Merge 4d4cf5e into 54a6c17

ark/segmentation/marker_quantification.py

@@ -23,7 +23,7 @@ def compute_marker_counts(input_images, segmentation_masks, nuclear_counts=False
             marker_counts (xarray): xarray containing segmented data of cells x markers
     """
 
-    unique_cell_num = len(np.unique(segmentation_masks.values).astype('int'))
+    unique_cell_ids = np.unique(segmentation_masks[..., 0].values)
 
     # define morphology properties to be extracted from regionprops
     object_properties = ["label", "area", "eccentricity", "major_axis_length",
@@ -34,12 +34,12 @@ def compute_marker_counts(input_images, segmentation_masks, nuclear_counts=False
                                     object_properties[:-1]), axis=None)
 
     # create np.array to hold compartment x cell x feature info
-    marker_counts_array = np.zeros((len(segmentation_masks.compartments), unique_cell_num,
+    marker_counts_array = np.zeros((len(segmentation_masks.compartments), len(unique_cell_ids),
                                     len(feature_names)))
 
     marker_counts = xr.DataArray(copy.copy(marker_counts_array),
                                  coords=[segmentation_masks.compartments,
-                                         np.unique(segmentation_masks.values).astype('int'),
+                                         unique_cell_ids.astype('int'),
                                          feature_names],
                                  dims=['compartments', 'cell_id', 'features'])
 
@@ -95,10 +95,10 @@ def compute_marker_counts(input_images, segmentation_masks, nuclear_counts=False
                 nuc_features = np.concatenate((nuc_counts, current_nuc_props), axis=None)
 
                 # add counts of each marker to appropriate column
-                marker_counts.loc['nuclear', nuc_id, marker_counts.features[1]:] = nuc_features
+                marker_counts.loc['nuclear', cell_id, marker_counts.features[1]:] = nuc_features
 
                 # add cell size to first column
-                marker_counts.loc['nuclear', nuc_id, marker_counts.features[0]] = \
+                marker_counts.loc['nuclear', cell_id, marker_counts.features[0]] = \
                     nuc_coords.shape[0]
 
     return marker_counts

ark/segmentation/marker_quantification_test.py

@@ -54,7 +54,7 @@ def test_compute_marker_counts():
     assert np.all(segmentation_output.loc['whole_cell', :, 'chan4'][3:] == 0)
 
     # check that cell sizes are correct
-    sizes = [np.sum(cell_mask == cell_id) for cell_id in [1, 2, 3, 4]]
+    sizes = [np.sum(cell_mask == cell_id) for cell_id in [1, 2, 3, 5]]
     assert np.array_equal(sizes, segmentation_output.loc['whole_cell', 1:, 'cell_size'])
 
     # check that regionprops size matches with cell size
@@ -116,7 +116,7 @@ def test_compute_marker_counts():
     assert np.all(segmentation_output_unequal.loc['nuclear', :, 'chan4'][3:] == 0)
 
     # check that cell sizes are correct
-    sizes = [np.sum(nuc_mask == cell_id) for cell_id in [1, 2, 3, 4]]
+    sizes = [np.sum(nuc_mask == cell_id) for cell_id in [1, 2, 3, 5]]
     assert np.array_equal(sizes, segmentation_output_unequal.loc['nuclear', 1:, 'cell_size'])
 
     assert np.array_equal(segmentation_output_unequal.loc['nuclear', 1:, 'cell_size'],
@@ -175,6 +175,9 @@ def test_generate_expression_matrix_multiple_compartments():
     # cell 2 in fov0 has no nucleus
     nuc_masks[0, nuc_masks[0, :, :, 0] == 2, 0] = 0
 
+    # all of the nuclei have a label that is 2x the label of the corresponding cell
+    nuc_masks *= 2
+
     unequal_masks = np.concatenate((cell_masks, nuc_masks), axis=-1)
     coords = [["Point0", "Point1"], range(40), range(40), ['whole_cell', 'nuclear']]
     dims = ['fovs', 'rows', 'cols', 'compartments']
@@ -190,16 +193,26 @@ def test_generate_expression_matrix_multiple_compartments():
         channel_data,
         nuclear_counts=True)
 
+    # 7 total cells
     assert normalized.shape[0] == 7
 
+    # channel 0 has a constant value of 1
     assert np.all(normalized['chan0'] == np.repeat(1, len(normalized)))
+
+    # channel 1 has a constant value of 5
     assert np.all(normalized['chan1'] == np.repeat(5, len(normalized)))
-    assert np.all(normalized['chan2'] == normalized['chan2'])
 
-    # check that missing nucleus has size 0
+    # these two channels should be equal for all cells
+    assert np.all(normalized['chan1'] == normalized['chan2'])
+
+    # check that cell with missing nucleus has size 0
     index = np.logical_and(normalized['label'] == 2, normalized['fov'] == 'Point0')
     assert normalized.loc[index, 'cell_size_nuclear'].values == 0
 
+    # check that correct nuclear label is assigned to all cells
+    normalized_with_nuc = normalized.loc[normalized['label'] != 2, ['label', 'label_nuclear']]
+    assert np.all(normalized_with_nuc['label'] * 2 == normalized_with_nuc['label_nuclear'])
+
 
 def test_compute_complete_expression_matrices():
 

ark/utils/segmentation_utils_test.py

@@ -12,7 +12,7 @@ def _create_test_extraction_data():
     cell_mask[4:10, 4:10] = 1
     cell_mask[15:25, 20:30] = 2
     cell_mask[27:32, 3:28] = 3
-    cell_mask[35:40, 15:22] = 4
+    cell_mask[35:40, 15:22] = 5
 
     # then create channels data
     channel_data = np.zeros((40, 40, 5), dtype="int16")