replace np.float/np.int

README.rst

@@ -10,6 +10,8 @@ optimize/test more on multi-gpu support in call_mods module
 
 add scipy==1.7.3 in requirements (floating point error in scipy>=1.9->tombo). fix this in deepsignal-plant, although tombo only is the pre-process tool for this tool.
 
+replace np.float/np.int with float/int in extract_features module
+
 
 0.1.6
 -----

deepsignal_plant/extract_features.py

@@ -82,7 +82,7 @@ def _get_label_raw(fast5_fn, correct_group, correct_subgroup):
  fast5_data.close()
  raise KeyError('no read_start_rel_to_raw in event attributes')
 
- lengths = event['length'].astype(np.int)
+ lengths = event['length'].astype(int)
  base = [x.decode("UTF-8") for x in event['base']]
  assert len(starts) == len(lengths)
  assert len(lengths) == len(base)
@@ -178,9 +178,9 @@ def _get_alignment_info_from_fast5(fast5_path, corrected_group='RawGenomeCorrect
 
 def _normalize_signals(signals, normalize_method="mad"):
  if normalize_method == 'zscore':
- sshift, sscale = np.mean(signals), np.float(np.std(signals))
+ sshift, sscale = np.mean(signals), float(np.std(signals))
  elif normalize_method == 'mad':
- sshift, sscale = np.median(signals), np.float(robust.mad(signals))
+ sshift, sscale = np.median(signals), float(robust.mad(signals))
  else:
  raise ValueError("")
  if sscale == 0.0:
@@ -271,7 +271,7 @@ def _get_scaling_of_a_read(fast5fp):
 
 
 def _rescale_signals(rawsignals, scaling, offset):
- return np.array(scaling * (rawsignals + offset), dtype=np.float)
+ return np.array(scaling * (rawsignals + offset), dtype=float)
 
 
 def _extract_features(fast5s, corrected_group, basecall_subgroup, normalize_method,