Two main issues were discussed: 1. After batch correction there is one sample that clusters mostly independently from the other samples. This sample did not stand out in QC (althogh this will be re-examined). There is some evidence that this donor might have been clinacally different from the other donors, and since the batch correction worked well for the other samples, there is the possibility that this is real biologal signal. However since there is only one such sample it would be really hard to confirm this, and the inclusion of this sample would counfound the biologcal question of interest. I advisied to drop this sample post-clustering on the grounds that this donor might be too differnt from the others. 2. Prelimnary results from DEG by cell type for the three groups find mitocondral genes are the most differntly expressed. They are concerned as MT RNA could be a technical artifact (contamination) this tech isn’t the best to explore this. But there is are other findings that cellular metabolism may play a role in this condition. I advisied to keep the DEG results with the MT, add the caveat in the paper that this might not be a perfect experiment for this but it could support real differnces in the cells (plus they already have pretty strict MT QC). Also run a second DEA w/o MT genes to see what other signal exists.

Full logs