Deseq2 normalization

Question

[Dana]

Hi! I am wondering why is the condition given as the design to the DESeqDataSetFromMatrix function is needed for normalization. For my understanding, the normalization factor for each sample is calculated as the median of all the geometric roots of the different genes in all samples - then why is the condition relevant? where is it taken into consideration? Thank you!

Answer 1

[ATpoint]

Condition is not relevant. The normalization works on the entire count matrix without any additional factors. You can see that by the fact that you can run the normalization just on a count matrix outside of the DESeqDataSet and on the fact that the size factors are always the same regardless of the design.