def nan_invalid_segments(rec):
"""
Currently a specialized signal for removing incorrect trials from data
collected using baphy during behavior.
TODO: Migrate to nems_db or make a more generic version
"""
# First, select the appropriate subset of data
rec['resp'] = rec['resp'].rasterize()
sig = rec['resp']
# get list of start and stop times (epoch bounds)
epoch_indices = np.vstack(
(ep.epoch_intersection(sig.get_epoch_bounds('HIT_TRIAL'),
sig.get_epoch_bounds('REFERENCE')),
ep.epoch_intersection(sig.get_epoch_bounds('REFERENCE'),
sig.get_epoch_bounds('PASSIVE_EXPERIMENT'))))
# Only takes the first of any conflicts (don't think I actually need this)
epoch_indices = ep.remove_overlap(epoch_indices)
epoch_indices2 = epoch_indices[0:1, :]
for i in range(1, epoch_indices.shape[0]):
if epoch_indices[i, 0] == epoch_indices2[-1, 1]:
epoch_indices2[-1, 1] = epoch_indices[i, 0]
else:
epoch_indices2 = np.concatenate(
(epoch_indices2, epoch_indices[i:(i + 1), :]), axis=0)
# add adjusted signals to the recording
newrec = rec.nan_times(epoch_indices2)
return newrec
def test_remove_overlap():
epochs = np.array([
[0, 10],
[1, 15],
[5, 10],
[11, 32],
[40, 50],
])
expected = np.array([
[0, 10],
[11, 32],
[40, 50],
])
actual = remove_overlap(epochs)
assert np.all(actual == expected)
def remove_invalid_segments(rec):
"""
Currently a specialized function for removing incorrect trials from data
collected using baphy during behavior.
TODO: Migrate to nems_lbhb or make a more generic version
"""
# First, select the appropriate subset of data
rec['resp'] = rec['resp'].rasterize()
if 'stim' in rec.signals.keys():
rec['stim'] = rec['stim'].rasterize()
sig = rec['resp']
# get list of start and stop indices (epoch bounds)
epoch_indices = np.vstack(
(ep.epoch_intersection(sig.get_epoch_indices('REFERENCE'),
sig.get_epoch_indices('HIT_TRIAL')),
ep.epoch_intersection(sig.get_epoch_indices('REFERENCE'),
sig.get_epoch_indices('PASSIVE_EXPERIMENT'))))
# Only takes the first of any conflicts (don't think I actually need this)
epoch_indices = ep.remove_overlap(epoch_indices)
# merge any epochs that are directly adjacent
epoch_indices2 = epoch_indices[0:1]
for i in range(1, epoch_indices.shape[0]):
if epoch_indices[i, 0] == epoch_indices2[-1, 1]:
epoch_indices2[-1, 1] = epoch_indices[i, 1]
else:
#epoch_indices2 = np.concatenate(
# (epoch_indices2, epoch_indices[i:(i + 1), :]), axis=0)
epoch_indices2 = np.append(epoch_indices2,
epoch_indices[i:(i + 1)],
axis=0)
# convert back to times
epoch_times = epoch_indices2 / sig.fs
# add adjusted signals to the recording
newrec = rec.select_times(epoch_times)
return newrec