Better strategy to compare two color arrays

[OnlyBelter]

When RGB color represented by 3D arrays, we cannot compare them as list vs list.

[codecov]

Codecov Report

Merging #1884 (d275026) into master (4dd8de9) will not change coverage.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##           master    #1884   +/-   ##
=======================================
  Coverage   71.22%   71.22%           
=======================================
  Files          92       92           
  Lines       11181    11181           
=======================================
  Hits         7964     7964           
  Misses       3217     3217           

Impacted Files	Coverage Δ
scanpy/plotting/_utils.py	54.61% <100.00%> (ø)

[ivirshup]

Can you tell us a bit about your use case? In general, we mostly assume colors are a list of hex values, and would suspect other things may break if you arrays here.

[OnlyBelter]

I don't know why sometimes it's broken when I use some specific column to set the parameter of color in function sc.pl.umap. But sometimes it works well. Is it random for using RGB (3D array) or hex values (a string) or at some specific situation it'll change to RGB?

[ivirshup]

There was an issue where values were being set as rgb values unintentionally which should be fixed by #1886.

Can you give an example of how this was causing problems? What would cause you to hit errors here?

[OnlyBelter]

I tried debug by myself and this is what I found at the break point: when the color represented by RGB values, such as [array([0.29803922, 0.44705882, 0.69019608]), array([0.86666667, 0.51764706, 0.32156863]), array([0.33333333, 0.65882353, 0.40784314]), array([0.76862745, 0.30588235, 0.32156863])], it throws error.

It may also cause by the default setting of seaborn. Since you have fixed it in #1886 , this problem should be fixed too.

[OnlyBelter]

Once again, I got this error, even if I didn't import seaborn.

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-10-a5c62c51242c> in <module>
      1 # plt.figure(figsize=(10, 10))
      2 rcParams['figure.figsize'] = 8, 8
----> 3 sc.pl.umap(all_dataset['pdac_pengj_02'], color='cluster', legend_loc='on data', legend_fontsize='small', title='', frameon=False)

f:\tools\miniconda3\envs\deside2\lib\site-packages\scanpy\plotting\_tools\scatterplots.py in umap(adata, **kwargs)
    601     If `show==False` a :class:`~matplotlib.axes.Axes` or a list of it.
    602     """
--> 603     return embedding(adata, 'umap', **kwargs)
    604 
    605 

f:\tools\miniconda3\envs\deside2\lib\site-packages\scanpy\plotting\_tools\scatterplots.py in embedding(adata, basis, color, gene_symbols, use_raw, sort_order, edges, edges_width, edges_color, neighbors_key, arrows, arrows_kwds, groups, components, layer, projection, scale_factor, color_map, cmap, palette, na_color, na_in_legend, size, frameon, legend_fontsize, legend_fontweight, legend_loc, legend_fontoutline, vmax, vmin, add_outline, outline_width, outline_color, ncols, hspace, wspace, title, show, save, ax, return_fig, **kwargs)
    244             groups=groups,
    245         )
--> 246         color_vector, categorical = _color_vector(
    247             adata,
    248             value_to_plot,

f:\tools\miniconda3\envs\deside2\lib\site-packages\scanpy\plotting\_tools\scatterplots.py in _color_vector(adata, values_key, values, palette, na_color)
   1128         return values, False
   1129     else:  # is_categorical_dtype(values)
-> 1130         color_map = _get_palette(adata, values_key, palette=palette)
   1131         color_vector = values.map(color_map).map(to_hex)
   1132 

f:\tools\miniconda3\envs\deside2\lib\site-packages\scanpy\plotting\_tools\scatterplots.py in _get_palette(adata, values_key, palette)
   1103         _utils._set_default_colors_for_categorical_obs(adata, values_key)
   1104     else:
-> 1105         _utils._validate_palette(adata, values_key)
   1106     return dict(zip(values.categories, adata.uns[color_key]))
   1107 

f:\tools\miniconda3\envs\deside2\lib\site-packages\scanpy\plotting\_utils.py in _validate_palette(adata, key)
    355         _palette.append(color)
    356     # Don't modify if nothing changed
--> 357     if _palette is not None and list(_palette) != list(adata.uns[color_key]):
    358         adata.uns[color_key] = _palette
    359 

ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

[OnlyBelter]

It seems that this problem has been corrected in version 1.8.0. So I close it.

[ivirshup]

Thanks for the update! Glad to hear it works now.