fixup! informed horseshoe: rename annotations_varm_key to annotations_mkey

also rename public API methods to contain the axis name

Author: ilia-kats

src/mofaflex/_core/priors/horseshoe.py

@@ -296,7 +296,7 @@ def n_informed_factors(self) -> int:
         return self._n_informed_factors
 
     @Prior._api(has_factors=False)
-    def get_significant_annotations(self) -> Mapping[str, pd.DataFrame]:
+    def get_significant_a̲x̲i̲s̲_annotations(self) -> Mapping[str, pd.DataFrame]:
         """Get the results of significance testing of annotations against factors.
 
         The significance testing is an implementation of PCGSE :cite:p:`pmid26300978`. While
@@ -310,6 +310,6 @@ def get_significant_annotations(self) -> Mapping[str, pd.DataFrame]:
 
     @Prior._api(has_factors=True, factors_subset="factors_subset")
     @property
-    def annotations(self) -> Mapping[str, pd.DataFrame]:
+    def a̲x̲i̲s̲_annotations(self) -> Mapping[str, pd.DataFrame]:
         """Annotation matrices for each view."""
         return MappingProxyType(self._annotations)

src/mofaflex/pl/_plotting.py

@@ -548,7 +548,7 @@ def factor_significance(
         alpha: False discovery rate threshold.
         figsize: Figure size in inches.
     """
-    pcgse_results = model.get_significant_annotations()
+    pcgse_results = model.get_significant_weight_annotations()
     if pcgse_results is None:
         raise ValueError("PCGSE results not available.")
 
@@ -581,7 +581,7 @@ def factor_significance(
     )
     r2 = model.get_r2("term")
     annotations = pd.concat(
-        {view_name: vannot.sum(axis=0) for view_name, vannot in model.get_annotations().items()}, axis=0
+        {view_name: vannot.sum(axis=0) for view_name, vannot in model.get_weight_annotations().items()}, axis=0
     ).rename_axis(index=("view", "annotation"))
     # Rank factors by the variance they explain within the selected views/groups, rather than overall, so that
     # restricting `views` reorders the plot accordingly.
@@ -1056,7 +1056,7 @@ def _prepare_weights_df(
 ):
     weights = model.get_weights(ordered=False)
     try:
-        annotations = model.get_annotations(ordered=False)
+        annotations = model.get_weight_annotations(ordered=False)
     except AttributeError:
         annotations = {}
     if views is None:

tests/test_mofaflex_integration.py

@@ -208,8 +208,8 @@ def test_integration(
         assert model.n_informed_factors > 0
         assert model.terms["_"].n_informed_factors > 0
         assert model.n_informed_factors == model.terms["_"].n_informed_factors
-        assert "get_annotations" in dir(model)
-        assert all(isinstance(annot, pd.DataFrame) for annot in model.get_annotations().values())
+        assert "get_weight_annotations" in dir(model)
+        assert all(isinstance(annot, pd.DataFrame) for annot in model.get_weight_annotations().values())
     elif argname == "guiding_vars_obs_keys":
         assert model.n_guided_factors == model.terms["_"].n_guided_factors == 3
     else:
@@ -220,9 +220,9 @@ def test_integration(
             == model.terms["_"].n_total_factors
             == 5
         )
-        assert "get_annotations" not in dir(model)
+        assert "get_weight_annotations" not in dir(model)
         with pytest.raises(AttributeError, match="is only available when using the 'InformedHorseshoe' prior"):
-            model.get_annotations()
+            model.get_weight_annotations()
 
     if fitargs.get("save_path") is not False:
         loaded_model = MOFAFLEX.load(path=next(iter(tmp_path.glob("*.h5"))))
@@ -330,14 +330,14 @@ def test_integration_dynamicapi_multiple_priors(anndata_dict, tmp_path, n_partic
         n_particles=n_particles,
         save_path=save_path,
     )
-    signif = model.get_significant_annotations()
+    signif = model.get_significant_weight_annotations()
     assert len(signif) == 1
     assert next(iter(signif.keys())) == "view_normal"
     assert signif["view_normal"]["factor"].cat.categories.size == 11
 
