Merge pull request #51 from KrishnaswamyLab/dev

scprep 0.12.2

scprep/plot/scatter.py

@@ -150,7 +150,11 @@ def discrete(self):
                     return True
                 else:
                     # guess based on number of unique elements
-                    return self.n_c_unique <= 20
+                    if self.n_c_unique > 20:
+                        return False
+                    else:
+                        # are the unique elements integer-like?
+                        return np.allclose(self.c_unique % 1, 0, atol=1e-4)
 
     @property
     def c_discrete(self):
@@ -173,12 +177,13 @@ def c_discrete(self):
 
     @property
     def c(self):
-        if self.constant_c() or self.array_c():
+        if self.constant_c():
             return self._c
-        elif self.discrete:
-            return self.c_discrete[self.plot_idx]
-        else:
+        elif self.array_c() or not self.discrete:
             return self._c[self.plot_idx]
+        else:
+            # discrete c
+            return self.c_discrete[self.plot_idx]
 
     @property
     def labels(self):

scprep/select.py

@@ -48,32 +48,32 @@ def _check_columns_compatible(*data):
     for d in data:
         if not _get_column_length(d) == _get_column_length(data[0]):
             raise ValueError(
-                "Expected all data to have the same number of "
+                "Expected `data` and `extra_data` to have the same number of "
                 "columns. Got {}".format(
                     [_get_column_length(d) for d in data]))
         if isinstance(d, (pd.DataFrame, pd.Series)) and \
                 isinstance(data[0], (pd.DataFrame, pd.Series)):
             if not np.all(_get_columns(data[0]) == _get_columns(d)):
                 raise ValueError(
-                    "Expected all pandas inputs to have the same columns. "
-                    "Fix with "
-                    "`scprep.select.select_cols(extra_data, data.columns)`")
+                    "Expected `data` and `extra_data` pandas inputs to have "
+                    "the same column names. Fix with "
+                    "`scprep.select.select_cols(*extra_data, data.columns)`")
 
 
 def _check_rows_compatible(*data):
     for d in data:
         if not _get_row_length(d) == _get_row_length(data[0]):
             raise ValueError(
-                "Expected all data to have the same number of "
+                "Expected `data` and `extra_data` to have the same number of "
                 "rows. Got {}".format(
                     [d.shape[0] for d in data]))
         if isinstance(d, (pd.DataFrame, pd.Series)) and \
                 isinstance(data[0], (pd.DataFrame, pd.Series)):
             if not np.all(data[0].index == d.index):
                 raise ValueError(
-                    "Expected all pandas inputs to have the same index. "
-                    "Fix with "
-                    "`scprep.select.select_rows(extra_data, data.index)`")
+                    "Expected `data` and `extra_data` pandas inputs to have "
+                    "the same index. Fix with "
+                    "`scprep.select.select_rows(*extra_data, data.index)`")
 
 
 def _convert_dataframe_1d(idx):

scprep/version.py

@@ -1,4 +1,4 @@
 # author: Scott Gigante <[email protected]>
 # (C) 2018 Krishnaswamy Lab GPLv2
 
-__version__ = "0.12.1"
+__version__ = "0.12.2"

setup.py

@@ -17,7 +17,6 @@
     'tables',
     'h5py',
     'rpy2>=3.0',
-    'matplotlib>=3.0.0',
     'coverage',
     'coveralls'
 ]
@@ -32,6 +31,10 @@
 
 if sys.version_info[:2] < (3, 5):
     raise RuntimeError("Python version >=3.5 required.")
+elif sys.version_info[:2] < (3, 6):
+    test_requires += ['matplotlib>=3.0,<3.1']
+else:
+    test_requires += ['matplotlib>=3.0']
 
 version_py = os.path.join(os.path.dirname(
     __file__), 'scprep', 'version.py')

