Recent changes in sc.tl.rank_genes_groups broke y-axis range in sc.pl.rank_genes_groups

y-axis range of sc.pl.rank_genes_groups seems unnecessarily long:

import scanpy as sc

sc.set_figure_params(dpi=72)
adata = sc.datasets.paul15()
sc.pp.log1p(adata)
sc.tl.rank_genes_groups(adata, groupby='paul15_clusters', method='t-test')
sc.pl.rank_genes_groups(adata, sharey=False)

Versions:

scanpy==1.5.2.dev52+g10937f5f anndata==0.7.4 umap==0.4.6 numpy==1.19.0 scipy==1.5.1 pandas==1.0.5 scikit-learn==0.23.1 statsmodels==0.11.1

[ivirshup]

My suspicion is that this is more likely to do with the plotting functions, than the calculation. I think the issue is that previously the axis limits weren’t being set (though they were calculated). Now they are set, but it turns out they were calculated on the full range of scores – not just the plotted ones. Here’s what a potential fix looks like:

Big images

Without fix:

With fix:

What do you think? (ping @fidelram)

[ivirshup]

Ah, that makes sense. Either way, I don’t think the intent of the function was to have the axis bounds determined by how many DE tests were saved.