Is strand specific library always better for RNA-seq?

Question

[Twitty]

Hi there, Biostars community!

I'm planning a sequencing experiment for the de novo assembly of a non-model organism transcriptome (certain ontogenetic stage of a flatworm). The goal is in its general characterization, looking for the expression of some genes related to reproduction. For this organism there're no genomic data apart of some very general sequences, e.g. COX subunits and rDNA .

Now we're communicating with a sequencing facility and they say that in case of a such non-model organism, without even a draft genome, strand-specific library preparation won't give any useful info. To my mind it should as we will be able to resolve "intersecting" transcripts encoded by different strands of DNA. Indirectly it's supported by existence of strand-specific flags in modern de novo transcriptome assemblers (e.g. Trinity, rnaSPAdes).

I'm curious who of us is right? Are there any cases when usage of strand-specific protocols is senseless, can someone advice something? As I haven't found lots in a web, hopefully, someone here (if not here then where?!) has an experience.

Thanks in advance!

Answer 1

[lieven.sterck]

in essence you have a point indeed but also the facility has some arguments.

Indeed, in de-novo transcript analysis the strand specific will be not crucial and might even not be informative neither. On the other hand if you no have stranded RNAseq you already have that at hand when you might have a genome in the (near) future.

Theoretical you will be able to do the analysis as you described with the stranded info but in practice you will encounter many issues before you end up at the situation or analysis you mention. (Analysing de-novo transcriptomes is cumbersome with lots of pitfalls along the way)

Moreover, nowadays the stranded protocol is the default/standard so I don't really see the point to go for a non common protocol (cost can not be much different) and get the non-stranded info.