Minimum cell cutoff in sc.pl.dotplot

[gokceneraslan]

• Additional function parameters / changed functionality / changed defaults?
• New analysis tool: A simple analysis tool you have been using and are missing in sc.tools?
• New plotting function: A kind of plot you would like to seein sc.pl?
• External tools: Do you know an existing package that should go into sc.external.*?
• Other?

Especially when we visualize large datasets with multiple categorical variables (e.g. patient, disease, cell type) using sc.pl.dotplot, and we use a sequence in the groupby argument (e.g. sc.pl.dotplot(ad, 'genex', groupby=['individual', 'disease_status', 'cell type'])), sometimes we end up with too few cells in some rows, in which summary statistics like fraction of nonzero expressors or mean expression are not very robust.

To avoid that, I think it'd be cool to have a minimum observation cutoff in the function, where e.g. min_cells=5 would show groupby combinations with at least 5 cells. Without this option, this sort of filtering becomes an annoying pandas exercise (which some might enjoy but possibly not everyone).

[ivirshup]

I wonder if min_cells is too specific, and if there is a more generalizable way to handle this.

Do we give any indication of how many cells are in a group right now? I feel like this would be important for the user to even know the stats could be unreliable.

Misc other thoughts:

• I think this could make sense to address in the plotting classes
• Could be cool to be able to pass a grouped anndata to sc.pl.dotplot, something like:

sc.pl.dotplot(
    adata.groupby([...]).select(lambda x: len(x) > 5),
    genes,
    ...
)

[gokceneraslan]

Do we give any indication of how many cells are in a group right now? I feel like this would be important for the user to even know the stats could be unreliable.

sc.pl.dotplot(..., return_fig=True).add_totals().show()

is one way to check the cell numbers, but there is no intuitive way to do the filtering.