test/test_plot.py

@@ -236,6 +236,9 @@ def test_array_c(self):
                                 c=self.array_c)
         assert params.array_c()
         assert not params.constant_c()
+        np.testing.assert_array_equal(params.x, params._x[params.plot_idx])
+        np.testing.assert_array_equal(params.y, params._y[params.plot_idx])
+        np.testing.assert_array_equal(params.c, params._c[params.plot_idx])
         assert params.discrete is None
         assert params.legend is False
         assert params.vmin is None
@@ -249,6 +252,9 @@ def test_continuous(self):
         params = _ScatterParams(x=self.x, y=self.y, c=self.c)
         assert not params.array_c()
         assert not params.constant_c()
+        np.testing.assert_array_equal(params.x, params._x[params.plot_idx])
+        np.testing.assert_array_equal(params.y, params._y[params.plot_idx])
+        np.testing.assert_array_equal(params.c, params._c[params.plot_idx])
         assert params.discrete is False
         assert params.legend is True
         assert params.cmap_scale == 'linear'
@@ -257,6 +263,9 @@ def test_continuous(self):
                                 c=np.round(self.c % 1, 1))
         assert not params.array_c()
         assert not params.constant_c()
+        np.testing.assert_array_equal(params.x, params._x[params.plot_idx])
+        np.testing.assert_array_equal(params.y, params._y[params.plot_idx])
+        np.testing.assert_array_equal(params.c, params._c[params.plot_idx])
         assert params.discrete is False
         assert params.legend is True
         assert params.labels is None
@@ -268,6 +277,10 @@ def test_discrete_tab10(self):
                                 c=np.where(self.c > 0, '+', '-'))
         assert not params.array_c()
         assert not params.constant_c()
+        np.testing.assert_array_equal(params.x, params._x[params.plot_idx])
+        np.testing.assert_array_equal(params.y, params._y[params.plot_idx])
+        np.testing.assert_array_equal(
+            params.c, params.c_discrete[params.plot_idx])
         assert params.discrete is True
         assert params.legend is True
         assert params.vmin is None
@@ -276,8 +289,8 @@ def test_discrete_tab10(self):
         np.testing.assert_equal(params.cmap.colors, plt.cm.tab10.colors[:2])
 
     def test_discrete_tab20(self):
-        params = _ScatterParams(x=self.x, y=self.y, discrete=True,
-                                c=np.round(self.c % 1, 1))
+        params = _ScatterParams(x=self.x, y=self.y,
+                                c=10 * np.round(self.c % 1, 1))
         assert not params.array_c()
         assert not params.constant_c()
         assert params.discrete is True
@@ -291,6 +304,19 @@ def test_discrete_tab20(self):
             params.cmap.colors,
             plt.cm.tab20.colors[:len(np.unique(np.round(self.c % 1, 1)))])
 
+    def test_continuous_less_than_20(self):
+        params = _ScatterParams(x=self.x, y=self.y,
+                                c=np.round(self.c % 1, 1))
+        assert not params.array_c()
+        assert not params.constant_c()
+        assert params.discrete is False
+        assert params.legend is True
+        assert params.vmin == 0
+        assert params.vmax == 1
+        assert params.cmap_scale == 'linear'
+        assert params.extend == 'neither'
+        assert params.cmap is matplotlib.cm.inferno
+
     def test_continuous_tab20_str(self):
         params = _ScatterParams(x=self.x, y=self.y, discrete=False,
                                 cmap='tab20', c=np.round(self.c % 1, 1))

test/test_select.py

@@ -314,7 +314,7 @@ def test_select_cols_2d_list_index(self):
     def test_select_cols_unequal_columns(self):
         assert_raise_message(
             ValueError,
-            "Expected all data to have the same number of "
+            "Expected `data` and `extra_data` to have the same number of "
             "columns. Got [100, 50]",
             scprep.select.select_cols,
             self.X,
@@ -323,7 +323,7 @@ def test_select_cols_unequal_columns(self):
     def test_select_rows_unequal_rows(self):
         assert_raise_message(
             ValueError,
-            "Expected all data to have the same number of "
+            "Expected `data` and `extra_data` to have the same number of "
             "rows. Got [100, 50]",
             scprep.select.select_rows,
             self.X,
@@ -332,19 +332,19 @@ def test_select_rows_unequal_rows(self):
     def test_select_cols_conflicting_data(self):
         assert_raise_message(
             ValueError,
-            "Expected all pandas inputs to have the same columns. "
-            "Fix with "
-            "`scprep.select.select_cols(extra_data, data.columns)`",
+            "Expected `data` and `extra_data` pandas inputs to have the same "
+            "column names. Fix with "
+            "`scprep.select.select_cols(*extra_data, data.columns)`",
             scprep.select.select_cols,
             self.X,
             scprep.select.subsample(self.X.T, n=self.X.shape[0]).T)
 
     def test_select_rows_conflicting_data(self):
         assert_raise_message(
             ValueError,
-            "Expected all pandas inputs to have the same index. "
-            "Fix with "
-            "`scprep.select.select_rows(extra_data, data.index)`",
+            "Expected `data` and `extra_data` pandas inputs to have the same "
+            "index. Fix with "
+            "`scprep.select.select_rows(*extra_data, data.index)`",
             scprep.select.select_rows,
             self.X,
             scprep.select.subsample(self.X, n=self.X.shape[0]))
@@ -397,8 +397,8 @@ def test_subsample_mismatch_size(self):
         libsize = self.libsize[:25]
         assert_raise_message(
             ValueError,
-            "Expected all data to have the same number of rows. "
-            "Got [100, 25]",
+            "Expected `data` and `extra_data` to have the same number of "
+            "rows. Got [100, 25]",
             scprep.select.subsample, self.X, libsize, n=20)
 
     def test_subsample_n_too_large(self):