Dear all,

I have two RNAseq data with 4 different treatments run in 2017 (Let's say treatment A and B) and 2019 (Treatment C and D); those have two different read lengths (I read for DE it does not matter the read length). For every experiment I have treated vs untreated. So I have 2 questions:

• Is it better to create one single dds object for all RNAseq experiments and then compare whatever I want using DESeq2 contrast option, or separate dds objects for each contrast group. I have tried to experiment a bit and I get different number of genes. Why is that and which method is more precise?.

• should I test treated vs untreated, or untreated vs treated? Also in this case i get inverse number of up/down regulated genes. Ie. In case of treated vs untreated I get more down-regulated genes and in case of untreated vs treated I get more up-regulated genes. I took one gene and controlled that its up/down regulation depends on this.