Hi,

I am analyzing gene expression from bulk RNA sequencing using Deseq2. Initially the normalised count matrix was generated using Deseq function. Also the raw count was normalised using rlog. I have read that for differential testing, rlog normalised counts should not be used. Can the rlog normalised count be used for visualizing (inferring using boxplot) the difference in expression between samples and a t-test be performed between samples?

Should we not use rlog normaiised count other than for PCA and machine learning applications?

Thank you.

Is it valid to do gene comparisons on VST or rlog transformed data?