def save_all_clusters_trials(self,
matching_criteria_dicts,
save_path='/home/skeshav/all_trials.csv'):
with temp_setattr(self, 'df_all_conditions', pd.DataFrame()):
for i, c in enumerate(self.cells()):
for matching_criteria_dict in matching_criteria_dicts:
with temp_setattr(c.block, 'current_condition',
matching_criteria_dict):
self.df_all_conditions = self.df_all_conditions.append(
c.block.to_df(), ignore_index=True)
self.df_all_conditions.to_csv(save_path)
def plot_all_sub_stimuli_histograms(self,
matching_dictionary,
n_bins=30,
save=False,
format='png'):
for c in tqdm(self.cells()):
with temp_setattr(c.block, 'current_condition',
matching_dictionary):
fig = plt.figure(facecolor='w', figsize=(8, 3))
plt.title('Cluster {}_Channel {}'.format(c.id, c.depth))
c.block.plot_all_sub_stimuli(n_bins=n_bins, fig=fig)
if save:
save_string = self.matching_dictionary_to_title_string(
matching_dictionary)
save_dir = os.path.join(self.main_dir, 'histogram_plots',
save_string)
if not os.path.isdir(save_dir):
os.makedirs(save_dir)
save_path = os.path.join(
save_dir, 'cluster_{}_channel_{}.{}'.format(
c.id, c.depth, format))
fig.savefig(save_path, format=format)
else:
plt.show()
def plot_overlay_histogram(self,
matching_dictionaries,
n_bins=20,
save=False,
format='png'): # FIXME:
for c in self.cells():
fig = plt.figure(facecolor='w', figsize=(8, 3))
for matching_dictionary in matching_dictionaries:
with temp_setattr(c.block, 'current_condition',
matching_dictionary):
fig = c.block.plot_all_sub_stimuli(n_bins=n_bins, fig=fig)
plt.suptitle('Cluster {}_ Channel {}'.format(c.id, c.depth))
histogram_axes = [
fig.axes[i] for i in range(len(fig.axes)) if i % 2 == 1
]
normalise_axes(histogram_axes, space=3)
#plt.show()
if save:
save_string = self.get_save_string(matching_dictionaries)
save_dir = os.path.join(self.main_dir, 'histogram_plots',
save_string)
if not os.path.isdir(save_dir):
os.makedirs(save_dir)
save_path = os.path.join(
save_dir,
'cluster_{}_channel_{}.{}'.format(c.id, c.depth, format))
fig.savefig(save_path, format=format)
else:
plt.show()
def plot_psth(self, extension):
for angle in self.angles:
plt.figure()
angle_bsl_psth_y = []
angle_spin_psth_y = []
angle_bsl2_psth_y = []
for cell in self:
with temp_setattr(cell.block, 'current_condition',
{'angle': angle}):
bsl_psth, spin_psth, bsl2_psth = cell.block.get_psth()
bsl_psth_x, bsl_psth_y = bsl_psth
spin_psth_x, spin_psth_y = spin_psth
bsl2_psth_x, bsl2_psth_y = bsl2_psth
angle_bsl_psth_y.append(bsl_psth_y)
angle_spin_psth_y.append(spin_psth_y)
angle_bsl2_psth_y.append(bsl2_psth_y)
angle_bsl_psth_y = np.array(angle_bsl_psth_y,
dtype=np.float64).sum(0)
angle_spin_psth_y = np.array(angle_spin_psth_y,
dtype=np.float64).sum(0)
angle_bsl2_psth_y = np.array(angle_bsl2_psth_y,
dtype=np.float64).sum(0)
plt.step(bsl_psth_x, angle_bsl_psth_y, where='post', color='blue')
plt.step(spin_psth_x,
angle_spin_psth_y,
where='post',
color='green')
plt.step(bsl2_psth_x, angle_bsl2_psth_y, where='post', color='red')
plt.savefig(
os.path.join(
self.main_dir, '{}_PSTH.{}'.format(self.get_str(angle),
extension)))
plt.close()
def plot_single_trial_detection(self, angle,
trial_id): # FIXME: trial plotter
with temp_setattr(self.block, 'current_condition', {'angle': angle}):
trial = self.block.trials[trial_id]
trace = trial.processed_trace
params = self.block.detection_params[angle]
delta_wave = diff(box_smooth(trace, params.n_pnts_peak),
box_smooth(trace, params.n_pnts_bsl),
params.n_pnts_rise_t)
plt.plot(trial.stimulus.x[:len(delta_wave)],
delta_wave,
linewidth=0.3
) # FIXME: find why different length # REFACTOR: extract
trace = signal.medfilt(trace.copy(), params.median_kernel_size)
events_params = trial.events.get_events_point_params(
) # WARNING: not analysed_events because for GUI
