Code review comments part 1 (no directory structure fix yet)

ark/phenotyping/som_utils.py

@@ -30,6 +30,10 @@ def compute_cluster_avg(fovs, channels, base_dir, cluster_col,
             Name of the column to group by
         cluster_dir (str):
             Name of the file containing the pixel data with cluster labels
+
+    Returns:
+        pandas.DataFrame:
+            Contains the average channel values for each SOM cluster
     """
 
     # define the cluster averages DataFrame
@@ -64,7 +68,7 @@ def compute_cluster_avg(fovs, channels, base_dir, cluster_col,
 
 
 def create_fov_pixel_data(fov, channels, img_data, seg_labels,
-                          blur_factor=2, subset_proportion=0.1, seed=None):
+                          blur_factor=2, subset_proportion=0.1, seed=42):
     """Preprocess pixel data for one fov
 
     Saves preprocessed data to pre_dir and subsetted data to sub_dir
@@ -87,10 +91,10 @@ def create_fov_pixel_data(fov, channels, img_data, seg_labels,
 
     Returns:
         tuple:
-            A tuple containing two pd.Dataframes:
+            Contains the following:
 
-            - The full preprocessed pixel dataset for a fov
-            - The subsetted pixel dataset for a fov
+            - pandas.DataFrame: Gaussian blurred and channel sum normalized pixel data for a fov
+            - pandas.DataFrame: subset of the preprocessed pixel dataset for a fov
     """
 
     # for each marker, compute the Gaussian blur
@@ -129,10 +133,10 @@ def create_fov_pixel_data(fov, channels, img_data, seg_labels,
 def create_pixel_matrix(fovs, base_dir, tiff_dir, seg_dir,
                         pre_dir='pixel_mat_preprocessed',
                         sub_dir='pixel_mat_subsetted', is_mibitiff=False,
-                        blur_factor=2, subset_proportion=0.1, seed=None):
+                        blur_factor=2, subset_proportion=0.1, seed=42):
     """For each fov, add a Gaussian blur to each channel and normalize channel sums for each pixel
 
-    Saves preprocessed data to pre_dir and subsetted data to sub_dir
+    Saves data to pre_dir and subsetted data to sub_dir
 
     Args:
         fovs (list):
@@ -366,6 +370,8 @@ def consensus_cluster(fovs, channels, base_dir, max_k=20, cap=3,
                       consensus_dir='pixel_mat_consensus', seed=42):
     """Run consensus clustering algorithm on summed data across channels
 
+    Saves data with consensus cluster labels to consensus_dir
+
     Args:
         fovs (list):
             The list of fovs to subset on

ark/utils/notebooks_test.py

@@ -132,7 +132,7 @@ def test_segment_image_data_folder(tb):
 def test_flowsom_cluster_folder(tb):
     with tdir() as base_dir:
         # create input files
-        notebooks_test_utils.flowsom_setup(tb, flowsom_dir=base_dir, seed=42)
+        notebooks_test_utils.flowsom_setup(tb, flowsom_dir=base_dir)
 
         # run the FlowSOM preprocessing and clustering
         notebooks_test_utils.flowsom_run(tb, fovs=['fov0', 'fov1'], channels=['chan0', 'chan1'])

ark/utils/notebooks_test_utils.py

@@ -80,7 +80,7 @@ def segment_notebook_setup(tb, deepcell_tiff_dir, deepcell_input_dir, deepcell_o
 
 
 def flowsom_setup(tb, flowsom_dir, img_shape=(50, 50), num_fovs=3, num_chans=3,
-                  seed=None, is_mibitiff=False, mibitiff_suffix="-MassCorrected-Filtered",
+                  is_mibitiff=False, mibitiff_suffix="-MassCorrected-Filtered",
                   dtype=np.uint16):
     """Creates the directories, data, and MIBItiff settings for testing FlowSOM clustering
 
@@ -95,8 +95,6 @@ def flowsom_setup(tb, flowsom_dir, img_shape=(50, 50), num_fovs=3, num_chans=3,
             The number of test fovs to generate
         num_chans (int):
             The number of test channels to generate
-        seed (int):
-            The random seed to set
         is_mibitiff (bool):
             Whether we're working with mibitiff files or not
         mibitiff_suffix (str):
@@ -151,12 +149,6 @@ def flowsom_setup(tb, flowsom_dir, img_shape=(50, 50), num_fovs=3, num_chans=3,
     """ % (flowsom_dir, tiff_dir, seg_dir, dt.now().strftime("%Y-%m-%dT%H:%M:%S"))
     tb.inject(define_paths, after='file_path')
 
-    # set a random seed
-    if seed is not None:
-        tb.inject("seed = %d" % seed, after='set_seed')
-    else:
-        tb.execute_cell('set_seed')
-
     # will set MIBItiff and MIBItiff_suffix
     tb.execute_cell('mibitiff_set')
     if is_mibitiff:

templates/example_flowsom_clustering.ipynb

@@ -63,20 +63,6 @@
     "consensus_clustered_dir = os.path.join(som_dir, 'pixel_mat_consensus')"
    ]
   },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {
-    "tags": [
-     "set_seed"
-    ]
-   },
-   "outputs": [],
-   "source": [
-    "# set a random seed if desired\n",
-    "seed = None"
-   ]
-  },
   {
    "cell_type": "code",
    "execution_count": 3,
@@ -131,7 +117,7 @@
    "source": [
     "# run FlowSOM preprocessing\n",
     "som_utils.create_pixel_matrix(\n",
-    "    fovs, base_dir, tiff_dir, segmentation_dir, pre_dir=preprocessed_dir, sub_dir=subsetted_dir, seed=seed\n",
+    "    fovs, base_dir, tiff_dir, segmentation_dir, pre_dir=preprocessed_dir, sub_dir=subsetted_dir,\n",
     ")"
    ]
   },
@@ -169,7 +155,7 @@
    "outputs": [],
    "source": [
     "# run the SOM clustering\n",
-    "som_utils.train_som(fovs, channels, base_dir, sub_dir=subsetted_dir, seed=seed)"
+    "som_utils.train_som(fovs, channels, base_dir, sub_dir=subsetted_dir)"
    ]
   },
   {
@@ -235,7 +221,7 @@
    "source": [
     "# run hierarchical clustering based on SOM cluster assignments\n",
     "som_utils.consensus_cluster(\n",
-    "    fovs, channels, base_dir, cluster_dir=som_clustered_dir, consensus_dir=consensus_clustered_dir, seed=seed\n",
+    "    fovs, channels, base_dir, cluster_dir=som_clustered_dir, consensus_dir=consensus_clustered_dir\n",
     ")"
    ]
   },