analysis within subset of cells

[Shubhra-B]

Hi, Thanks for your effort making this. I have been looking into the package and trying to evaluate how different the results are when using conventional pseudo-bulking + standard DEG calling v/s LEMUR. In the worked-out example we also notice that the location of the sample collection seems to have some importance as you can see in the umap, I attach here. So if, I were to evaluate just the effect of the treatment on lets say left-frontal cells, would just include the location as co-variate in the design or would you calculate the DEGs as it is, and see there are some genes which are lets say enriched within that region. Since, you will capture any difference in these cells in latent space and you do not want to decrease the n?

I hope I could explain myself

Best S