I'm analyzing the single nuclei RNA seq data of 8 individuals (two brain regions from each individual). I have 4 patients and 4 controls and am trying to compare two regions in patients in terms of vulnerability to disease. I have some questions regarding it:
First, I already have controls to remove the tissue specificity in my comparison. But I found papers comparing different regions and don't have controls. So, I was wondering when having control samples are necessary? In the mentioned paper they integrated each region separately and concluded based on comparison of integrated two regions data.
Second, in my case how should I know whether integration should be done or not. Sorry if it's a naive question. But I saw in one of the post that if we are not looking for patient-specific differences there is no need for integration and it was not clear to me.
I hope you share your thought.