estimate expression levels by taking into account read quantities (with a method like Cufflinks for instance).

I just aligned the transcripts against CDS and cut the transcripts after the stopcodon alignment. Just want to know if it's right or if there is a tool to retrieve my 3UTRs.

When an annotated genome is available the mapping strategy is usually preferred because de novo transcript assembly is not a straightforward process. I guess that your alignments can tell you how much your assembly match with the annotation.