Analyzing CITE-seq data

Hello,

I wonder if someone could please advise on two points in the scanpy guidance for CITEseq analysis please.

  1. When clustering protein and RNA together it suggests using sc.tl.leiden_multiplex, however I get this error: AttributeError: module ‘scanpy.tools’ has no attribute ‘leiden_multiplex’ … I cannot find leiden_multiplex on the sc.tl documents … does it exist?

  2. On line 36 it performs pbmc.X[:, (pbmc.var[“feature_types”] == “Gene Expression”).values] = rna.X … however i get a shape mismatch (ValueError: shape mismatch in assignment), despite following the steps meticulously. I’m wondering whether the RNA filtering steps (which are part of the tutorial) maybe caused this? Has anyone else come across this?

I believe these functions you’re trying to use are not officially in Scanpy and the CITE-seq tutorial is more exploratory at the moment.

As a shameless plug, I would encourage you to try scvi-tools, and specifically totalVI (tutorial here)

Thank you. I’ll be sure to give that a try.

To join @adamgayoso with another shameless plug, we’ve performed some CITE-seq analysis with the Python multimodal framework (see some notebooks e.g. here).
That should also address some of the points from the original post e.g. regarding multiplex clustering.

Thanks for the info. I’ll definitely take a look at using this.

1 Like