How should non-model species be analyzed?

[zhangpan-2018]

Your software is very helpful. if it is a non-model species and the library has already been rebuilt for CellChat analysis, do you need to rebuild the library again for FlowSight analysis? How should I proceed with the analysis?

[axelalmet]

Hi zhangpan-2018,

Firstly, I am so sorry about taking so long to reply to your question. You bring up an excellent question. There's a couple of ways you could go about this:

If you're using CellChat, the only step of the FlowSig analysis that you'd need to modify for a non-human/mouse organism is construct_inflow_signals_cellchat, because I had used OmniPath to infer potential downstream TFs activated by the ligand-receptor interactions listed by CellChat. If you have such a modified CellChat interactions list, you can use that directly to construct the potential inflow variables. In that case, I should modify FlowSig so that you can specify this database directly, e.g., as a Pandas dataframe, as an optional argument.

Otherwise, I think you would need to create a custom function that does everything construct_flows_from_cellchat does (see here), but allows you to work with non-model organisms.

Once you're past construct_flows_from_cellchat, however, everything else remains the same.