STYLE: Prefer importing numpy as np

Prefer importing `numpy` as `np` to be consistent with current common
practice and for the sake of compactness.

bin/wm_append_diffusion_measures_across_subjects.py

@@ -5,7 +5,7 @@
 import glob
 import os
 
-import numpy
+import numpy as np
 import pandas
 
 
@@ -82,8 +82,8 @@ def main():
 
         data.append(stats_data_vec)
 
-    data = numpy.array(data)
-    data = numpy.concatenate([numpy.array(subject_IDs)[:, numpy.newaxis], data], axis = 1)
+    data = np.array(data)
+    data = np.concatenate([np.array(subject_IDs)[:, np.newaxis], data], axis = 1)
 
     df = pandas.DataFrame(data, columns=appended_fields)
 
@@ -111,19 +111,19 @@ def main():
 
     if len(clusters) == 800 and len(tracts) == 73:
         comm_clusters = [3, 8, 33, 40, 46, 52, 56, 57, 62, 68, 69, 73, 86, 91, 109, 110, 114, 142, 144, 145, 146, 159, 163, 250, 251, 252, 257, 262, 263, 268, 271, 305, 311, 314, 322, 330, 334, 338, 342, 350, 363, 371, 375, 403, 410, 437, 440, 448, 456, 465, 468, 475, 484, 485, 488, 519, 522, 525, 543, 545, 549, 557, 576, 577, 582, 587, 591, 597, 601, 614, 620, 623, 627, 633, 645, 653, 658, 663, 670, 677, 683, 702, 770, 781]
-        comm_clusters = numpy.array(comm_clusters)
-        hemi_clusters = numpy.setdiff1d(numpy.arange(800), comm_clusters)
+        comm_clusters = np.array(comm_clusters)
+        hemi_clusters = np.setdiff1d(np.arange(800), comm_clusters)
     elif len(clusters) == 800 and len(tracts) == 74: # CPC separated
         comm_clusters = [3, 8, 33, 40, 46, 52, 56, 57, 62, 68, 69, 73, 86, 91, 109, 110, 114, 142, 144, 146, 163, 250, 251, 252, 257, 262, 263, 268, 271, 305, 311, 314, 322, 330, 334, 338, 342, 350, 363, 371, 375, 403, 410, 437, 440, 448, 456, 465, 468, 475, 484, 485, 488, 519, 522, 525, 543, 545, 549, 576, 577, 582, 587, 591, 597, 601, 614, 620, 623, 627, 633, 645, 653, 658, 663, 670, 683, 702, 781]
-        comm_clusters = numpy.array(comm_clusters)
-        hemi_clusters = numpy.setdiff1d(numpy.arange(800), comm_clusters)
+        comm_clusters = np.array(comm_clusters)
+        hemi_clusters = np.setdiff1d(np.arange(800), comm_clusters)
     else:
         comm_clusters = None
         hemi_clusters = None
 
     locations = ['left_hemisphere', 'right_hemisphere', 'commissural']
     appended_fields = ['subjectkey']
-    clusters = numpy.array(clusters)
+    clusters = np.array(clusters)
     for loc in locations:
         if loc == 'left_hemisphere' or loc == 'right_hemisphere':
             clusters_loc = clusters[hemi_clusters]
@@ -161,16 +161,16 @@ def main():
         stats_data = stats_data[comm_clusters, :]
         c_stats_data_vec = stats_data.flatten()
 
-        stats_data_vec = numpy.concatenate([l_stats_data_vec, r_stats_data_vec, c_stats_data_vec])
+        stats_data_vec = np.concatenate([l_stats_data_vec, r_stats_data_vec, c_stats_data_vec])
 
         if stats_data_vec.shape[0] != len(appended_fields) - 1:
             print("Error: Check if the diffusion measure file has the same rows and columns with other subjects!")
             exit()
 
         data.append(stats_data_vec)
 
-    data = numpy.array(data)
-    data = numpy.concatenate([numpy.array(subject_IDs)[:, numpy.newaxis], data], axis = 1)
+    data = np.array(data)
+    data = np.concatenate([np.array(subject_IDs)[:, np.newaxis], data], axis = 1)
 
     df = pandas.DataFrame(data, columns=appended_fields)
 

bin/wm_assess_cluster_location_by_hemisphere.py

@@ -17,7 +17,7 @@
 # and this can be learned. At that point all data are midsagitally aligned, which this requires.
 # For running this per-subject, the alignment should be performed to handle the tracts near the midline better.
 # That should be added as an option.
-import numpy
+import numpy as np
 import vtk
 
 import whitematteranalysis as wma
@@ -116,7 +116,7 @@ def _read_location(_inpd):
 
             return _flag_location
 
-        mask_location = numpy.zeros(pd.GetNumberOfLines())
+        mask_location = np.zeros(pd.GetNumberOfLines())
 
         inpointdata = inpd.GetPointData()
         flag_location = False
@@ -151,7 +151,7 @@ def _read_location(_inpd):
             print(f"<{os.path.basename(__file__)}> Cluster location file is assigned but the file", clusterLocationFile, "does not exist.")
             exit()
         else:
-            location_data = numpy.loadtxt(open(clusterLocationFile, "rb"),
+            location_data = np.loadtxt(open(clusterLocationFile, "rb"),
                                           dtype={'names': ('Cluster Index', 'Location Label'), 'formats': ('S17', 'S1')},
                                           delimiter="\t", skiprows=1)
 
