def generate_rasters(self):
filename = self.filename
# Validate filename
if not self.valid_filename(filename):
return
target_test = self.ui.comboBox_test_num.currentText()
thresh = self.ui.doubleSpinBox_threshold.value()
h_file = h5py.File(unicode(filename), 'r')
# Find target segment
for segment in h_file.keys():
for test in h_file[segment].keys():
if target_test == test:
target_seg = segment
target_test = test
fs = h_file[target_seg].attrs['samplerate_ad']
reps = h_file[target_seg][target_test].attrs['reps']
start_time = h_file[target_seg][target_test].attrs['start']
trace_data = h_file[target_seg][target_test].value
if len(trace_data.shape) == 4:
pass
stim_info = eval(h_file[target_seg][target_test].attrs['stim'])
duration = trace_data.shape[-1] / fs * 1000
if stim_info[1]['components'][0]['stim_type'] != 'Pure Tone':
self.add_message('Cannot generate raster with stim type "' + str(stim_info[1]['components'][0]['stim_type']) + '".')
h_file.close()
return
autoRasters = {}
for tStim in range(1, len(stim_info)):
freq = int(stim_info[tStim]['components'][0]['frequency'])
spl = int(stim_info[tStim]['components'][0]['intensity'])
trace_key = str(freq) + '_' + str(spl)
num_reps = trace_data.shape[1] # ???: Same as reps = h_file[target_seg][target_test].attrs['reps']
spikeTrains = pd.DataFrame([])
nspk = 0
for tRep in range(reps):
if len(trace_data.shape) == 3:
trace = trace_data[tStim][tRep]
pass
elif len(trace_data.shape) == 4:
tchan = int(self.ui.comboBox_channel.currentText().replace('channel_', '')) - 1
trace = trace_data[tStim][tRep][tchan]
pass
else:
self.add_message('Cannot handle trace_data of shape: ' + str(trace_data.shape))
return
spike_times = 1000 * np.array(get_spike_times(trace, thresh, fs))
spike_times_s = pd.Series(spike_times)
if spike_times_s.size > nspk:
spikeTrains = spikeTrains.reindex(spike_times_s.index)
nspk = spike_times_s.size
spikeTrains[str(tRep)] = spike_times_s
autoRasters[trace_key] = spikeTrains
h_file.close()
return autoRasters
def graph_multi_io_test(self):
h_file = h5py.File(unicode(self.filename), 'r')
target_rows = []
for i in range(len(self.checkboxes)):
if self.checkboxes[i].checkState():
target_rows.append(i)
axes = []
for row in target_rows:
for segment in h_file.keys():
for test in h_file[segment].keys():
if self.checkboxes[row].text() == test:
target_seg = segment
target_test = test
fs = h_file[target_seg].attrs['samplerate_ad']
reps = h_file[target_seg][target_test].attrs['reps']
start_time = h_file[target_seg][target_test].attrs['start']
trace_data = h_file[target_seg][target_test].value
stim_info = eval(h_file[target_seg][target_test].attrs['stim'])
autoRasters = {}
for tStim in range(1, len(stim_info)):
spl = int(stim_info[tStim]['components'][0]['intensity'])
traceKey = 'None_' + str(spl).zfill(2)
spikeTrains = pd.DataFrame([])
nspk = 0
for tRep in range(reps):
if len(trace_data.shape) == 3:
trace = trace_data[tStim][tRep]
pass
elif len(trace_data.shape) == 4:
tchan = int(self.comboboxes[row].currentText().replace('channel_', '')) - 1
trace = trace_data[tStim][tRep][tchan]
pass
else:
self.add_message('Cannot handle trace_data of shape: ' + str(trace_data.shape))
return
spike_times = 1000 * np.array(get_spike_times(trace, self.spnboxes[row].value(), fs))
spike_times_s = pd.Series(spike_times)
if spike_times_s.size >= nspk:
spikeTrains = spikeTrains.reindex(spike_times_s.index)
nspk = spike_times_s.size
spikeTrains[str(tRep)] = spike_times_s
autoRasters[traceKey] = spikeTrains
rasters = autoRasters
tuning = []
sortedKeys = sorted(rasters.keys())
for traceKey in sortedKeys:
