def createNewProject(self, file_name, pen_data):
PenDataSegmentCategory.clearSegmentCache()
self._project_settings = None
self._name = file_name
# Normalize pen sample times so first sample starts at 0.0 sec.
self._original_timebase_offset=pen_data[0]['time']
pen_data['time']-=self._original_timebase_offset
pen_data['time']=pen_data['time']/1000.0
# to flip data vertically by changing y pos values...
# Maybe better to see if pyqtgraph has a built in option
# to flip a plot vertically.
#
# pmax = pen_data['y'].max()
# pmin = pen_data['y'].min()
# pen_data['y'] = pmax-pen_data['y']+pmin
self._pendata = pen_data
self.nonzero_pressure_mask=self._pendata['pressure']>0
# nonzero_regions_ix will be a tuple of (starts, stops, lengths) arrays
self.nonzero_region_ix=contiguous_regions(self.nonzero_pressure_mask)
self._segmentset=PenDataSegmentCategory(name=self.name,project=self)
self._pendata['segment_id']=self._segmentset.id
if self._selectedtimeregion is None:
MarkWriteProject._selectedtimeregion = SelectedTimePeriodItem(project=self)
else:
MarkWriteProject._selectedtimeregion.project = self
MarkWriteProject._selectedtimeregion.setBounds(bounds=(self.pendata['time'][0], self.pendata['time'][-1]))
MarkWriteProject._selectedtimeregion.setRegion([self.pendata['time'][0], self.pendata['time'][0] + 1.0])
def createNewProject(self, file_name, pen_data, condvars=None, stime_var=None, etime_var=None, file_type=None):
PenDataSegmentCategory.clearSegmentCache()
self._project_settings = None
self._name = file_name
self._filetype = file_type
self._expcondvars = condvars
self._stimevar = stime_var
self._etimevar = etime_var
# go through each trial, select only the samples within
# the trial period, add the trials sample array to list of trial
# sample data.
samples_by_trial = []
trial_times = None
if self._expcondvars is not None and self._stimevar is not None and self._etimevar is not None:
trial_times = []
for t in self._expcondvars:
try:
trialstart = float(t[self._stimevar])
trialend = float(t[self._etimevar])
trial_times.append((trialstart,trialend))
trial_samples = pen_data[(pen_data['time'] >= trialstart) & (pen_data['time'] <= trialend)]
samples_by_trial.append(trial_samples)
#print "trial samples:",trial_samples.shape
except:
print("Error getting trial time period:")
import traceback
traceback.print_exc()
if samples_by_trial:
# make pen_data == concat'ed samples_by_trial
pen_data = np.concatenate(samples_by_trial)
# turn trial start, stop time list into np array
trial_times = np.asarray(trial_times)
# Normalize pen sample times so first sample starts at 0.0 sec.
self._original_timebase_offset=pen_data[0]['time']
pen_data['time']-=self._original_timebase_offset
if trial_times is not None:
trial_times-=self._original_timebase_offset
self._trialtimes = trial_times
# Change time stamps to sec.msec format, if needed
if file_type != 'hdf5':
# data from iohub hdf5 file is already in sec.msec format
pen_data['time']=pen_data['time']/1000.0
self._pendata = pen_data
self.nonzero_pressure_mask=self._pendata['pressure']>0
# nonzero_regions_ix will be a tuple of (starts, stops, lengths) arrays
self.nonzero_region_ix=contiguous_regions(self.nonzero_pressure_mask)
self._segmentset=PenDataSegmentCategory(name=self.name,project=self)
self._pendata['segment_id']=self._segmentset.id
if self._selectedtimeregion is None:
MarkWriteProject._selectedtimeregion = SelectedTimePeriodItem(project=self)
else:
MarkWriteProject._selectedtimeregion.project = self
def postprocess(cls,pendata):
# Set initial sample state to HOVERING or PRESSED based on pressure
pendata['state'][pendata['pressure']==0]=SAMPLE_STATES["HOVERING"]
pendata['state'][pendata['pressure']>0]=SAMPLE_STATES["PRESSED"]
# Calculate ISI value for samples 1-N
sample_dts = pendata['time'][1:]-pendata['time'][:-1]
# Handle non monotonic time stamped samples
non_mono_ix = (sample_dts < 0.0).nonzero()[0]
if len(non_mono_ix)>0:
non_mono_ix = non_mono_ix + 1
print ">>>>>>>"
print "WARNING: Non-monotonic Timestamps Detected."
