This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Compare expression level of isoforms among conditions - TPM or deseq2 normalised counts?

I am now performing transcript-level DE analysis using salmon-tximport-deseq2.

I have two questions: 1. what are the differentially expressed transcripts among conditions. 2. How isoforms of the same gene are differentially expressed among conditions.

  1. I simply input the salmon result by tximport with txOut = TRUE, then perform DE analysis in deseq2 at the transcript-level.
  2. I will only pick some genes of interest so it would not be a global expression analysis. For example, a gene has transcript A and B, I hope to plot their expression levels against different conditions in one line chart or box plot. From my understanding, I should not use the normalized counts generated by deseq2 since they are only normalized for library size. Since I will compare/plot expressions of different isoforms in one plot, the reads should be normalized for both library size and transcript length right? Then TPM might be good to use? In that case, I can directly use the abundance of the tximport output?
rna-seq deseq2 tximport salmon
• 1,959 views • link
written by tianshenbio tianshenbio

I would answer with a No to the strategies you suggest.

Can you please read existing literature first before posting? http://www.bioconductor.org/packages/devel/workflows/vignettes/rnaseqDTU/inst/doc/rnaseqDTU.html

Also here: https://mikelove.github.io/counts-model/transcripts.html

The difficulty is the uncertainty of transcript level estimates due to the large overlap of exons between transcript of the same gene so you cannot simply use txOut=TRUE and be done with it.

Please use search engines first, there are plenty of resources that will help you get a background.

• 0 views • link
written by ATpoint ATpoint

Hi, thank you for your reply and the links, will definitely read more about it. Do you mean the transcript counts generated by salmon are not a real representative of transcript expression? I think salmon is able to assign reads to the transcripts where they originated, so reads mapping to overlapping exons are already sorted during salmon quantification, no? Sorry if I understood it wrong.

• 0 views • link
written by tianshenbio tianshenbio

Please read the resources first. They will answer most of your questions.

• 0 views • link
written by ATpoint ATpoint

0 answers

No answers yet.

Log in to answer this question.

More posts like this