def get_avg_firing_rates_around_events(spike_rates, event_time_points, ev_video_df, time_around_event=5,
keep_trials=False):
frames_of_events = sync_funcs.time_point_to_frame_from_video_df(ev_video_df, event_time_points)
frames_around_event = time_around_event * VIDEO_FRAME_RATE
firing_rate_around_events = np.zeros((len(event_time_points), len(spike_rates), 2 * frames_around_event))
for f in np.arange(len(frames_of_events)):
frame = frames_of_events[f]
firing_rate_around_events[f, :, :] = spike_rates[:, frame - frames_around_event:
frame + frames_around_event]
avg_firing_rate_around_events = firing_rate_around_events.mean(axis=0)
if keep_trials:
return firing_rate_around_events, avg_firing_rate_around_events
return avg_firing_rate_around_events
time_points_of_non_trial_pokes_selected.append(
all_pokes[non_trial_pokes_in_all_pokes_to_select_from[np.squeeze(
selecetd_index)]])
dif_of_non_trial_pokes_to_previous_poke_to_select_from = \
np.delete(dif_of_non_trial_pokes_to_previous_poke_to_select_from, selecetd_index)
all_pokes_to_select_from = np.delete(
non_trial_pokes_in_all_pokes_to_select_from, selecetd_index)
time_points_of_non_trial_pokes_selected = np.squeeze(
np.array(time_points_of_non_trial_pokes_selected))
# Create the corresponding time windows (and in frames)
window_size_frames = np.array(trial_durations / const_comm.SAMPLING_FREQUENCY *
120,
dtype=np.int32)
trial_frames = sync_funcs.time_point_to_frame_from_video_df(
event_dataframes['ev_video'], time_points_of_trial_pokes)
non_trial_frames = sync_funcs.time_point_to_frame_from_video_df(
event_dataframes['ev_video'], time_points_of_non_trial_pokes_selected)
number_of_trial_pokes = len(trial_frames)
number_of_non_trial_pokes = len(non_trial_frames)
trial_windows_in_frames = [
np.arange(trial_frames[tf] - window_size_frames[tf],
trial_frames[tf] + window_size_frames[tf], 1)
for tf in np.arange(len(trial_frames))
]
non_trial_windows_in_frames = [
np.arange(non_trial_frames[ntf] - window_size_frames[ntf],
non_trial_frames[ntf] + window_size_frames[ntf], 1)
for ntf in np.arange(len(non_trial_frames))
def tp_to_frame(frame):
return sync_funcs.time_point_to_frame_from_video_df(ev_video, frame)[0] - frames_before_tb
# Patterned behaviour is defined as going to the ball and then requiring a short time interval to go to the poke with
# no or very little digression from the "touch ball -> go to poke" pattern
time_between_tb_and_p = 15
frames_between_tb_and_p = time_between_tb_and_p * 120
frames_before_tb = 1 * 120
index_of_events_tb_followed_by_fast_p = np.squeeze(np.argwhere((events_successful_trial_pokes -
events_touch_ball_successful_trial) /
const.SAMPLING_FREQUENCY < time_between_tb_and_p))
# The following delete is because during the 30th time span the rat just sits doing nothing
index_of_events_tb_followed_by_fast_p = np.delete(index_of_events_tb_followed_by_fast_p, 30)
frames_tb_followed_by_fast_p = sync_funcs.time_point_to_frame_from_video_df(ev_video,
events_touch_ball_successful_trial[
index_of_events_tb_followed_by_fast_p])
frames_fast_p = sync_funcs.time_point_to_frame_from_video_df(ev_video,
events_successful_trial_pokes[
index_of_events_tb_followed_by_fast_p])
number_of_patterned_events = len(index_of_events_tb_followed_by_fast_p)
windows_of_patterned_behaviour = [np.arange(frames_tb_followed_by_fast_p[frame],
frames_fast_p[frame], 1) for
frame in np.arange(number_of_patterned_events)]
# <editor-fold desc="Have a look at the selected patterned behaviour events">
globals(),
'index',
'output',
'show_rasters_decrease',
'args',
slider_limits=[0, len(decreasing_firing_rates_neuron_index) - 1])
# </editor-fold>
# -------------------------------------------------
# <editor-fold desc="Looking at the time of touching ball. Nothing happens!">
plt.plot(cdt.space_data(avg_firing_rate_around_suc_trials, 10).transpose())
touch_ball_trials = event_dataframes['ev_rat_touch_ball']
frames_of_touch_ball_trials = sync_funcs.time_point_to_frame_from_video_df(
event_dataframes['ev_video'], touch_ball_trials['AmpTimePoints'].values)
firing_rate_around_touch_trials = np.zeros(
(len(touch_ball_trials), len(spike_rates), 2 * frames_around_beam_break))
for f in np.arange(len(firing_rate_around_touch_trials)):
frame = frames_of_touch_ball_trials[f]
firing_rate_around_touch_trials[
f, :, :] = spike_rates[:, frame - frames_around_beam_break:frame +
frames_around_beam_break]
avg_firing_rate_around_touch_trials = firing_rate_around_touch_trials.mean(
axis=0)
avg_firing_rate_around_touch_trials -= np.expand_dims(
avg_firing_rate_around_touch_trials[:, :600].mean(axis=1), axis=1)
plt.imshow(avg_firing_rate_around_touch_trials,
def tp_to_frame(tp):
return sync_funcs.time_point_to_frame_from_video_df(
event_dataframes['ev_video'], tp)[0]
