CellSign results have lost interaction orientation

[rchapple2]

I am having difficulty interpreting the results of a recent CellphoneDB DEGs analysis ran with the Cellsign module. I provided an active_TF.txt file to my DEGs analysis, but the results don’t seem to make sense. I’ve attached a snippet of the degs_analysis_CellSign_active_interactions_deconvoluted.txt file.

id_cp_interaction interacting_pair partner_a partner_b gene_a gene_b active_TF celltype_pairs active_celltype
CPI-SS0A28DCA72 CXCL12_CXCR4 simple:P48061 simple:P61073 CXCL12 CXCR4 STAT4 GEP1|GEP10 GEP1

If I am interpreting this correctly, this line would indicate that cells labeled as GEP10 are expressing CXCL12, and cells labeled as GEP1 are expressing the CXCR4 receptor and the STAT4 TF. If that is true, then I’m confused because neither CXCL12 is expressed by GEP10 cells nor is CXCR4 expressed by GEP1 cells. I know that STAT4 is expressed, but the upstream interaction doesn’t appear to be in the correct configuration. The deconvoluted.txt file is filled with hundreds of lines that show a similar pattern. Am I misinterpreting the output? Could I please get some assistance on this issue? Thank you for providing this resource!

[datasome]

Hi rchapple2,

Thank you for reaching out to us and for using CellphoneDB.

The correct interpretation of the above entry degs_analysis_CellSign_active_interactions_deconvoluted.txt file is that GEP1 expresses CXCL12 and that GEP10 expresses CXCR4. If you still think the method doesn't work as expected, please could you send a link to the output files of your DEG analysis to [email protected] and I will take a deeper look. Many thanks.

Best,

Robert.