SPOTlight deconvolution logFC, logFC cohen or mean.AUC

Question

[M'zelle]

Hello,

I am working on spatial transcriptomics and I would like to do deconvolution using SPOTlight, in the tutorial they use mean.AUC but I would like to use logFC. There are two different logFC available in my data: logFC.detected and logFC.cohen. Which one is the most relevant? How to determine the threshold to use?

In the tutorial, to keep only those genes that are relevant for each cell identity a filter is used for relevant marker genes, those with AUC > 0.8 are kept

I never heard about logFC cohen before and I don't find documentation about it, if someone can help me please do.

I hope you can understand what I explained, English is not my language