spl = int(traceKey.split('_')[-1])
raster = rasters[traceKey]
res = ResponseStats(raster)
tuning.append({'intensity': spl, 'response': res[0], 'responseSTD': res[1]})
tuningCurves = pd.DataFrame(tuning)
if axes:
tuningCurves.plot(x='intensity', y='response', ax=axes, yerr='responseSTD', capthick=1, label=str(target_test) + ' : ' + str(self.spnboxes[row].value()) + ' V')
else:
axes = tuningCurves.plot(x='intensity', y='response', yerr='responseSTD', capthick=1, label=str(target_test) + ' : ' + str(self.spnboxes[row].value()) + ' V')
plt.legend(loc='upper left', fontsize=12, frameon=True)
sns.despine()
plt.grid(False)
plt.xlabel('Intensity (dB)', size=14)
plt.ylabel('Response Rate (Hz)', size=14)
plt.tick_params(axis='both', which='major', labelsize=14)
plt.title(str.split(str(self.filename), '/')[-1].replace('.hdf5', '') + ' Multi I/O')
plt.show()
h_file.close()
def graph_io_test(self):
filename = self.filename = self.ui.lineEdit_file_name.text()
# Validate filename
if not self.valid_filename(filename):
return
target_test = self.ui.comboBox_test_num.currentText()
thresh = self.ui.doubleSpinBox_threshold.value()
h_file = h5py.File(unicode(filename), 'r')
# Find target segment
for segment in h_file.keys():
for test in h_file[segment].keys():
if target_test == test:
target_seg = segment
target_test = test
fs = h_file[target_seg].attrs['samplerate_ad']
reps = h_file[target_seg][target_test].attrs['reps']
start_time = h_file[target_seg][target_test].attrs['start']
trace_data = h_file[target_seg][target_test].value
if len(trace_data.shape) == 4:
pass
stim_info = eval(h_file[target_seg][target_test].attrs['stim'])
duration = trace_data.shape[-1] / fs * 1000
autoRasters = {}
for tStim in range(1, len(stim_info)):
spl = int(stim_info[tStim]['components'][0]['intensity'])
traceKey = 'None_' + str(spl).zfill(2)
spikeTrains = pd.DataFrame([])
nspk = 0
for tRep in range(reps):
if len(trace_data.shape) == 3:
trace = trace_data[tStim][tRep]
pass
elif len(trace_data.shape) == 4:
tchan = int(self.ui.comboBox_channel.currentText().replace('channel_', '')) - 1
trace = trace_data[tStim][tRep][tchan]
pass
else:
self.add_message('Cannot handle trace_data of shape: ' + str(trace_data.shape))
return
spike_times = 1000 * np.array(get_spike_times(trace, thresh, fs))
spike_times_s = pd.Series(spike_times)
if spike_times_s.size >= nspk:
spikeTrains = spikeTrains.reindex(spike_times_s.index)
nspk = spike_times_s.size
spikeTrains[str(tRep)] = spike_times_s
autoRasters[traceKey] = spikeTrains
rasters = autoRasters
h_file.close()
tuning = []
sortedKeys = sorted(rasters.keys())
for traceKey in sortedKeys:
spl = int(traceKey.split('_')[-1])
raster = rasters[traceKey]
res = ResponseStats(raster)
tuning.append({'intensity': spl, 'response': res[0], 'responseSTD': res[1]})
tuningCurves = pd.DataFrame(tuning)
tuningCurves.plot(x='intensity', y='response', yerr='responseSTD', capthick=1, label=str(target_test))
plt.legend(loc='upper left', fontsize=12, frameon=True)
sns.despine()
plt.grid(False)
plt.xlabel('Intensity (dB)', size=14)
plt.ylabel('Response Rate (Hz)', size=14)
plt.tick_params(axis='both', which='major', labelsize=14)
plt.title(str.split(str(filename), '/')[-1].replace('.hdf5', '') + ' '
+ str(self.ui.comboBox_test_num.currentText()).replace('test_', 'Test '))
plt.show()
def graph_historgram(self):
filename = self.filename = self.ui.lineEdit_file_name.text()
# Validate filename
if not self.valid_filename(filename):
return
target_test = self.ui.comboBox_test_num.currentText()
thresh = self.ui.doubleSpinBox_threshold.value()
h_file = h5py.File(unicode(filename), 'r')
