Two gates for 1 marker for different subsetting operations in prior knowledge driven phenotyping #144
Description
Dear Arj,
I have an edge case here and would like to ask you, if there is a good way of doing this in scimap, even if it is hacky.
One of our marker has a problematic SNR (surprisingly its aSMA). We use it to classify smooth muscle cells (SMC) and also myofibroblasts.
I attached the relevant part of the gating matrix:
SMCs are super super high for aSMA, so I could use a quite high and conservative gate to not get a lot of falsely labeled SMC cells (as they already branch from all' cells.
However, CAF_S3 (myofibroblasts) express aSMA quite a lot lower, in a range where many cells would be labeled false positive for aSMA given the problematic staining. Setting the gate low would however work for CAF_S3 as we pregate on Fibroblasts with other pos/neg markers, so in the case of CAF_S3 a lower threshold works. However, the lower threshold does not work for SMC.
Essentially I would need to use 2 different thresholds, one conservative for SMCs and one less conservative but only for subsetting fibroblasts.
I already thought of duplicating aSMA to aSMA_low and rescale this channel only for CAF_S3. However this is quite hacky and also results probably in a channel mismatch with the ome.tiff for napariGater.
How would you approach this @ajitjohnson ?
Thanks!