Hi,

I am new to scRNAseq and I am very confused about few things when it1s time to integrate two datasets/conditions (control vs treated).

So I "characterized" each dataset (control or treated) and identified the various clusters (which cells are what) tutorial. But I need to integrate the two datasets to see what happens to the cells when I treat my tissue with a drug so to my understanding I should follow this integration but:

1. why I have to split the dataset into 2 objects since in each dataset has one only condition? Should I not merge first all?

]2. once I merged, will I have to characterized again the dataset (which I assume contains only cells and genes shared across the two conditions) doing PCA, map, cluster identification? is this correct?

1. After the integration (harmony) when I look the the pre vd post umps, there is not much difference (below an example). how do I find then only those few genes that change within each cluster?

I am sorry if these questions are stupid but I find a lot of script with very few explanations why should I use X parameter or X approach then I think I am doing random things without understand them.

Thank you

Camilla