Advanced Downstream Analysis with new mudata object

[cedmo001]

Hello!

Thanks for creating such an amazing tool! Just wondering when the downstream analysis for the new pipeline will be available?

Best Wishes,
C

[SeppeDeWinter]

Hi @cedmo001

Currently I'm quite busy (hence the late reply ;) ), so I'm not sure when this will be finished.

However, in the mean time you can convert your output to the old SCENIC+ object which you can use for all downstream analysis.

See: "Converting mudata output to old-style SCENIC+ object." in the tutorial: https://scenicplus.readthedocs.io/en/latest/human_cerebellum.html#Converting-mudata-output-to-old-style-SCENIC+-object.

Best,

Seppe

[cedmo001]

Hi Seppe,

Thanks for the reply!

Do you have a recommendation for a downstream analysis tutorial I should follow? Was trying to use the old 10x PBMC multiome tutorial but it seems like the converted mudata to scplus_obj is missing some of the data that was originally stored in the unstructured data, so some of the functions don't seem to work? Unless I'm missing something?

Best Wishes,
C

[SeppeDeWinter]

Hi @cedmo001

Are there particular functionalities you are missing at the moment?
I recently implemented the perturbation simulation: https://scenicplus.readthedocs.io/en/latest/Perturbation_simulation.html

Best,

Seppe

[wgao688]

Hi @SeppeDeWinter , where do we access the codebase for working with the "old-style" SCENIC+ object after the "Converting mudata output to old-style SCENIC+ object" section? It seems like that was able to produce figures like this. Thanks for your help!

[SeppeDeWinter]

Hi @wgao688

I'm afraid to say that this figure was drawn using cytoscape. We never head function to automatically draw such figures.

Best,

Seppe

[cedmo001]

Hi @SeppeDeWinter,

I guess I'm struggling with filtering only the (+/+) and (-/+) regulons, as this isn't included in the new tutorial, and can't seem to figure out how to do this on my own (python isn't my forte haha). As well, I would like to enrich for eregulons that have a correlation with gene expression and AUC values, and then be able to view these eregulons on the dot plot. As of now I can't figure out how to have the color modality in the dot plot represent the gene expression values and the size modality represent my filtered eregulons. Apologies if these are really easy fixes, but just can't seem to get it to work. Thank you!

Best,
C

[wgao688]

Hi @SeppeDeWinter, similarly I am dealing with the issue of having too many eRegulons and trying to narrow on some "stronger" candidates. Do you have any suggestions regarding filtering eRegulons, based for example on AUC, RSS, or other metrics? Also would you recommend focusing on "+/+" interactions out of the 4 possibilities since activators are more reliably identified than repressors?