events_pos, peaks_pos, half_rises, _ = events_params
_plot_trace(trace,
'trial_{}'.format(trial_id),
events_pos,
peaks_pos,
color='grey',
x=trial.stimulus.x,
y_shift=0)
def save_detection(self, processing_type):
for angle in self.angles:
with temp_setattr(self.block, 'current_condition', {
'keep': True,
'angle': angle
}):
self.block.save_detection(angle, processing_type)
def get_results_as_df(self, angle):
with temp_setattr(self.block, 'current_condition', {
'keep': True,
'angle': angle
}):
return pd.concat(
[self._get_base_df(angle),
self.block.get_results_df()],
axis=1)
def do_stats(self, matching_dictionary):
for c in self.cells():
with temp_setattr(c.block, 'current_condition',
matching_dictionary):
c.set_main_dir(self.main_dir)
c.block.save_stats()
self.stats.append(c.block.stats_df)
print(self.stats)
def population_data(self, matching_dictionary, labels, time_match_label):
all_groups = {}
for label in labels:
spikes_in_label = []
for c in self.cells():
with temp_setattr(c.block, 'current_condition',
matching_dictionary):
spikes_in_label.extend(
c.block.get_events_in_period(label, time_match_label))
all_groups.setdefault(label, spikes_in_label)
return all_groups
def analyse(self, processed=True):
"""
Loop over all recordings for each angle
self.noises with the high frequency noise of each recording
and self.events_params with the tuple of lists: (events_pos, peaks_pos, half_rises, peak_ampls)
"""
for angle in self.angles:
with temp_setattr(self.block, 'current_condition', {
'keep': True,
'angle': angle
}):
self.block.analyse(angle, processed=processed)
def get_detection_params_path(self, angle):
with temp_setattr(self, 'current_condition', {
'keep': True,
'angle': angle
}):
expected_path = os.path.join(
self.cell.dir, '{}_detection_params.csv'.format(self))
if os.path.exists(expected_path):
return expected_path
else:
warnings.warn(
'File {} not found to reload detection parameters, creating new file'
.format(expected_path))
return None
def analyse(
self,
angle,
processed=True
): # WARNING: all trials # TODO: rename? + DOUBLE WARNING should be only one angle
"""
All trials but one angle
:param processed:
:return:
"""
with temp_setattr(self, 'current_condition', {'angle': angle}):
self.__reset_detection()
for trial in self.kept_trials:
trial.detect(self.detection_params[angle],
processed) # detect all whether kept or not
def get_all_cells_all_trials_spiking_tables(self,
angle,
levels_var_name='position'):
spiking_tables = []
for i, cell in enumerate(tqdm(self)):
block = cell.block
with temp_setattr(block, 'current_condition', {
'keep': True,
'angle': angle
}):
FieldOfView.check_all_trials_have_same_n_repeats(
block) # TODO: ideally check all cells have same ?
spiking_table, levels, cmd = block.get_freqs_from_timetable(
levels_var_name)
spiking_tables.append(spiking_table)
spiking_tables = np.dstack(spiking_tables) # TEST: check shape
spiking_tables = np.transpose(spiking_tables, (2, 0, 1))
return levels, spiking_tables, cmd
def get_events_integral(self, angle, trial_id):
"""
All trials but one angle (because for detection)
:param trial_id:
:return:
"""
with temp_setattr(self, 'current_condition', {'angle': angle}):
integrals = []
trial = self.trials[trial_id]
trace = trial.processed_trace
filtered_trace = signal.medfilt(trace.copy(), 3)
for event in trial.events:
integrals.append(event.get_integral(filtered_trace))
if integrals:
if __debug__:
print(integrals)
return np.array(integrals).mean()
else:
return 0
def analyse_block(self, angle):
with temp_setattr(self.block, 'current_condition', {
'keep': True,
'angle': angle
}):
self.block.save_stats()
def plot_all(self, angle):
with temp_setattr(self.block, 'current_condition', {
'keep': True,
'angle': angle
}):
self.block.plot()
