A lot of Poly-A after RNAseq completed

[memento]

Hi,
Is there any obvious explanation why there would be a lot (~50%) of poly-A (or Poly-T, since this was bidirectional lib prep) in the output ?
RNA was purified using Poly-A+ enrichment (purified twice). Is it possible that the library was overamplified, hence the "PCR-winners" (polyA/T have lower GC) ? Is the library usable and the only drawback would be a lot of sequencing wasted on poly-A ?

Is it possible that the RNA was already digested, hence the only fragment retained in enrichment were the poly-A them self?

Thanks!

[nucacidhunter]

Providing more information might make it easier to answer the questions.
For example, was the input RNA QCed by Bioanalaser or other methods? Are poly A streches the whole read or just part of the reads? How the 1st strand synthesis was primed and what kit was used for library prep? How many cycles were used for sequencing and what was average library fragment size?

[memento]

1) RNA was qced using Nanodrop + 2d gel
2) Whole reads are polyA (or T), both on 50bp SE (HiSeq) and 250bp(MiSeq, did it just out of curiosity as a part of other run). mRNA polyA is between 100-500bp so that would still make sense.
3) Priming/1st strand etc was done according to Wang2011 (PLOS), 13 cycles were used to amplify.
4) Average library size (after amp) was between 250-400 (need to subtract adapters ~100bp).

[nucacidhunter]

Is there any obvious explanation why there would be a lot (~50%) of poly-A (or Poly-T, since this was bidirectional lib prep) in the output ?

Priming/1st strand etc was done according to Wang2011 (PLOS), 13 cycles were used to amplify.

I cannot find any explanation. If bidirectional lib prep here means non-stranded, it would be contrary to the reference (Wang 2011 PLOS) because the method described in that paper is stranded library prep.

Is it possible that the RNA was already digested, hence the only fragment retained in enrichment were the poly-A them self?

RNA was qced using Nanodrop + 2d gel

mRNA polyA is between 100-500bp so that would still make sense.

This is less likely for two reasons: 1- you mention that you have checked RNA prior to use by gel, so you would have noticed if it was degraded, 2- I cannot think of any way that random primers used for 1st strand synthesis would prime polyA to a extent that comprises 50% of reads.

Is it possible that the library was overamplified, hence the "PCR-winners" (polyA/T have lower GC) ?

Less likely for such an extreme bias.

Is the library usable and the only drawback would be a lot of sequencing wasted on poly-A ?

It would be reasonable to have explanation for this before using the data.

You have not mentioned species or estimated GC content. There might be something there to explore. FastQC plots of data are also usually good start point for troubleshooting.

[ipeikon]

Any updates to this thread? Did you ever find out what was going wrong?