sc.queries.enrich throws AssertionError with floats

I’m trying to run an enrichment analysis after filtering out certain genes via sc.tl.filter_rank_genes_groups, so I use key='rank_genes_groups_filtered' as an argument for sc.queries.enrich. Since the filtered values are replaced with nan I hoped they’d by ignored in the enrichment analysis, but it actually leads to an uninformative AssertionError.

My suggestion here is simply to filter nan values from the gene list around here and 2 lines later: https://github.com/theislab/scanpy/blob/249fc572471683357b86b8bbf41d3284118bc8f8/scanpy/queries/_queries.py#L296

I can make a little PR if we agree with this simple fix

Note you can reproduce this very simply without an adata object (but of course the likely use case is with an adata object as outlined above):

sc.queries.enrich([float('nan')])

Output:

AssertionError: query failed with error 500

Versions:

scanpy==1.4.5.post2 anndata==0.6.22.post1 umap==0.3.10 numpy==1.18.1 scipy==1.2.1 pandas==1.0.1 scikit-learn==0.22.1 statsmodels==0.11.0 python-igraph==0.8.0

[ivirshup]

@chris-rands, that should be fixed now with #1054

[chris-rands]

Thanks!