print " Removing %d sample(s) with negative ISIs...."%(len(non_mono_ix))
print " Sample Indices:",non_mono_ix
print "<<<<<<<"
print
# TODO: This code needs updating, it is a quick and dirty fix.
# Simply removes samples with negative ISI. In Guidos files,
# this seems to only occur within the first few samples
# of the file, if at all.
pendata = np.delete(pendata,non_mono_ix)
sample_dts = pendata['time'][1:]-pendata['time'][:-1]
#
# Detect Series Boundaries Using an ISI Threshold alg.
#
# Determine ISI threshold value to use
if SETTINGS['series_detect_max_isi_msec'] > 0:
series_isi_thresh = SETTINGS['series_detect_max_isi_msec']/1000.0
else:
try:
series_isi_thresh = np.percentile(sample_dts,99.0,interpolation='nearest')*2.5
except TypeError:
series_isi_thresh = np.percentile(sample_dts,99.0)*2.5
print "series_isi_thresh:",series_isi_thresh
print
# Find sample array ix where ISI threshold is exceeded.
# These are the series start ix
series_start_ixs = (sample_dts > series_isi_thresh).nonzero()[0]
if len(series_start_ixs)>0:
series_start_ixs=series_start_ixs+1
# First sample in data file is always a series start.
series_start_ixs = np.insert(series_start_ixs, 0, 0)
# Create list of series_start, series_end indexs for slicing
series_bounds = [] # holds (series_start_ix, series_end_ix,
# series_len, stime, etime, series_dur, series_dur/series_len)
for i, start_ix in enumerate(series_start_ixs[:-1]):
end_ix = series_start_ixs[i+1]
stime, etime = pendata['time'][[start_ix,end_ix-1]]
sdur = etime - stime
slen = end_ix-start_ix
series_bounds.append((start_ix,end_ix,
slen,stime, etime,sdur,sdur/slen))
start_ix, end_ix = series_start_ixs[-1], len(pendata)
stime, etime = pendata['time'][[start_ix,end_ix-1]]
sdur = etime - stime
slen = end_ix-start_ix
series_bounds.append((start_ix,end_ix,
slen,stime, etime,sdur,sdur/(slen-1)))
# For each Sample Series in the data file,
# 1) handle duplicate sample timestamps
# 2) set FIRST_PRESS and FIRST_HOVER sample status as needed.
for i, sb in enumerate(series_bounds):
# Cleanup duplicate timestamps after series bounds have been found
# TODO: This code needs updating, it is a quick and dirty fix.
# A constant ISI across all series should really be used.
dup_ix = (sample_dts[sb[0]:sb[1]] == 0.0).nonzero()[0]# + sb[0]
if len(dup_ix)>0:
start_ix, end_ix, slen, stime, etime, sdur, sampling_interval = sb
calc_sample_times, calc_isi = np.linspace(etime-slen*sampling_interval,etime,slen,retstep=True)
print ">>>>>>>"
print "Warning: Duplicate timestamps found in series", i+1, (stime, etime)
print " Sample / Duplicate Count:",slen,'/',len(dup_ix)
print " Calc. Series ISI:",calc_isi
print " Updating Series sample times......."
print "<<<<<<<"
print
pendata['time'][start_ix:end_ix] = calc_sample_times[:]
# Update sample state field to include FIRST_PRESS as needed
run_start_ixs, _, _ = contiguous_regions(pendata['state'][sb[0]:sb[1]] == SAMPLE_STATES["PRESSED"])
if len(run_start_ixs):
run_start_ixs = np.array(run_start_ixs, dtype=np.uint32) + sb[0]
pendata['state'][run_start_ixs]+= SAMPLE_STATES['FIRST_PRESS']
# Update sample state field to include FIRST_HOVER as needed
hover_start_ixs, _, _ = contiguous_regions(pendata['state'][sb[0]:sb[1]] == SAMPLE_STATES["HOVERING"])
if len(hover_start_ixs):
hover_start_ixs = np.array(hover_start_ixs, dtype=np.uint32) + sb[0]
pendata['state'][hover_start_ixs]+= SAMPLE_STATES['FIRST_HOVER']
# For whole sample array, update series start sample's status
# with FIRST_ENTER value.
if len(series_start_ixs):
pendata['state'][series_start_ixs] += SAMPLE_STATES['FIRST_ENTER']
# Return pendata array with sample state field populated.
return pendata
def postprocess(cls, pendata):
# Set initial sample state to HOVERING or PRESSED based on pressure
pendata['state'][pendata['pressure'] == 0] = SAMPLE_STATES["HOVERING"]
pendata['state'][pendata['pressure'] > 0] = SAMPLE_STATES["PRESSED"]
# Calculate ISI value for samples 1-N
sample_dts = pendata['time'][1:] - pendata['time'][:-1]
# Handle non monotonic time stamped samples
non_mono_ix = (sample_dts < 0.0).nonzero()[0]
if len(non_mono_ix) > 0:
non_mono_ix = non_mono_ix + 1
print ">>>>>>>"
print "WARNING: Non-monotonic Timestamps Detected."
