How can one identify "real" antisense transcripts in prokaryote genomes?

[dsher]

Hello everyone,

We are using directional RNA-seq (Illumina TruSeq® Stranded mRNA Sample Preparation Low Sample (LS) kit) to look at transcriptomes of marine microbes and have noticed that ~3-6% of our reads (after removal of rRNA both experimentally and computationally) map to the antisense of known genes. This can be very interesting biologically, but in some cases the antisense reads seem "sporadic" when compared to the sense reads, so we are looking for a computational method to differentiate between real antisense transcripts and "noise" generated during the library preparation.

Does anyone have any idea what is the level of "strand fidelity" one can expect during library prep, or ideas on how to computationally select "bona-fide" antisense transcripts?

Thanks
Daniel

[TiborNagy]

Maybe you can measure the antisense read coverage of the transcripts and calculate FPKM values, just like in "normal" RNA-Seq.

[cascoamarillo]

First, if you used directional RNA-seq protocol you should be able to obtain directional libraries. So, if you made it right, antisense or reverse reads just reflect that; reverse reads to the gene direction.
Second, I don't know too much about antisense or regulatory RNAs on prokaryotes; but they do exist. Maybe you have to leave it as a side project, but bacteria carries argonaute proteins which can be associated with small RNAs having regulatory activities over foreign gene, for instance:

http://www.ncbi.nlm.nih.gov/pubmed/24034694

As TiborNagy mentioned, look at the levels of these kind of antisense RNAs. You can use FPKM values or just measure count levels with htseq-count (-s reverse option). Don't know, on total mRNA experiment, the proportion of reverse reads that can be reflected (especially small RNAs). You normally have to do a specific library method for this kind of RNAs.