binarization

[Flu09]

i want to ask if binarization should be done for each condition alone or merged composed of multiple conditions. thank you.

[robertzeibich]

Hi @Flu09,

Have you made any progress? I have also opened an issue because I was curious how they generated the following:

I am working with snRNA-seq case-control data.

I would binarize and compare two conditions at a time.

I ran pyscenic on oligo cells.
Binarized the auc_mtx.
Split the auc_mtx between cases and controls.
Plotted auc_mtx, auc_mtx_case and auc_mtx_control.
The regulon FOSB(+) seems to be active in cases but not in controls.

I needed to adjust the threshold. So, if you want to use the binarized output, you would need to check the thresholds for each regulon and adjust accordingly.
Then you would need to binarize the adjusted regulons again.

np.where(auc_mtx['FOSB(+)'] > 0.07,1,0)

I also ran pyscenic on all cells. Extracted oligo cells. Picked a regulon. However, the results were less promising/clear.