print " Removing %d sample(s) with negative ISIs...." % (
len(non_mono_ix))
print " Sample Indices:", non_mono_ix
print "<<<<<<<"
print
# TODO: This code needs updating, it is a quick and dirty fix.
# Simply removes samples with negative ISI. In Guidos files,
# this seems to only occur within the first few samples
# of the file, if at all.
pendata = np.delete(pendata, non_mono_ix)
sample_dts = pendata['time'][1:] - pendata['time'][:-1]
#
# Detect Series Boundaries Using an ISI Threshold alg.
#
# Determine ISI threshold value to use
if SETTINGS['series_detect_max_isi_msec'] > 0:
series_isi_thresh = SETTINGS['series_detect_max_isi_msec'] / 1000.0
else:
try:
series_isi_thresh = np.percentile(
sample_dts, 99.0, interpolation='nearest') * 2.5
except TypeError:
series_isi_thresh = np.percentile(sample_dts, 99.0) * 2.5
print "series_isi_thresh:", series_isi_thresh
print
# Find sample array ix where ISI threshold is exceeded.
# These are the series start ix
series_start_ixs = (sample_dts > series_isi_thresh).nonzero()[0]
if len(series_start_ixs) > 0:
series_start_ixs = series_start_ixs + 1
# First sample in data file is always a series start.
series_start_ixs = np.insert(series_start_ixs, 0, 0)
# Create list of series_start, series_end indexs for slicing
series_bounds = [] # holds (series_start_ix, series_end_ix,
# series_len, stime, etime, series_dur, series_dur/series_len)
for i, start_ix in enumerate(series_start_ixs[:-1]):
end_ix = series_start_ixs[i + 1]
stime, etime = pendata['time'][[start_ix, end_ix - 1]]
sdur = etime - stime
slen = end_ix - start_ix
series_bounds.append(
(start_ix, end_ix, slen, stime, etime, sdur, sdur / slen))
start_ix, end_ix = series_start_ixs[-1], len(pendata)
stime, etime = pendata['time'][[start_ix, end_ix - 1]]
sdur = etime - stime
slen = end_ix - start_ix
series_bounds.append(
(start_ix, end_ix, slen, stime, etime, sdur, sdur / (slen - 1)))
# For each Sample Series in the data file,
# 1) handle duplicate sample timestamps
# 2) set FIRST_PRESS and FIRST_HOVER sample status as needed.
for i, sb in enumerate(series_bounds):
# Cleanup duplicate timestamps after series bounds have been found
# TODO: This code needs updating, it is a quick and dirty fix.
# A constant ISI across all series should really be used.
dup_ix = (sample_dts[sb[0]:sb[1]] == 0.0).nonzero()[0] # + sb[0]
if len(dup_ix) > 0:
start_ix, end_ix, slen, stime, etime, sdur, sampling_interval = sb
calc_sample_times, calc_isi = np.linspace(
etime - slen * sampling_interval,
etime,
slen,
retstep=True)
print ">>>>>>>"
print "Warning: Duplicate timestamps found in series", i + 1, (
stime, etime)
print " Sample / Duplicate Count:", slen, '/', len(
dup_ix)
print " Calc. Series ISI:", calc_isi
print " Updating Series sample times......."
print "<<<<<<<"
print
pendata['time'][start_ix:end_ix] = calc_sample_times[:]
# Update sample state field to include FIRST_PRESS as needed
run_start_ixs, _, _ = contiguous_regions(
pendata['state'][sb[0]:sb[1]] == SAMPLE_STATES["PRESSED"])
if len(run_start_ixs):
run_start_ixs = np.array(run_start_ixs,
dtype=np.uint32) + sb[0]
pendata['state'][run_start_ixs] += SAMPLE_STATES['FIRST_PRESS']
# Update sample state field to include FIRST_HOVER as needed
hover_start_ixs, _, _ = contiguous_regions(
pendata['state'][sb[0]:sb[1]] == SAMPLE_STATES["HOVERING"])
if len(hover_start_ixs):
hover_start_ixs = np.array(hover_start_ixs,
dtype=np.uint32) + sb[0]
pendata['state'][hover_start_ixs] += SAMPLE_STATES[
'FIRST_HOVER']
# For whole sample array, update series start sample's status
# with FIRST_ENTER value.
if len(series_start_ixs):
pendata['state'][series_start_ixs] += SAMPLE_STATES['FIRST_ENTER']
# Return pendata array with sample state field populated.
return pendata
