AttributeError: 'tuple' object has no attribute '__module__' when using plot_and_annotate_facets about statannotations

[trevismd]

Hello @xinyuejohn , thank you for the question. Sorry this is yet poorly documented, and part 2 of the tutorial is long overdue...
(There is an example in the test suite:

statannotations/tests/test_integrate_annotator.py

Lines 73 to 87 in 2a5ebe1

	def test_plot_and_annotate_facets(self):
	annotator = Annotator(None, self.simple_pairs)
	g = sns.FacetGrid(self.params_df.pop("data"),
	col=self.params_df.pop("hue"),
	height=10, sharey=False)
	self.params_df.pop("hue_order")
	g.map_dataframe(annotator.plot_and_annotate_facets,
	plot="boxplot",
	plot_params=self.params_df,
	configuration={'test': 'Mann-Whitney',
	'text_format': 'simple'},
	annotation_func='apply_test',
	ax_op_after=[['set_xlabel', ['Group'], None]],
	annotation_params={'num_comparisons': 'auto'}
	)

)

So basically should be using g.map_dataframe instead of g.map (this should really be in the docstring, I'm sorry), and thus provide the arguments directly to that function.

For your snippet, something like this:

annot = Annotator(None, pairs)
plot_and_annotate_facets = annot.plot_and_annotate_facets(
# deducing
plot_params = {
    'x': feature_name,
    'y': value_name
}  
g = sns.FacetGrid(
    data=plot_df,
    col="Cell type"
)
g.map_dataframe(
    annot.plot_and_annotate_facets,
    plot='boxplot', 
    plot_params=plot_params,
    configuration=configuration, 
    annotation_func='set_custom_annotations', 
    annotation_params=annotation_params, 
    **kwargs
)

Please let us know if something is still not working.

[xinyuejohn]

Hello @xinyuejohn , thank you for the question. Sorry this is yet poorly documented, and part 2 of the tutorial is long overdue... (There is an example in the test suite:

statannotations/tests/test_integrate_annotator.py

Lines 73 to 87 in 2a5ebe1

	def test_plot_and_annotate_facets(self):
	annotator = Annotator(None, self.simple_pairs)
	g = sns.FacetGrid(self.params_df.pop("data"),
	col=self.params_df.pop("hue"),
	height=10, sharey=False)
	self.params_df.pop("hue_order")
	g.map_dataframe(annotator.plot_and_annotate_facets,
	plot="boxplot",
	plot_params=self.params_df,
	configuration={'test': 'Mann-Whitney',
	'text_format': 'simple'},
	annotation_func='apply_test',
	ax_op_after=[['set_xlabel', ['Group'], None]],
	annotation_params={'num_comparisons': 'auto'}
	)

)
So basically should be using g.map_dataframe instead of g.map (this should really be in the docstring, I'm sorry), and thus provide the arguments directly to that function.

@trevismd Thank you so much for your reply! Your reply really helped a lot. But I still have some issue about pairs.

For example, in your test suite, you used self.simple_pairs as pairs. But if I want to only plot for some of the pairs ([(("a", "blue"), ("b", "blue"))]) instead of [["a", "b"]], I will get an error: Missing x value(s) "('a', 'blue')", "('b', 'blue')" in x (specified in pairs) in data

Could you please tell me if I can use such pairs when I plot? Thank you so much!

[hmassalha]

hello @trevismd. you have here a great tool that a lot are interested in, however, it is not documented well. I find it a complete waste of time running between the different parts of codes/informations you gave for each comment trying hard to link them together. Would it be possible to provide a working example, from start to end, well documented, for adding annotations to catplot (grouped box plot output)? many thanks

[hmassalha]

after 2h I managed to came over the following...
https://github.com/trevismd/statannotations/issues/36#issuecomment-1866210714
I still see a detailed doc will help with tweaking this amazing tool (e.g. between group stat rather than within group stat from catplot output, and more...)