Does result data after DESeq2 analysis contain normalised counts ?

Question

[hariprasadp.iitkgp]

Hi, I have perdormed DGE analysis on a dataset using DESeq2. But when I compare the counts of genes in result data and the normalised counts (done using normalized_counts <- counts(dds, normalized=TRUE)) there is a lot difference in the counts. So which data do I need to use ? Should I take normalised counts as input for deseq ? My dataset contains different genes and different samples(that are in different conditions .. two control and two expreimental ). If those values need to be converted as reads per million, how to do so?

Answer 1

[ATpoint]

DESeq2, as clearly states in the manual, starts from raw counts.

If you want the normalized counts either use the code you show or, if you need it for clustering or similar applications, better transform the raw data with vst to get variance-stabilized counts on approximately log2-scale. Please read the manual, this is all in the FAQ section.

https://bioconductor.org/packages/release/bioc/vignettes/DESeq2/inst/doc/DESeq2.html