     # Loading without an explicit map_location uses the serialized training device.
     reloaded = MOFAFLEX.load(save_path)
-    assert reloaded.get_significant_annotations().keys() == signif.keys()
+    assert reloaded.get_significant_weight_annotations().keys() == signif.keys()
 
 
 def test_load_uses_training_device_for_gp(anndata_dict, tmp_path, monkeypatch):

tests/test_pcgse.py

@@ -8,7 +8,7 @@
 def test_test_annotation_significance_data_None_corr_True(mousebrain_model):
     with pytest.raises(ValueError):
         mfl.tl.test_annotation_significance(
-            mousebrain_model, mousebrain_model.get_annotations(), data=None, corr_adjust=True
+            mousebrain_model, mousebrain_model.get_weight_annotations(), data=None, corr_adjust=True
         )
 
 
@@ -19,7 +19,7 @@ def test_test_annotation_significance_annotations_empty(mousebrain_model):
 
 def test_test_annotation_significance(mousebrain_model):
     results = mfl.tl.test_annotation_significance(
-        mousebrain_model, mousebrain_model.get_annotations(), data=None, corr_adjust=False
+        mousebrain_model, mousebrain_model.get_weight_annotations(), data=None, corr_adjust=False
     )
 
     assert isinstance(results, dict)
@@ -33,7 +33,7 @@ def test_pcgse_test(mousebrain_model):
     results = pcgse_test(
         data=None,
         nonnegative_weights={"view_1": True},
-        annotations=mousebrain_model.get_annotations(),
+        annotations=mousebrain_model.get_weight_annotations(),
         weights=mousebrain_model.get_weights(),
         corr_adjust=False,
     )
@@ -48,7 +48,7 @@ def test_test_single_view_nonnegative(mousebrain_model):
         _test_single_view(
             "view_1",
             nonnegative_weights=True,
-            feature_sets=mousebrain_model.get_annotations()["view_1"],
+            feature_sets=mousebrain_model.get_weight_annotations()["view_1"],
             factor_loadings=mousebrain_model.get_weights()["view_1"],
             y=None,
             sign="neg",
@@ -59,7 +59,7 @@ def test_test_single_view_nonnegative(mousebrain_model):
 
 
 def test_test_single_view_empty(mousebrain_model):
-    feature_sets_empty = mousebrain_model.get_annotations()["view_1"] & False
+    feature_sets_empty = mousebrain_model.get_weight_annotations()["view_1"] & False
 
     assert (
         _test_single_view(
@@ -80,7 +80,7 @@ def test_test_single_view_repeated_factor_names(mousebrain_model):
     result = _test_single_view(
         "view_1",
         nonnegative_weights=True,
-        feature_sets=mousebrain_model.get_annotations()["view_1"],
+        feature_sets=mousebrain_model.get_weight_annotations()["view_1"],
         factor_loadings=pd.concat((factors, factors), axis=1),
         y=None,
         sign="pos",
@@ -95,7 +95,7 @@ def test_test_single_view(mousebrain_model, sign):
     results = _test_single_view(
         "view_1",
         nonnegative_weights=False,
-        feature_sets=mousebrain_model.get_annotations()["view_1"],
+        feature_sets=mousebrain_model.get_weight_annotations()["view_1"],
         factor_loadings=mousebrain_model.get_weights()["view_1"],
         y=None,
         sign=sign,
@@ -112,7 +112,7 @@ def test_single_guided_factor(mousebrain_model):
     results = _test_single_view(
         "view_1",
         nonnegative_weights=False,
-        feature_sets=mousebrain_model.get_annotations()["view_1"].iloc[:, :1],
+        feature_sets=mousebrain_model.get_weight_annotations()["view_1"].iloc[:, :1],
         factor_loadings=mousebrain_model.get_weights()["view_1"],
         y=None,
         sign="all",