BayraktarLab / cell2location

Comprehensive mapping of tissue cell architecture via integrated single cell and spatial transcriptomics (cell2location model)
https://cell2location.readthedocs.io/en/latest/
Apache License 2.0
324 stars 58 forks source link

Using per-celltype counts for downstream analysis #376

Open Dillon214 opened 3 months ago

Dillon214 commented 3 months ago

Hello Cell2location devs,

I've had a good time using cell2location for deconvolution and generation of per-celltype gene counts. I am especially interested in the latter functionality, as not many packages currently support it. I wonder, is it possible to use these celltype specific counts for things such as clustering or differential expression? Could you, say, cluster celltype A specific counts and get different populations compared to clustering celltype B specific counts? I would be greatly appreciative of any input your team could give on this, both practically (what normalizations do you reccomend, or filtering steps) and qualitiatively (is this a good idea, or should it not be attempted).

sincerely, Dillon Brownell

note: here is the relevant section of the tutorial for generation of celltype specific counts: image