question about stranded RNA seq

[StephaniePi83]

Dear all,

I read a paper in which they performed a "strand specific RNA-seq librairies". I wonder how the strand is lost in a standard protocole (they said "although powerful, these strategies were limited in that the transcribed strand of origin was not captured " , and how a stranded mRNA seq protocole preserves this information ?

Thanks in advance for your reply.

[TonyBrooks]

Non stranded protocols convert mRNA (single stranded) into cDNA (double stranded) before library prep.
When you do this, you are then unable to tell the strand of the genomic DNA from which the mRNA was originally transcribed. The data obtained from the sequencer could feasible come from either cDNA strand.
There are two main methods of creating a stranded library. Firstly, you could adapt the small RNA protocol where adapters are ligated directly to the RNA one at a time. That way you get different adapters on the 5' and 3' end and only sequence in one direction.
A second method is to incorporate dUTP into the second strand cDNA synthesis. At the end of library prep, you then treat with USER which will digest the Uracil tagged strand, leaving you with only library from the first cDNA strand.

[pmiguel]

The third method I have seen (commercial kit) is anneal 3' extension-blocked random oligos to the 1st strand cDNA and extend the cDNA into specific adapter sequence.

--
Phillip

[huijia]

I am new in this field, so maybe my question is stupid. If the library is non-strand specific, does it mean when we align reads to genome, for each exonic region, there are about 50% of reads in the same strand of that exon and 50% on the other strand?

[pmiguel]

I am new in this field, so maybe my question is stupid. If the library is non-strand specific, does it mean when we align reads to genome, for each exonic region, there are about 50% of reads in the same strand of that exon and 50% on the other strand?

Yes,that is correct.
--
Phillip

[huijia]

Yes,that is correct.
--
Phillip

In that case, if half of mapped reads with wrong strand information. How can the downstream analysis be correct?

[pmiguel]

In that case, if half of mapped reads with wrong strand information. How can the downstream analysis be correct?

If the library is non-strand specific, you do not get strand information from it. Count and position may still be correct. If you need strand information you must make a strand-specific library.

--
Phillip

[huijia]

If the library is non-strand specific, you do not get strand information from it. Count and position may still be correct. If you need strand information you must make a strand-specific library.

--
Phillip

If my understanding is correct, reads of both strand in one position will be counted in exon in that position. If there is transcript only on one strand at that position, then the reads count is correct. If there are transcripts on both strand at that position, then with non-strand specific library, we can not get right reads count.

[pmiguel]

If my understanding is correct, reads of both strand in one position will be counted in exon in that position. If there is transcript only on one strand at that position, then the reads count is correct. If there are transcripts on both strand at that position, then with non-strand specific library, we can not get right reads count.

I don't see why this would be the case. All the genes will mostly be transcribed on one strand but both cDNA strands will be mapped. So comparisons among genes should be valid. The only issue is that it will not be possible to distinguish sense from anti-sense transcripts. As long as you are aware of this, I don't see a major problem.

However, if the strandedness of your template strands is critical to you for any reason, then you must construct a strand-specific library. It can be done, it just requires some extra work.

--
Phillip

[huijia]

I don't see why this would be the case. All the genes will mostly be transcribed on one strand but both cDNA strands will be mapped. So comparisons among genes should be valid. The only issue is that it will not be possible to distinguish sense from anti-sense transcripts. As long as you are aware of this, I don't see a major problem.

However, if the strandedness of your template strands is critical to you for any reason, then you must construct a strand-specific library. It can be done, it just requires some extra work.

--
Phillip

Philip, thank you for your detail answer. Why I am concern this issue is because our group focus on expression level of sense-antisense gene pair and the data is from other group with collaboration with us. So the strand problem bother me. Now I am more clear with this problem. Thank you.

[aslihan]

Thank you for the explanation. I am wondering how can quantify antisense transcripts from non-strand Specific RNA seq ?

Thanks

[frozenlyse]

You can assign strandedness to any reads across a (canonical) splice boundary (the GU/AG splicing signals in the intron being asymmetrical).