Some basic question on quantification of RNA-seq mapped reads

Question

[Hamid]

Hello everyone, I'm trying to use stringtie to quantify some previously mapped RNA-seq reads. And I have some basic questions. I appreciate if someone explains to me in simple words:

1. How does certain quantifier's can only quantify the output of spliced aligners? Does it make any difference? After all, these are all mapped reads, what is specific about spliced vs non-spliced aligner outputs?

2. In case of stringtie and i'm sure many other qunitifiers, there is an option for specifying the strand, e.g.: --rf Assumes a stranded library fr-firststrand. --fr Assumes a stranded library fr-secondstrand.

first of all, what is a library? Is it the very first set of raw reads? Eventually, each read will map to one strand, so we will know the strand, is it not? What am I missing?

And my last question is about the XS tags, why does it matter only when we are talking about the junctions? Thanks

Answer 1

[Devon Ryan]

1. I suppose things like stringTie would sort of need spliced alignments to be normally useful, but in general normal quantification methods don't need spliced output. Either whoever told you that mistated things or didn't know what they were talking about.
2. You don't just extract RNA and sequence it, you have to process it such that you have something that actually can be sequenced. That final "something" is a "library".
3. In and of itself, the orientation with which a given read aligns (commonly incorrectly referred to as the strand it aligns to) has absolutely nothing to do with the strand from which the RNA it came from originated. Whether this orientation has any meaning at all is completely dependent on how the library was made. 90+% of libraries I see these day are directional (i.e., the alignment orientation has meaning) and use a dUTP based method, so --fr for stringTie or -s 2 for featureCounts is appropriate.
4. This refers to the splice acceptor and donor (so you can only find them for splice reads), which can be used to determine the likely orientation of a gene.

[Hamid]

Thank you so much for your response. So on 4, does it mean that XS is only reported if a read spans two exons (i.e. a junction)? and if yes, how is it used to determine the likely orientation of a gene?

[Devon Ryan]

1. I presume so, but I've never checked.
2. You can figure that out with a modicum of effort.