Many test warnings

Signed-off-by: zethson <[email protected]>

ehrapy/anndata/anndata_ext.py

@@ -303,7 +303,7 @@ def move_to_x(adata: AnnData, to_x: list[str] | str) -> AnnData:
         )
 
     if cols_not_in_x:
-        new_adata = concat([adata, AnnData(adata.obs[cols_not_in_x], dtype="object")], axis=1)
+        new_adata = concat([adata, AnnData(adata.obs[cols_not_in_x])], axis=1)
         new_adata.obs = adata.obs[adata.obs.columns[~adata.obs.columns.isin(cols_not_in_x)]]
         # update uns (copy maybe: could be a costly operation but reduces reference cycles)
         # users might save those as separate AnnData object and this could be unexpected behaviour if we dont copy

ehrapy/core/meta_information.py

@@ -76,7 +76,6 @@ def print_header(*, file=None):  # pragma: no cover
         ("sklearn", "scikit-learn"),
         "statsmodels",
         ("igraph", "python-igraph"),
-        "louvain",
         "leidenalg",
         "pynndescent",
     ]

ehrapy/io/_read.py

@@ -334,7 +334,9 @@ def _do_read_h5ad(file_path: Path | Iterator[str]) -> AnnData:
     Returns:
         An AnnData object.
     """
-    adata = read_h5(file_path)
+    import anndata as ad
+
+    adata = ad.read_h5ad(file_path)
     if "ehrapy_dummy_encoding" in adata.uns.keys():
         # if dummy encoding was needed, the original dtype of X could not be numerical, so cast it to object
         adata.X = adata.X.astype("object")

tests/anndata/test_anndata_ext.py

@@ -119,17 +119,17 @@ def test_move_to_x_move_to_obs(self):
         # case 1:  move some column from obs to X and this col was copied previously from X to obs
         move_to_obs(adata, ["name"], copy_obs=True)
         adata = move_to_x(adata, ["name"])
-        assert {"name"}.issubset(set(adata.var_names))  # check if the copied column is still in X
-        assert adata.X.shape == adata_dim_old  # the shape of X should be the same as previously
-        assert "name" in [item for sublist in adata.uns.values() for item in sublist]  # check if the column in in uns
-        delete_from_obs(adata, ["name"])  # delete the column from obs to restore the original adata state
+        assert {"name"}.issubset(set(adata.var_names))
+        assert adata.X.shape == adata_dim_old
+        assert "name" in [item for sublist in adata.uns.values() for item in sublist]
+        delete_from_obs(adata, ["name"])
 
         # case 2: move some column from obs to X and this col was previously moved inplace from X to obs
         move_to_obs(adata, ["clinic_id"], copy_obs=False)
         adata = move_to_x(adata, ["clinic_id"])
-        assert not {"clinic_id"}.issubset(set(adata.obs.columns))  # check if the copied column was removed from obs
-        assert {"clinic_id"}.issubset(set(adata.var_names))  # check if the copied column is now in X
-        assert adata.X.shape == adata_dim_old  # the shape of X should be the same as previously
+        assert not {"clinic_id"}.issubset(set(adata.obs.columns))
+        assert {"clinic_id"}.issubset(set(adata.var_names))
+        assert adata.X.shape == adata_dim_old
         assert "clinic_id" in [
             item for sublist in adata.uns.values() for item in sublist
         ]  # check if the column in in uns
@@ -142,12 +142,10 @@ def test_move_to_x_move_to_obs(self):
         assert not {"los_days"}.issubset(
             set(adata.obs.columns)
         )  # check if the copied column was removed from obs by delete_from_obs()
-        assert not {"b12_values"}.issubset(set(adata.obs.columns))  # check if the moved column was removed from obs
-        assert {"los_days", "b12_values"}.issubset(set(adata.var_names))  # check if the copied column is now in X
-        assert adata.X.shape == adata_dim_old  # the shape of X should be the same as previously
-        assert {"los_days", "b12_values"}.issubset(
-            {item for sublist in adata.uns.values() for item in sublist}
-        )  # check if the column in in uns
+        assert not {"b12_values"}.issubset(set(adata.obs.columns))
+        assert {"los_days", "b12_values"}.issubset(set(adata.var_names))
+        assert adata.X.shape == adata_dim_old
+        assert {"los_days", "b12_values"}.issubset({item for sublist in adata.uns.values() for item in sublist})
 
     def test_delete_from_obs(self):
         adata = ep.io.read_csv(CUR_DIR / "../io/test_data_io/dataset_move_obs_mix.csv")

tests/io/test_read.py

@@ -3,6 +3,7 @@
 import numpy as np
 import pandas as pd
 import pytest
+from pandas import CategoricalDtype
 
 from ehrapy.io._read import read_csv, read_fhir, read_h5ad
 
@@ -135,7 +136,7 @@ def test_read_csv_with_bools_and_cats_obs_only(self):
         assert set(adata.obs.columns) == {"b12_values", "survival", "name"}
         assert pd.api.types.is_bool_dtype(adata.obs["survival"].dtype)
         assert pd.api.types.is_numeric_dtype(adata.obs["b12_values"].dtype)
-        assert pd.api.types.is_categorical_dtype(adata.obs["name"].dtype)
+        assert isinstance(adata.obs["name"].dtype, CategoricalDtype)
 
     def test_set_default_index(self):
         adata = read_csv(dataset_path=f"{_TEST_PATH}/dataset_index.csv")

