Transcript count normalization using Deseq2

Question

[Aiswarya]

Hi,

I need to look for difference’s in expression of two different transcripts from the same gene across three different stages of disease. Ie two different comparisons a) comparing expression of these two transcripts b) comparisons across different stages of disease

To do this I am trying to use the “transcript count matrix “ generated by stringtie. I ran this output via deseq2 and got the normalized counts using “counts( dds, normalized =True) “ but I am not sure what kind of normalization it is doing and if it is the right step for my purpose.

Could someone help me out here.

Thanks for any help!

[ATpoint]

Bioconductor has a guided tutorial for differential transcript usage, I suggest you start there: https://bioconductor.org/packages/release/workflows/vignettes/rnaseqDTU/inst/doc/rnaseqDTU.html

Since transcripts of a single gene have large overlaps in terms of sequence content it is important to have a quantification and differential analysis method that is aware of that and allows to incorporate that "inferential uncertainty" into analysis. The linked workflow covers some basics, plus you can go through existing literature to read up on it. I think only looking at DESeq2 counts based on the stringtie output is not going to give you much insight.