# Find target segment
for segment in h_file.keys():
for test in h_file[segment].keys():
if target_test == test:
target_seg = segment
target_test = test
target_trace = int(self.ui.comboBox_trace.currentText().replace('trace_', '')) - 1
fs = h_file[target_seg].attrs['samplerate_ad']
reps = h_file[target_seg][target_test].attrs['reps']
start_time = h_file[target_seg][target_test].attrs['start']
trace_data = h_file[target_seg][target_test].value
if len(trace_data.shape) == 4:
pass
stim_info = eval(h_file[target_seg][target_test].attrs['stim'])
duration = trace_data.shape[-1] / fs * 1000
autoRasters = {}
tStim = target_trace
spikeTrains = pd.DataFrame([])
nspk = 0
for tRep in range(reps):
if len(trace_data.shape) == 3:
trace = trace_data[tStim][tRep]
pass
elif len(trace_data.shape) == 4:
tchan = int(self.ui.comboBox_channel.currentText().replace('channel_', '')) - 1
trace = trace_data[tStim][tRep][tchan]
pass
else:
self.add_message('Cannot handle trace_data of shape: ' + str(trace_data.shape))
return
spike_times = 1000 * np.array(get_spike_times(trace, thresh, fs))
spike_times_s = pd.Series(spike_times)
if spike_times_s.size > nspk:
spikeTrains = spikeTrains.reindex(spike_times_s.index)
nspk = spike_times_s.size
spikeTrains[str(tRep)] = spike_times_s
rasters = spikeTrains
h_file.close()
if len(rasters.shape) > 1:
spks = np.array([])
trns = np.array([])
for trnNum in range(len(rasters.columns)):
spkTrn = np.array(rasters.iloc[:, trnNum].dropna())
trns = np.hstack([trns, (trnNum + 1) * np.ones(len(spkTrn))])
spks = np.hstack([spks, spkTrn])
spikeTimes = rasters.stack()
else:
spikeTimes = rasters.dropna()
# --- Histogram of spike times (1 ms bins)---
sns.set_style("white")
sns.set_style("ticks")
histogram_f = plt.figure(figsize=(8, 3))
axHist = spikeTimes.hist(bins=int(duration / 1), range=(0, duration)) # , figsize=(8,3))
sns.despine()
plt.xlim(0, duration)
plt.xlabel('Time (ms)', size=14)
plt.ylabel('Number of spikes', size=14)
plt.title(str.split(str(filename), '/')[-1].replace('.hdf5', '') + ' '
+ str(self.ui.comboBox_test_num.currentText()).replace('test_', 'Test '))
plt.tick_params(axis='both', which='major', labelsize=14)
plt.grid(False)
histogram_f.show()
return axHist
def generate_tuning_curve(self):
if self.valid_filename():
h_file = h5py.File(unicode(self.filename), 'r')
target_test = self.ui.comboBox_test_num.currentText()
else:
return
target_seg = []
# Find target segment
for segment in h_file.keys():
for test in h_file[segment].keys():
if target_test == test:
target_seg = segment
target_test = test
trace_data = h_file[target_seg][target_test].value
fs = h_file[target_seg].attrs['samplerate_ad']
samples = trace_data.shape[-1]
traces = trace_data.shape[0]
reps = trace_data.shape[1]
if len(h_file[target_seg][target_test].value.shape) > 3:
channels = h_file[target_seg][target_test].value.shape[2]
else:
channels = 1
stim_info = eval(h_file[target_seg][target_test].attrs['stim'])
# Get the values from the combo boxes
if self.ui.comboBox_trace.currentText() != '':
target_trace = int(self.ui.comboBox_trace.currentText().replace(
'trace_', '')) - 1
if self.ui.comboBox_channel.currentText() != '':
target_chan = int(self.ui.comboBox_channel.currentText().replace(
'channel_', '')) - 1
# Get the values from the spinbox
thresh = self.ui.doubleSpinBox_threshold.value()
# print 'test:', target_test
# print 'trace:', target_trace
# print 'chan:', target_chan
# print ''
frequency = []
intensity = []
spike_count = {}
if len(trace_data.shape) == 4:
for t in range(traces):