tests/preprocessing/test_encode.py

@@ -2,6 +2,7 @@
 
 import pandas as pd
 import pytest
+from pandas import CategoricalDtype
 
 from ehrapy.io._read import read_csv
 from ehrapy.preprocessing._encode import DuplicateColumnEncodingError, _reorder_encodings, encode
@@ -70,7 +71,7 @@ def test_autodetect_encode():
         ]
     )
     assert pd.api.types.is_bool_dtype(encoded_ann_data.obs["survival"].dtype)
-    assert pd.api.types.is_categorical_dtype(encoded_ann_data.obs["clinic_day"].dtype)
+    assert isinstance(encoded_ann_data.obs["clinic_day"].dtype, CategoricalDtype)
 
 
 def test_autodetect_num_only(capfd):
@@ -113,7 +114,7 @@ def test_autodetect_custom_mode():
         for column in ["ehrapycat_survival", "ehrapycat_clinic_day"]
     )
     assert pd.api.types.is_bool_dtype(encoded_ann_data.obs["survival"].dtype)
-    assert pd.api.types.is_categorical_dtype(encoded_ann_data.obs["clinic_day"].dtype)
+    assert isinstance(encoded_ann_data.obs["clinic_day"].dtype, CategoricalDtype)
 
 
 def test_autodetect_encode_again():
@@ -169,7 +170,7 @@ def test_custom_encode():
         ]
     )
     assert pd.api.types.is_bool_dtype(encoded_ann_data.obs["survival"].dtype)
-    assert pd.api.types.is_categorical_dtype(encoded_ann_data.obs["clinic_day"].dtype)
+    assert isinstance(encoded_ann_data.obs["clinic_day"].dtype, CategoricalDtype)
 
 
 def test_custom_encode_again_single_columns_encoding():
@@ -199,7 +200,7 @@ def test_custom_encode_again_single_columns_encoding():
     }
     assert id(encoded_ann_data_again.X) != id(encoded_ann_data_again.layers["original"])
     assert pd.api.types.is_bool_dtype(encoded_ann_data.obs["survival"].dtype)
-    assert pd.api.types.is_categorical_dtype(encoded_ann_data.obs["clinic_day"].dtype)
+    assert isinstance(encoded_ann_data.obs["clinic_day"].dtype, CategoricalDtype)
 
 
 def test_custom_encode_again_multiple_columns_encoding():
@@ -233,7 +234,7 @@ def test_custom_encode_again_multiple_columns_encoding():
     }
     assert id(encoded_ann_data_again.X) != id(encoded_ann_data_again.layers["original"])
     assert pd.api.types.is_bool_dtype(encoded_ann_data.obs["survival"].dtype)
-    assert pd.api.types.is_categorical_dtype(encoded_ann_data.obs["clinic_day"].dtype)
+    assert isinstance(encoded_ann_data.obs["clinic_day"].dtype, CategoricalDtype)
 
 
 def test_update_encoding_scheme_1():

tests/preprocessing/test_quality_control.py

@@ -86,19 +86,19 @@ def test_obs_nan_qc_metrics():
     adata.X[0][4] = np.nan
     adata2 = encode(adata, encodings={"one-hot": ["clinic_day"]})
     obs_metrics = _obs_qc_metrics(adata2)
-    assert obs_metrics.iloc[0][0] == 1
+    assert obs_metrics.iloc[0].iloc[0] == 1
 
 
 def test_var_nan_qc_metrics():
     adata = read_csv(dataset_path=f"{_TEST_PATH_ENCODE}/dataset1.csv")
     adata.X[0][4] = np.nan
     adata2 = encode(adata, encodings={"one-hot": ["clinic_day"]})
     var_metrics = _var_qc_metrics(adata2)
-    assert var_metrics.iloc[0][0] == 1
-    assert var_metrics.iloc[1][0] == 1
-    assert var_metrics.iloc[2][0] == 1
-    assert var_metrics.iloc[3][0] == 1
-    assert var_metrics.iloc[4][0] == 1
+    assert var_metrics.iloc[0].iloc[0] == 1
+    assert var_metrics.iloc[1].iloc[0] == 1
+    assert var_metrics.iloc[2].iloc[0] == 1
+    assert var_metrics.iloc[3].iloc[0] == 1
+    assert var_metrics.iloc[4].iloc[0] == 1
 
 
 def test_calculate_qc_metrics(missing_values_adata):

tests/tools/test_sa.py

@@ -12,8 +12,8 @@ def test_ols(self):
         formula = "tco2_first ~ pco2_first"
         var_names = ["tco2_first", "pco2_first"]
         ols = ep.tl.ols(adata, var_names, formula, missing="drop")
-        s = ols.fit().params[1]
-        i = ols.fit().params[0]
+        s = ols.fit().params.iloc[1]
+        i = ols.fit().params.iloc[0]
         assert isinstance(ols, statsmodels.regression.linear_model.OLS)
         assert 0.18857179158259973 == pytest.approx(s)
         assert 16.210859352601442 == pytest.approx(i)
@@ -24,8 +24,8 @@ def test_glm(self):
         var_names = ["day_28_flg", "age"]
         family = "Binomial"
         glm = ep.tl.glm(adata, var_names, formula, family, missing="drop", as_continuous=["age"])
-        Intercept = glm.fit().params[0]
-        age = glm.fit().params[1]
+        Intercept = glm.fit().params.iloc[0]
+        age = glm.fit().params.iloc[1]
         assert isinstance(glm, statsmodels.genmod.generalized_linear_model.GLM)
         assert 5.778006344870297 == pytest.approx(Intercept)
         assert -0.06523274132877163 == pytest.approx(age)