[BioC] DESeq and transcript-wise analysis

[Ann Loraine]

Hello,

Is there anything that can assess differences at the individual splice
site level?

Testing at the isoform level is sometimes not useful because many,
many factors go into determining the final structure of a
fully-processed mRNA transcript, such as the transcription start site,
cleavage at the three prime end, splicing of many internal introns,
and so on.

I think it is much more interesting and useful biologically to focus
in on one or two aspects of transcription and test whether a condition
or treatment has affected that one aspect.

For example, I'd like to know if a condition or treatment affects
individual splice site preference.

ex):

5'
V1 XXXXX------XXXXXXXXX 3'
V2 XXXXX--------------XXXXX

In this simple example, we have two isoforms arising from the same gene.

To assess whether the condition has changed splicing, i.e., changed
which splice site is preferred, then I need to know how many
transcripts used acceptor site V1 versus acceptor site V2.

I can do this by counting reads that align across the intron. A
spliced read can support the V1 acceptor or the V2 acceptor, but never
both. So I can be sure of which choice the spliced read represents.

But let's say I have data from three treatments and three controls.
How can I determine whether the treatment changed the ratio of V1 to
V2 spliced reads?

Can I do something like calculate the percentage of V1 reads overall
and then compare the percentages using a t test?

Sorry if this question is horribly naive!

Best wishes,

Ann