# print 'trace:', t
# print 'stim_type:', stim_info[t]['components'][0]['stim_type']
# print 'intensity:', stim_info[t]['components'][0]['intensity']
# if stim_info[t]['components'][0]['stim_type'] != 'silence':
# print 'frequency:', stim_info[t]['components'][0]['frequency']
if stim_info[t]['components'][0]['stim_type'] != 'silence':
intensity.append(
stim_info[t]['components'][0]['intensity'])
frequency.append(
stim_info[t]['components'][0]['frequency'] / 1000)
spikes = 0
for r in range(reps):
if self.ui.groupBoxWindow.isChecked():
x_min = np.floor(
self.ui.doubleSpinBox_xmin.value() * fs)
x_max = np.floor(
self.ui.doubleSpinBox_xmax.value() * fs)
temp_trace = trace_data[t][r][target_chan]
trace = temp_trace[x_min:x_max]
pass
else:
trace = trace_data[t][r][target_chan]
spike_times = 1000 * np.array(
get_spike_times(trace, thresh, fs,
self.ui.view._abs))
spikes += len(spike_times)
spike_count[(stim_info[t]['components'][0]['frequency'] /
1000,
stim_info[t]['components'][0]['intensity']
)] = float(spikes) / float(reps)
# print 'spikes:', float(spikes)
# print 'avg_spikes:', float(spikes)/float(reps)
if len(trace_data.shape) == 3:
for t in range(traces):
if stim_info[t]['components'][0]['stim_type'] != 'silence':
intensity.append(
stim_info[t]['components'][0]['intensity'])
frequency.append(
stim_info[t]['components'][0]['frequency'] / 1000)
spikes = 0
for r in range(reps):
trace = trace_data[t][r]
spike_times = 1000 * np.array(
get_spike_times(trace, thresh, fs, self.ui.view._abs))
spikes += len(spike_times)
if stim_info[t]['components'][0]['stim_type'] != 'silence':
spike_count[(stim_info[t]['components'][0]['frequency'] /
1000,
stim_info[t]['components'][0]['intensity']
)] = float(spikes) / float(reps)
# Get only the unique values
frequency = sorted(list(set(frequency)))
intensity = sorted(list(set(intensity)))
# print 'freq', len(frequency), frequency
# print 'inten', len(intensity), intensity
xlist = np.linspace(min(frequency), max(frequency), len(frequency))
ylist = np.linspace(min(intensity), max(intensity), len(intensity))
# print 'xlist', len(xlist), xlist
# print 'ylist', len(ylist), ylist
# Initialize Z
Z = np.empty([len(intensity), len(frequency)])
X, Y = np.meshgrid(xlist, ylist)
for y in range(len(intensity)):
for x in range(len(frequency)):
Z[y][x] = spike_count[(frequency[x], intensity[y])]
# print '(y, x):', y, x
# print 'freq:', frequency[x]
# print 'inten:', intensity[y]
plt.figure()
if self.ui.groupBoxWindow.isChecked():
# Set the min, max and number of contour levels
levels = np.linspace(self.ui.doubleSpinBox_zmin.value(),
self.ui.doubleSpinBox_zmax.value(),
num=(self.ui.spinBoxContourLevels.value() +
1))
cp = plt.contourf(X, Y, Z, levels)
else:
# Auto assign contour levels
cp = plt.contourf(X, Y, Z)
plt.colorbar(cp, label='Mean Spikes Per Presentation')
# print 'X:', X
# print 'Y:', Y
# print 'Z:', Z
# plt.title('Tuning Curve')
if channels == 1:
plt.title(
str.split(str(self.filename), '/')[-1].replace('.hdf5', '') +
' ' + str(self.ui.comboBox_test_num.currentText()).replace(
'test_', 'Test '))
else:
plt.title(
str.split(str(self.filename), '/')[-1].replace('.hdf5', '') +
' ' + str(self.ui.comboBox_test_num.currentText()).replace(
'test_', 'Test ') + ' ' +
str(self.ui.comboBox_channel.currentText()).replace(
'channel_', 'Channel '))
plt.xlabel('Frequency (kHz)')
# plt.xlabel('Frequency (Hz)')
plt.ylabel('Intensity (dB)')
plt.figtext(
.02, .02, 'Threshold: ' +
str(self.ui.doubleSpinBox_threshold.value()) + ' V')
# Idea to try interpolation, didn't work as intended
# plt.imshow(Z, aspect='equal', interpolation=None, origin='lower')
plt.show()
