Ensure cluster_helpers.PixelSOMCluster receives the fovs argument

ark/phenotyping/cell_cluster_utils.py

@@ -483,7 +483,7 @@ def train_cell_som(fovs, channels, base_dir, pixel_data_dir, cell_table_path,
 
     # define the cell SOM cluster object
     cell_pysom = cluster_helpers.CellSOMCluster(
-        cluster_counts_size_norm_path, som_weights_path, cluster_count_cols,
+        cluster_counts_size_norm_path, som_weights_path, fovs, cluster_count_cols,
         num_passes=num_passes, xdim=xdim, ydim=ydim, lr_start=lr_start, lr_end=lr_end
     )
 

ark/phenotyping/cell_cluster_utils_test.py

@@ -633,7 +633,7 @@ def test_cluster_cells(pixel_cluster_prefix):
         # error test: no weights assigned to cell pysom object
         with pytest.raises(ValueError):
             cell_pysom_bad = cluster_helpers.CellSOMCluster(
-                cluster_counts_size_norm_path, 'bad_path.feather', cluster_cols
+                cluster_counts_size_norm_path, 'bad_path.feather', [-1], cluster_cols
             )
 
             cell_cluster_utils.cluster_cells(base_dir=temp_dir, cell_pysom=cell_pysom_bad)
@@ -648,7 +648,7 @@ def test_cluster_cells(pixel_cluster_prefix):
             feather.write_dataframe(weights, weights_path)
 
             cell_pysom_bad = cluster_helpers.CellSOMCluster(
-                cluster_counts_size_norm_path, weights_path, cluster_cols
+                cluster_counts_size_norm_path, weights_path, [-1], cluster_cols
             )
 
             cell_cluster_utils.cluster_cells(base_dir=temp_dir, cell_pysom=cell_pysom_bad)
@@ -662,7 +662,7 @@ def test_cluster_cells(pixel_cluster_prefix):
 
         # define a CellSOMCluster object
         cell_pysom = cluster_helpers.CellSOMCluster(
-            cluster_counts_size_norm_path, cell_som_weights_path, cluster_cols
+            cluster_counts_size_norm_path, cell_som_weights_path, [-1], cluster_cols
         )
 
         # error test: bad cluster_col provided

ark/phenotyping/cluster_helpers.py

@@ -116,7 +116,7 @@ def generate_som_clusters(self, external_data: pd.DataFrame) -> np.ndarray:
 
 class PixelSOMCluster(PixieSOMCluster):
     def __init__(self, pixel_subset_folder: pathlib.Path, norm_vals_path: pathlib.Path,
-                 weights_path: pathlib.Path, columns: List[str],
+                 weights_path: pathlib.Path, fovs: List[str], columns: List[str],
                  num_passes: int = 1, xdim: int = 10, ydim: int = 10,
                  lr_start: float = 0.05, lr_end: float = 0.01):
         """Creates a pixel SOM cluster object derived from the abstract PixieSOMCluster
@@ -128,6 +128,8 @@ def __init__(self, pixel_subset_folder: pathlib.Path, norm_vals_path: pathlib.Pa
                 The name of the feather file containing the normalization values.
             weights_path (pathlib.Path):
                 The path to save the weights to.
+            fovs (List[str]):
+                The list of FOVs to subset the data on.
             columns (List[str]):
                 The list of columns to subset the data on.
             num_passes (int):
@@ -151,10 +153,14 @@ def __init__(self, pixel_subset_folder: pathlib.Path, norm_vals_path: pathlib.Pa
         # load the normalization values in
         self.norm_data = feather.read_dataframe(norm_vals_path)
 
+        # define the fovs used
+        self.fovs = fovs
+
         # list all the files in pixel_subset_folder and load them to train_data
         fov_files = list_files(pixel_subset_folder, substrs='.feather')
         self.train_data = pd.concat(
-            [feather.read_dataframe(os.path.join(pixel_subset_folder, fov)) for fov in fov_files]
+            [feather.read_dataframe(os.path.join(pixel_subset_folder, fov)) for fov in fov_files
+             if os.path.splitext(fov)[0] in fovs]
         )
 
         # we can just normalize train_data now since that's what we'll be training on
@@ -220,15 +226,17 @@ def assign_som_clusters(self, external_data: pd.DataFrame) -> pd.DataFrame:
 
 class CellSOMCluster(PixieSOMCluster):
     def __init__(self, cell_data_path: pathlib.Path, weights_path: pathlib.Path,
-                 columns: List[str], num_passes: int = 1, xdim: int = 10, ydim: int = 10,
-                 lr_start: float = 0.05, lr_end: float = 0.01):
+                 fovs: List[str], columns: List[str], num_passes: int = 1,
+                 xdim: int = 10, ydim: int = 10, lr_start: float = 0.05, lr_end: float = 0.01):
         """Creates a cell SOM cluster object derived from the abstract PixieSOMCluster
 
         Args:
             cell_data_path (pathlib.Path):
                 The name of the cell dataset to use for training
             weights_path (pathlib.Path):
                 The path to save the weights to.
+            fovs (List[str]):
+                The list of FOVs to subset the data on.
             columns (List[str]):
                 The list of columns to subset the data on.
             num_passes (int):
@@ -253,6 +261,12 @@ def __init__(self, cell_data_path: pathlib.Path, weights_path: pathlib.Path,
         # load the cell data in
         self.cell_data = feather.read_dataframe(cell_data_path)
 
