sc.pl.violin doesn't work with `groupby` and `ax` passed together

scottgigante-immunai commented 2 years ago

[X] I have checked that this issue has not already been reported.
[X] I have confirmed this bug exists on the latest version of scanpy.
[ ] (optional) I have confirmed this bug exists on the master branch of scanpy.

Minimal code sample (that we can copy&paste without having any data)

Option 1: group with two keys, passing two axes

import scanpy as sc
import matplotlib.pyplot as plt
adata = sc.datasets.pbmc3k()
adata.obs['group'] = adata.obs.index.to_series().str.startswith("A").astype(str)
fig, axes = plt.subplots(1, 2)
sc.pl.violin(adata2, keys=['CD8A', 'CD8B'], groupby="group", ax=axes)

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
Input In [51], in <module>
      4 adata2.obs['group'] = adata2.obs.index.to_series().str.startswith("A").astype(str)
      5 fig, axes = plt.subplots(1, 2)
----> 6 sc.pl.violin(adata2, keys=['CD8A', 'CD8B'], groupby="group", ax=axes)

File /opt/conda/envs/analysis/lib/python3.8/site-packages/scanpy/plotting/_anndata.py:835, in violin(adata, keys, groupby, log, use_raw, stripplot, jitter, size, layer, scale, order, multi_panel, xlabel, ylabel, rotation, show, save, ax, **kwds)
    833     axs = [ax]
    834 for ax, y, ylab in zip(axs, ys, ylabel):
--> 835     ax = sns.violinplot(
    836         x=x,
    837         y=y,
    838         data=obs_tidy,
    839         order=order,
    840         orient='vertical',
    841         scale=scale,
    842         ax=ax,
    843         **kwds,
    844     )
    845     if stripplot:
    846         ax = sns.stripplot(
    847             x=x,
    848             y=y,
   (...)
    854             ax=ax,
    855         )

File /opt/conda/envs/analysis/lib/python3.8/site-packages/seaborn/_decorators.py:46, in _deprecate_positional_args.<locals>.inner_f(*args, **kwargs)
     36     warnings.warn(
     37         "Pass the following variable{} as {}keyword arg{}: {}. "
     38         "From version 0.12, the only valid positional argument "
   (...)
     43         FutureWarning
     44     )
     45 kwargs.update({k: arg for k, arg in zip(sig.parameters, args)})
---> 46 return f(**kwargs)

File /opt/conda/envs/analysis/lib/python3.8/site-packages/seaborn/categorical.py:2408, in violinplot(x, y, hue, data, order, hue_order, bw, cut, scale, scale_hue, gridsize, width, inner, split, dodge, orient, linewidth, color, palette, saturation, ax, **kwargs)
   2405 if ax is None:
   2406     ax = plt.gca()
-> 2408 plotter.plot(ax)
   2409 return ax

File /opt/conda/envs/analysis/lib/python3.8/site-packages/seaborn/categorical.py:1043, in _ViolinPlotter.plot(self, ax)
   1041 def plot(self, ax):
   1042     """Make the violin plot."""
-> 1043     self.draw_violins(ax)
   1044     self.annotate_axes(ax)
   1045     if self.orient == "h":

File /opt/conda/envs/analysis/lib/python3.8/site-packages/seaborn/categorical.py:761, in _ViolinPlotter.draw_violins(self, ax)
    759 def draw_violins(self, ax):
    760     """Draw the violins onto `ax`."""
--> 761     fill_func = ax.fill_betweenx if self.orient == "v" else ax.fill_between
    762     for i, group_data in enumerate(self.plot_data):
    764         kws = dict(edgecolor=self.gray, linewidth=self.linewidth)

AttributeError: 'numpy.ndarray' object has no attribute 'fill_betweenx'

Option 2: group with two keys, passing first of two axes

import scanpy as sc
import matplotlib.pyplot as plt
adata = sc.datasets.pbmc3k()
adata.obs['group'] = adata.obs.index.to_series().str.startswith("A").astype(str)
fig, axes = plt.subplots(1, 2)
sc.pl.violin(adata2, keys=['CD8A', 'CD8B'], groupby="group", ax=axes[0])

No traceback, but the second axis is simply not plotted.

