def load_traces_and_frametimes(self, key):
# -- find number of recording depths
pipe = (fuse.Activity() & key).fetch('pipe')
assert len(
np.unique(pipe)) == 1, 'Selection is from different pipelines'
pipe = dj.create_virtual_module(pipe[0], 'pipeline_' + pipe[0])
k = dict(key)
k.pop('field', None)
ndepth = len(dj.U('z') & (pipe.ScanInfo.Field() & k))
frame_times = (stimulus.Sync()
& key).fetch1('frame_times').squeeze()[::ndepth]
soma = pipe.MaskClassification.Type() & dict(type='soma')
spikes = (dj.U('field', 'channel') * pipe.Activity.Trace() * StaticScan.Unit() \
* pipe.ScanSet.UnitInfo() & soma & key)
traces, ms_delay, trace_keys = spikes.fetch(
'trace',
'ms_delay',
dj.key,
order_by='animal_id, session, scan_idx, unit_id')
delay = np.fromiter(ms_delay / 1000, dtype=np.float)
frame_times = (delay[:, None] + frame_times[None, :])
traces = np.vstack(
[fill_nans(tr.astype(np.float32)).squeeze() for tr in traces])
traces, frame_times = self.adjust_trace_len(traces, frame_times)
return traces, frame_times, trace_keys
def load_eye_traces(self, key):
#r, center = (pupil.FittedPupil.Ellipse() & key).fetch('major_r', 'center', order_by='frame_id ASC')
r, center = (pupil.FittedPupil.Circle() & key).fetch(
'radius', 'center', order_by='frame_id')
detectedFrames = ~np.isnan(r)
xy = np.full((len(r), 2), np.nan)
xy[detectedFrames, :] = np.vstack(center[detectedFrames])
xy = np.vstack(map(partial(fill_nans, preserve_gap=3), xy.T))
if np.any(np.isnan(xy)):
log.info('Keeping some nans in the pupil location trace')
pupil_radius = fill_nans(r.squeeze(), preserve_gap=3)
if np.any(np.isnan(pupil_radius)):
log.info('Keeping some nans in the pupil radius trace')
eye_time = (pupil.Eye() & key).fetch1('eye_time').squeeze()
return pupil_radius, xy, eye_time