+        # define the fovs used
+        self.fovs = fovs
+
+        # subset cell_data on just the FOVs specified
+        self.cell_data = self.cell_data[self.cell_data['fov'].isin(self.fovs)]
+
         # since cell_data is the only dataset, we can just normalize it immediately
         self.normalize_data()
 

ark/phenotyping/cluster_helpers_test.py

@@ -99,14 +99,14 @@ def pixel_pyflowsom_object(pixel_som_base_dir) -> Iterator[
     # define a PixelSOMCluster object with weights
     pixel_som_with_weights = PixelSOMCluster(
         pixel_subset_folder=pixel_sub_path, norm_vals_path=norm_vals_path,
-        weights_path=weights_path, columns=channels, xdim=20, ydim=10
+        weights_path=weights_path, fovs=fovs, columns=channels, xdim=20, ydim=10
     )
 
     # define a PixelSOMCluster object without weights
     pixel_som_sans_weights = PixelSOMCluster(
         pixel_subset_folder=pixel_sub_path, norm_vals_path=norm_vals_path,
         weights_path=pixel_som_base_dir / 'weights_new.feather',
-        columns=channels, xdim=20, ydim=10
+        fovs=fovs[:2], columns=channels, xdim=20, ydim=10
     )
 
     yield pixel_som_with_weights, pixel_som_sans_weights
@@ -154,13 +154,13 @@ def cell_pyflowsom_object(cell_som_base_dir) -> Iterator[
     # define a CellSOMCluster object with weights
     cell_som_with_weights = CellSOMCluster(
         cell_data_path=cell_data_path, weights_path=weights_path,
-        columns=count_cols, xdim=20, ydim=10
+        fovs=['fov0', 'fov1'], columns=count_cols, xdim=20, ydim=10
     )
 
     # define a CellSOMCluster object without weights
     cell_som_sans_weights = CellSOMCluster(
         cell_data_path=cell_data_path, weights_path=cell_som_base_dir / 'weights_new.feather',
-        columns=count_cols, xdim=20, ydim=10
+        fovs=['fov0'], columns=count_cols, xdim=20, ydim=10
     )
 
     yield cell_som_with_weights, cell_som_sans_weights

ark/phenotyping/pixel_cluster_utils.py

@@ -929,7 +929,7 @@ def train_pixel_som(fovs, channels, base_dir,
 
     # define the pixel SOM cluster object
     pixel_pysom = cluster_helpers.PixelSOMCluster(
-        subsetted_path, norm_vals_path, som_weights_path, channels,
+        subsetted_path, norm_vals_path, som_weights_path, fovs, channels,
         num_passes=num_passes, xdim=xdim, ydim=ydim, lr_start=lr_start, lr_end=lr_end
     )
 

ark/phenotyping/pixel_cluster_utils_test.py

@@ -1362,7 +1362,7 @@ def test_run_pixel_som_assignment():
         # define example PixelSOMCluster object
         sample_pixel_cc = cluster_helpers.PixelSOMCluster(
             os.path.join(temp_dir, 'pixel_mat_data'), sample_norm_vals_path,
-            sample_som_weights_path, chans
+            sample_som_weights_path, fovs, chans
         )
 
         fov_status = pixel_cluster_utils.run_pixel_som_assignment(
@@ -1468,13 +1468,14 @@ def test_cluster_pixels_base(multiprocess):
         with pytest.raises(ValueError):
             pixel_pysom_bad = cluster_helpers.PixelSOMCluster(
                 os.path.join(temp_dir, 'pixel_mat_data'), norm_vals_path,
-                'bad_path.feather', chan_list
+                'bad_path.feather', fovs, chan_list
             )
             pixel_cluster_utils.cluster_pixels(fovs, chan_list, temp_dir, pixel_pysom_bad)
 
         # create a sample PixelSOMCluster object
         pixel_pysom = cluster_helpers.PixelSOMCluster(
-            os.path.join(temp_dir, 'pixel_mat_data'), norm_vals_path, som_weights_path, chan_list
+            os.path.join(temp_dir, 'pixel_mat_data'), norm_vals_path, som_weights_path,
+            fovs, chan_list
         )
 
         # run SOM cluster assignment
@@ -1505,7 +1506,7 @@ def test_cluster_pixels_corrupt(multiprocess, capsys):
 
         # create a sample PixelSOMCluster object
         pixel_pysom = cluster_helpers.PixelSOMCluster(
-            os.path.join(temp_dir, 'pixel_mat_data'), norm_vals_path, som_weights_path, chans
+            os.path.join(temp_dir, 'pixel_mat_data'), norm_vals_path, som_weights_path, fovs, chans
         )
 
         # corrupt a fov for this test
@@ -1566,13 +1567,13 @@ def test_generate_som_avg_files(capsys):
         # error test: weights not assigned to PixelSOMCluster object
         with pytest.raises(ValueError):
             pixel_pysom_bad = cluster_helpers.PixelSOMCluster(
-                pixel_data_path, norm_vals_path, 'bad_path.feather', chan_list
+                pixel_data_path, norm_vals_path, 'bad_path.feather', fovs, chan_list
             )
             pixel_cluster_utils.generate_som_avg_files(fovs, chan_list, temp_dir, pixel_pysom_bad)
 
         # define an example PixelSOMCluster object
         pixel_pysom = cluster_helpers.PixelSOMCluster(
-            pixel_data_path, norm_vals_path, weights_path, chan_list
+            pixel_data_path, norm_vals_path, weights_path, fovs, chan_list
         )
 
         # test base generation with all subsetted FOVs