Preferable metric to infer cell-to-cell communication about nichenetr

[Eisuan]

Dear Joshua,
thank you for using our tool.

I believe the approach you would like to utilize resembles the classical data-driven exploration for the inference of sender-receiver behaviours, such as the one proposed in the CellChat frameworks. Unfortunately, we currently do not provide filtering criteria to select a specific set of "presumably high-confidence" ligand-receptor interactions. This has partially to do with the variability observed in the AUPR scores among different datasets.

However, NicheNet and CellChat are complementary, because of the different strategies utilized to prioritize ligand-receptor interactions. I would approach the analysis by analyzing the communication dynamics using CellChat to clearly identify putative sender/receiver relationships. Then I would tailor my NicheNet analysis accordingly and I would examine the top-ranked ligands according to NicheNet and compare them with CellChat results.