Fix a few pep8 warnings

dwi_ml/data/processing/streamlines/post_processing.py

@@ -82,7 +82,8 @@ def _get_all_n_previous_dirs(streamlines_dirs: List[torch.Tensor],
         for n in range(nb_previous_dirs):
             # The n^e previous dir is just the list of dirs, shifted right
             if n > 0:
-                no_n_prev_dirs = torch.cat((unexisting_val, no_n_prev_dirs[:-1,:]))
+                no_n_prev_dirs = torch.cat((unexisting_val,
+                                            no_n_prev_dirs[:-1, :]))
             else:
                 no_n_prev_dirs = torch.cat((unexisting_val, no_n_prev_dirs))
 
@@ -309,7 +310,16 @@ def compute_triu_connectivity_from_labels(streamlines, data_labels,
     data_labels: np.ndarray
         The loaded nifti image.
     binary: bool
-        If true, return a binary matrix.
+        If True, return a binary matrix.
+    use_scilpy: bool
+        If True, uses scilpy's method:
+          'Strategy is to keep the longest streamline segment
+           connecting 2 regions. If the streamline crosses other gray
+           matter regions before reaching its final connected region,
+           the kept connection is still the longest. This is robust to
+           compressed streamlines.'
+        Else, uses simple computation from endpoints. Faster. Also, works with
+        incomplete parcellation.
     """
     real_labels = np.unique(data_labels)[1:]
     nb_labels = len(real_labels)

dwi_ml/training/utils/batch_samplers.py

@@ -54,6 +54,4 @@ def prepare_batch_sampler(dataset, args, sub_loggers_level):
             cycles=args.cycles,
             rng=args.rng, log_level=sub_loggers_level)
 
-
-
     return batch_sampler

dwi_ml/unit_tests/test_submethods_connectivity_matrix.py

@@ -24,4 +24,4 @@ def test_connectivity():
 
 
 if __name__ == '__main__':
-    test_connectivity()
+    test_connectivity()

dwi_ml/unit_tests/test_submethods_interpolation.py

@@ -10,7 +10,6 @@
     interpolate_volume_in_neighborhood
 
 
-
 nb_points = 2
 nb_features = 2
 fake_data = torch.zeros(15, 15, 15, nb_features)

dwi_ml/unit_tests/test_submethods_losses.py

@@ -99,6 +99,7 @@ def format_model_outputs(one_fake_model_output):
 
     return fake_model_outputs
 
+
 def _verify_loss(streamline: Union[torch.Tensor, list],
                  fake_model_outputs: Union[list, Tuple[list]],
                  expected_loss: Union[torch.Tensor, np.ndarray, float],

dwi_ml/unit_tests/test_train_batch_sampler.py

@@ -7,7 +7,8 @@
 from dwi_ml.data.dataset.multi_subject_containers import MultiSubjectDataset
 from dwi_ml.unit_tests.utils.expected_values import (
     TEST_EXPECTED_SUBJ_NAMES, TEST_EXPECTED_NB_STREAMLINES)
-from dwi_ml.unit_tests.utils.data_and_models_for_tests import create_test_batch_sampler, fetch_testing_data
+from dwi_ml.unit_tests.utils.data_and_models_for_tests import (
+    create_test_batch_sampler, fetch_testing_data)
 
 
 def test_batch_sampler():

dwi_ml/unit_tests/visual_tests/analyze_batch_loader_visually.py

@@ -58,9 +58,8 @@ def save_loaded_batch_for_visual_assessment(dataset, ref):
     logging.info("*********")
     logging.info("Neighborhood:")
     # Using ModelForTest with both parents OneInput and Neighborhood.
-    model = ModelForTest(experiment_name='test',
-                         neighborhood_type='axes',
-                         neighborhood_radius=[1, 2])
+    model = ModelForTest(experiment_name='test', neighborhood_type='axes',
+                         neighborhood_radius=2)
     batch_loader = create_batch_loader(dataset, model)
     batch_loader.set_context('training')
     _load_directly_and_verify(batch_loader, batch_idx_tuples, ref, 'neighb')

dwi_ml/version.py

@@ -85,4 +85,4 @@
 PYTHON_SCRIPTS = glob.glob("scripts_python/*.py")
 BASH_SCRIPTS = glob.glob("bash_utilities/*.sh")
 
-PREVIOUS_MAINTAINERS=[]
+PREVIOUS_MAINTAINERS = []

scripts_python/dwiml_compute_connectivity_matrix_from_labels.py

@@ -43,7 +43,7 @@ def _build_arg_parser():
              "   connecting 2 regions. If the streamline crosses other gray \n"
              "   matter regions before reaching its final connected region, \n"
              "   the kept connection is still the longest. This is robust to \n"
-             "   compressed streamlines.\n"
+             "   compressed streamlines.'\n"
              "Else, uses simple computation from endpoints. Faster. Also, "
              "works with incomplete parcellation.")