search

saeyslab / multinichenetr

MultiNicheNet: a flexible framework for differential cell-cell communication analysis from multi-sample multi-condition single-cell transcriptomics data
GNU General Public License v3.0
107 stars 14 forks source link

How do you determine which group the results are up-regulated in? why the circolplot incoherence with heatmap? #52

Closed SMU-TK-2022 closed 7 months ago

SMU-TK-2022 commented 8 months ago

dear author:

![Uploading result.png…]()

in the case ( MultiNicheNet analysis: MIS-C pairwise comparison - wrapper function). PDGFD-PDGFRB is down in M group .but in the circoplot PDGFD-PDGFRB is appear in M group but not in S group

browaeysrobin commented 7 months ago

Hi @SUM-TK-2022

There is no incoherence between the circos plot and the heatmap in https://github.com/saeyslab/multinichenetr/blob/main/vignettes/pairwise_analysis_MISC.md

PDGFD-PDGFRB is M-specific in both circos and heatmap (meaning, it's in the top n of interactions with highest prioritization score). However, and this is probably what you are referring to, the ligand activity of PDGFD-PDGFRB is more downregulatory in the M-group.

Why is this then an M-specific interaction? In the current version of the software, both up- and downregulatory activity are considered to prioritize interactions. Meaning that if an interaction, like PDGFD-PDGFRB is more strongly expressed in the M-group, but its target genes are enriched in the downregulated genes in M, we will still prioritize it because some interactions can suppress their target genes.

In the next version of MultiNicheNet (now on dev-branch), we will offer the possibility to the user to choose between including both up-and downregulatory activities for the prioritization (like now, considers potential suppressive effects of interactions) or only upregulatory activity (more easy to interpret).