Strand-specific library appears not strand-specific

[oligo]

I just received sequencing results of small RNA fraction (21-25nt) sequenced with HISeq and got puzzling results. Many of the sequences map to known small RNAs, but the mapping is on both strands, whereas the library was prepared as strand-specific.

Summary of library preparation:
(1) Small RNA fraction was isolated
(2) ligation to 3' adapter
(3) ligation to 5' adapter
(4) Reverse-transcription with specific primer
(5) all other steps, finally sequencing of 40nt.

Due to the specific ligation of the 5' adapter to the 5' end of the RNA I can't see how the library has lost its strand specificity completely (about 1:1 ratio).

Has anyone had a similar experience with such lack of "strandiness" with strand-specific library?

[scooter]

Since this is a small RNA library that presumably includes miRNAs, one possiblity is that you are detecting the mature miRNAs along with the 'passenger' strand RNA derived from the pre-miRNAs. During miRNA biogenesis the precursor is digested in a small double-stranded RNA assembled into the RISC complex. Either strand can be used as a miRNA, while the other is called the passenger strand. Both of these can be detected in an NGS experiment. In many cases only one will be more dominant, but that is not always true. In addition, I have seens tissue specific differences between the levels of the two possible miRNAs.

I guess one way to determine whether this is what you are observing is to see how these map to precursor miRNAs.

[oligo]

Hi Scooter, thanks for the answer, although it doesn't explain what I'm seeing.

This symmetric mapping can be found in regions that not contain perfect hairpins (so mapping to the passenger strand could not be explained by sequence similarity), and moreover, in regions that do not contain anything that resembles miRNA and hairipin structures. Finally, I'm considering right now only unique mappings, so I don't think it is a mapping ambiguity problem

[pmiguel]

Any possibility of it being genomic DNA contamination? (That is, did you do a DNAse treatment of your RNA?)

--
Phillip

[oligo]

Probably not the answer:
1. The intergenic regions are pretty much clean.
2. The sense to antisense expression is similar per gene, but is very different between genes.

[pmiguel]

Okay. That doesn't leave much. Any chance your data set actually derives from a PE run?

--
Phillip

[oligo]

Don't believe so. I got only one file and read identifiers were unique without pairs.

[pmiguel]

That leaves only increasingly crazy possibilities. Have you done this sort of library prep previously and had it give you strand-specific results? Are you seeing the correct adapter at the 3' ends of your sequence? What species is this?

--
Phillip