Identifying antisense transcript with stranded RNA-seq.

Question

[a.kmg]

Hello to all,

I have one dataset that contains samples (2 conditions in 2 replicates) sequenced with "TruSeq Stranded mRNA Sample Prep Kit" : directional mRNA-seq.

How is it possible to study the antisense transcripts? In fact, I have theoretical and technical questions about it.

• antisense transcripts have as template the sense strand of DNA (and thus have the same sequence as the antisense strand of DNA) ? So antisense transcripts are transcribed from + strand ? and conversely, sense transcripts have as template the antisense strand of DNA (and thus have the same sequence as the sense strand of DNA) ?
• when I align my data with the TopHat tool (with fr-firstrand option, because this is a dUTP protocol), I have a XS:A tag by read, following by + or -. So, reads that are tagged by XS:A:+ correspond to sense transcripts and reads that are tagged by XS:A:- correspond to antisense transcripts ?
• in this topic "http://genomebytes.wordpress.com/2013/07/08/fixing-the-xs-tag-in-tophat-output-bug-fixing/", Chung-Chau Hon said : "If a read is mapped on a feature specified in the GTF, the XS tag would always be on the same strand of the feature not matter what strand the read is mapped to. This would create wrong counts is there are unannotated antisense transcripts.". How do you study these antisense transcripts?

Thank you in advance for your response.

[Devon Ryan]

Hello velt!

It appears that your post has been cross-posted to another site: SEQanswers

This is typically not recommended as it runs the risk of annoying people in both communities.

Answer 1

[Istvan Albert]

As the Sequence Ontology states: antisense RNA is RNA that is transcribed from the coding, rather than the template, strand of DNA. It is therefore complementary to mRNA.

What this means is that the data will show as aligning to the opposite strand and orientation that you expect it.

In your specific a stranded protocol this (strangely) means that all read pairs where the read in the first file maps to the same strand as the feature that you are looking for will indicate an antisense strand.

[stan]

Hi Istvan,

Can you clarify for me the point you raised here in light of picking antisense transcription using HTSEQ-count.

You said "all read pairs where the read in the first file maps to the same strand as the feature that you are looking for will indicate an antisense strand." And when making read counts using Htseq-count it says "For stranded=yes, for paired-end reads, the first read has to be on the same strand and the second read on the opposite strand.

So my confusion or question is fo strand-specific data (dUTP protocol) mapped using tophat2 fr-firststrand, which --stranded option on HTSeq will give me read counts for antisense transcription. Because I was under the impression that the --stranded reverse option will give me the read counts for antisense transcription.

Thanks.

[Devon Ryan]

Please post questions as such and not as answers on old posts.

To your actual question, dUTP-based libraries convey strand with read #2, so htseq-count --stranded=reverse will produce sense counts for such datasets. Using --stranded=yes would then yield anti-sense counts. Htseq-count was unfortunately designed at a time when non-dUTP methods were still popular, so its default stranded option is opposite of what's now common.

[stan]

Thanks for the clarification Ryan. (Point taken about posts etc..)