Hi. Thank you for the great package.
I’m wondering if I understand the DEA process right and I’m also curious as to how it handles the batch.
My several questions:
- Do I get it right that the main difference between the vanilla and change methods is that vanilla just compares the means (which one is larger), while the change method requires a certain distance between the means as well ( |lfc| > delta ) ?
- How do the methods sample from posterior distribution? We provide the cell indices for the groups to compare, but what happens then? Looking here my guess is that
n_samplesrandom cells with replacement are drawn from each group, and their expression values are compared. Are those expression values corrected by the model, i.e. what I would get from
- How batch is treated in this process? Looking at the code I can see that batch information is passed around, and initially taken from the model, if it is not provided by the caller, but I couldn’t figure out how it affects the output.
I apologize if I couldn’t find where you have this written already.