Option 3: group with two keys, passing one axis

import scanpy as sc
import matplotlib.pyplot as plt
adata = sc.datasets.pbmc3k()
adata.obs['group'] = adata.obs.index.to_series().str.startswith("A").astype(str)
fig, ax = plt.subplots()
sc.pl.violin(adata, keys=['CD8A', 'CD8B'], groupby="group", ax=ax)

No traceback, even though this should error. Plots just the first of the two keys.

Versions

``` WARNING: If you miss a compact list, please try `print_header`! ----- anndata 0.7.8 scanpy 1.8.2 sinfo 0.3.1 ----- PIL 8.4.0 anndata 0.7.8 asttokens NA attr 21.2.0 backcall 0.2.0 cffi 1.15.0 colorama 0.4.4 cycler 0.10.0 cython_runtime NA dateutil 2.8.0 debugpy 1.5.1 decorator 5.1.1 defusedxml 0.7.1 django 4.0 executing 0.8.2 google NA h5py 2.10.0 idna 3.1 igraph 0.9.8 importlib_resources NA ipykernel 6.7.0 ipython_genutils 0.2.0 ipywidgets 7.6.5 jedi 0.18.1 jinja2 3.0.3 joblib 1.1.0 jsonschema 4.4.0 jupyter_server 1.13.3 kiwisolver 1.3.2 leidenalg 0.8.8 llvmlite 0.37.0 markupsafe 2.0.1 matplotlib 3.5.1 matplotlib_inline NA mpl_toolkits NA natsort 8.0.2 nbformat 5.1.3 numba 0.54.1 numexpr 2.8.0 numpy 1.19.5 packaging 21.3 pandas 1.1.5 parso 0.8.3 pexpect 4.8.0 pickleshare 0.7.5 pkg_resources NA prometheus_client NA prompt_toolkit 3.0.24 psutil 5.8.0 ptyprocess 0.7.0 pure_eval 0.2.1 pvectorc NA pydev_ipython NA pydevconsole NA pydevd 2.6.0 pydevd_concurrency_analyser NA pydevd_file_utils NA pydevd_plugins NA pydevd_tracing NA pygments 2.10.0 pyparsing 3.0.6 pyrsistent NA pytz 2021.3 scanpy 1.8.2 scipy 1.5.3 scprep 1.1.0 seaborn 0.11.2 send2trash NA setuptools 58.0.4 setuptools_scm NA sinfo 0.3.1 six 1.16.0 sklearn 0.24.2 sphinxcontrib NA stack_data 0.1.4 statsmodels 0.13.1 tables 3.6.1 terminado 0.12.1 texttable 1.6.4 tornado 6.1 tqdm 4.62.3 traitlets 5.1.1 typing_extensions NA wcwidth 0.2.5 yaml 6.0 zipp NA zmq 22.3.0 ----- IPython 8.0.0 jupyter_client 6.1.12 jupyter_core 4.9.1 jupyterlab 3.2.8 notebook 6.4.7 ----- Python 3.8.12 | packaged by conda-forge | (default, Oct 12 2021, 21:59:51) [GCC 9.4.0] Linux-5.4.0-1064-gcp-x86_64-with-glibc2.10 16 logical CPU cores ----- Session information updated at 2022-02-10 16:38 ```

ivirshup commented 2 years ago

Thanks for the report. This is essentially a documentation issue, as the ax argument cannot work with groupby in this context. This is a holdout from implementations before seaborn made the distinction between Figure and Axes level plots.

scottgigante-immunai commented 2 years ago

Thanks @ivirshup ! In addition to changing the documentation, it would probably make sense to throw an error if these two arguments are passed together.

scverse / scanpy