@@ -289,20 +289,20 @@ def _read_location(_inpd):
             wma.io.write_polydata(pd, fname)
 
         # for sanity check 
-        if len(numpy.where(mask_location ==0)[0]) > 1:
+        if len(np.where(mask_location ==0)[0]) > 1:
             print("Error: Not all fibers in", fname, "is labeled with hemisphere location infromation.")
             exit()
 
         if outdir is not None:
 
             # output separated clusters
-            mask_right = numpy.zeros(pd.GetNumberOfLines())
-            mask_left = numpy.zeros(pd.GetNumberOfLines())
-            mask_commissure = numpy.zeros(pd.GetNumberOfLines())
+            mask_right = np.zeros(pd.GetNumberOfLines())
+            mask_left = np.zeros(pd.GetNumberOfLines())
+            mask_commissure = np.zeros(pd.GetNumberOfLines())
 
-            mask_left[numpy.where(mask_location==1)[0]] = 1
-            mask_right[numpy.where(mask_location==2)[0]] = 1
-            mask_commissure[numpy.where(mask_location==3)[0]] = 1
+            mask_left[np.where(mask_location==1)[0]] = 1
+            mask_right[np.where(mask_location==2)[0]] = 1
+            mask_commissure[np.where(mask_location==3)[0]] = 1
 
             pd_right = wma.filter.mask(pd, mask_right, preserve_point_data=True, preserve_cell_data=True, verbose=False)
             pd_left = wma.filter.mask(pd, mask_left, preserve_point_data=True, preserve_cell_data=True, verbose=False)

bin/wm_average_tract_measures.py

@@ -5,7 +5,7 @@
 import os
 import re
 
-import numpy
+import numpy as np
 import pandas
 
 
@@ -45,7 +45,7 @@ def main():
     fields = []
     for col in stats.columns:
         fields.append(col)
-    fields = numpy.array(fields)
+    fields = np.array(fields)
 
     print(fields)
 
@@ -65,7 +65,7 @@ def main():
         print(fields[indices])
         append_list.append(indices)
 
-    append_list = numpy.array(append_list)
+    append_list = np.array(append_list)
     append_measures = [field.replace(args.tractList[-1], '') for field in fields[indices]]
 
     print("Output measures:", append_measures)
@@ -75,21 +75,21 @@ def main():
 
         if m_name == '.Num_Points':
 
-            avg_stat = numpy.sum(stats.to_numpy()[:, append_list[:, m_idx]], axis=1)
+            avg_stat = np.sum(stats.to_numpy()[:, append_list[:, m_idx]], axis=1)
 
         elif m_name == '.Num_Fibers':
 
-            avg_stat = numpy.sum(stats.to_numpy()[:, append_list[:, m_idx]], axis=1)
+            avg_stat = np.sum(stats.to_numpy()[:, append_list[:, m_idx]], axis=1)
 
         elif m_name == '.Mean_Length': # weighted by NoS
 
             weight = stats.to_numpy()[:, append_list[:, 1]]
             val = stats.to_numpy()[:, append_list[:, m_idx]]
 
-            val_weighted_sum = numpy.sum(val * weight, axis=1)
-            weight_sum = numpy.sum(weight, axis=1)
+            val_weighted_sum = np.sum(val * weight, axis=1)
+            weight_sum = np.sum(weight, axis=1)
 
-            empty_indices = numpy.where(weight_sum == 0)[0]
+            empty_indices = np.where(weight_sum == 0)[0]
             weight_sum[empty_indices] = 1
 
             avg_stat = val_weighted_sum / weight_sum
@@ -100,18 +100,18 @@ def main():
             weight = stats.to_numpy()[:, append_list[:, 0]]
             val = stats.to_numpy()[:, append_list[:, m_idx]]
 
-            val_weighted_sum = numpy.sum(val.astype(numpy.double) * weight.astype(numpy.double), axis=1)
-            weight_sum = numpy.sum(weight, axis=1)
+            val_weighted_sum = np.sum(val.astype(np.double) * weight.astype(np.double), axis=1)
+            weight_sum = np.sum(weight, axis=1)
 
-            empty_indices = numpy.where(weight_sum == 0)[0]
+            empty_indices = np.where(weight_sum == 0)[0]
             weight_sum[empty_indices] = 1
 
             avg_stat = val_weighted_sum / weight_sum
             avg_stat[empty_indices] = -1
 
         avg_stats.append(avg_stat)
 
-    avg_stats = numpy.array(avg_stats)
+    avg_stats = np.array(avg_stats)
     avg_stats = avg_stats.transpose()
 
     column_names = ['subjectkey']

bin/wm_change_nrrd_dir.py

@@ -6,7 +6,7 @@
 import multiprocessing
 import os
 
-import numpy
+import numpy as np
 import vtk
 from joblib import Parallel, delayed