Identifying polyA tail sequences

[Ephraim Trakhtenberg]

Is it possible to use STAR for extracting info on presence or absence of polyA tail sequences for novel transcripts predicted by Cufflinks?

[Alexander Dobin]

Hi Ephraim,

it is possible to extract information about polyadenylation sites from STAR alignments, please see this post:

https://groups.google.com/d/msg/rna-star/5wqQ7HSzjEM/ns1BNvUZxBkJ

Then you can check whether these sites agree with Cufflinks predictions.

However, in my experience, standard RNA-seq protocols heavily deplete the 3' ends of polyA molecules, so you will find very few reads with A-tails, and only for highly expressed genes.

Cheers

Alex

[Ephraim Trakhtenberg]

Alex, thank you! Which library prep kit did you use when you encountered the issue with low polyA-tail yield?

I did two RNAseqs: one using Illumina polyA-selection kit (so maybe polyA tails get stuck to and lost with the beads), and the other RiboZero (Illumina) where total RNA depleted of ribosomal RNA gets reverse transcribed, so here it would be only the matter of primers (I assume?) though I am not certain if polyT primer is included in the random primer hexamer mix. Would appreciate any feedback.

Best,

Ephraim

[Alexander Dobin]

Hi Ephraim,

these are ENCODE RNA-seq libraries, both polyA-selected and Total Ribo-Zero, with dUTP stranded protocols. They are quite similar to various flavors of Illumina Tru-seq.

We added extra poly-dT primers to the standard random hexamers, however, surprisingly, it did not boost the A-tailed reads. Other groups had been successful with anchored poly-dT primers.

Cheers

Alex

[Ephraim Trakhtenberg]

Thank you Alex, this is very helpful. So polyA-selected and Ribo-Zero approaches both had similarly poor yields for polyA tails?
I can see the logic in using the anchored poly-dT primers, and (I assume) the only issue this approach could address is that poly-dT hexamers in the mix are being syphoned within the long polyA tails, but then simply adding more poly-dTs (like you did) should have also addressed this issue.

Best,
Ephraim

On Friday, August 15, 2014 4:16:03 PM UTC-4, Alexander Dobin wrote: