def estimate_spindle_band_power(lfps,
sampling_frequency,
spindle_band=(10, 16),
start_time=0.00,
multitaper_params=_DEFAULT_MULTITAPER_PARAMS):
"""Estimates the spindle power of each LFP.
Parameters
----------
lfps : ndarray, shape (n_time, n_signals)
sampling_frequency : float
spindle_band : (start_freq, end_freq)
start_time : float
multitaper_params : dict
Returns
-------
time : ndarray, shape (n_time_windows,)
spindle_band_power : ndarray (n_time_windows, n_signals)
"""
m = Multitaper(lfps,
sampling_frequency=sampling_frequency,
**multitaper_params)
c = Connectivity.from_multitaper(m)
freq_ind = ((c.frequencies > spindle_band[0]) &
(c.frequencies < spindle_band[1]))
power = c.power()[..., freq_ind, :]
return c.time, power
def get_replay_triggered_power(lfps,
replay_info,
tetrode_info,
multitaper_params,
window_offset=(-0.250, 0.250),
sampling_frequency=1500):
ripple_locked_lfps = reshape_to_segments(
lfps,
replay_info.loc[:, ['start_time', 'end_time']],
window_offset=window_offset,
sampling_frequency=sampling_frequency)
ripple_locked_lfps = (ripple_locked_lfps.to_xarray().to_array().rename({
'variable':
'tetrodes'
}).transpose('time', 'replay_id', 'tetrodes').dropna('replay_id'))
ripple_locked_lfps = (ripple_locked_lfps -
ripple_locked_lfps.mean(['replay_id']))
start_time = ripple_locked_lfps.time.min().values / np.timedelta64(1, 's')
m = Multitaper(ripple_locked_lfps.values,
**multitaper_params,
start_time=start_time)
c = Connectivity.from_multitaper(m)
dimension_names = ['time', 'frequency', 'tetrode']
data_vars = {'power': (dimension_names, c.power())}
coordinates = {
'time': _center_time(c.time),
'frequency': c.frequencies + np.diff(c.frequencies)[0] / 2,
'tetrode': lfps.columns,
}
return (xr.Dataset(data_vars,
coords=coordinates).sel(frequency=slice(0, 300)))
def estimate_theta_power(time, tetrode_info, multitaper_params=None):
if multitaper_params is None:
multitaper_params = MULTITAPER_PARAMETERS['4Hz']
is_brain_areas = (
tetrode_info.area.astype(str).str.upper().isin(BRAIN_AREAS))
tetrode_keys = tetrode_info.loc[is_brain_areas].index
lfps = get_LFPs(tetrode_keys, ANIMALS)
m = Multitaper(lfps.values,
**multitaper_params,
start_time=lfps.index[0].total_seconds())
c = Connectivity.from_multitaper(m)
coordinates = {
'time': pd.to_timedelta(c.time, unit='s'),
'frequency': c.frequencies,
'tetrode': lfps.columns,
}
dimension_names = ['time', 'frequency', 'tetrode']
data_vars = {'power': (dimension_names, c.power())}
theta_power = (xr.Dataset(data_vars, coords=coordinates).sel(
frequency=slice(4, 12)).mean('frequency').dropna('tetrode').interp(
time=time).to_dataframe().unstack(level=0))
theta_power_change = theta_power.transform(lambda df: df / df.mean())
theta_power_zscore = theta_power.transform(lambda df:
(df - df.mean()) / df.std())
return dict(
theta_power=theta_power,
theta_power_change=theta_power_change,
theta_power_zscore=theta_power_zscore,
)
def estimate_ripple_band_power(lfps, sampling_frequency):
"""Estimates the 200 Hz power of each LFP.
Parameters
----------
lfps : ndarray, shape (n_time, n_signals)
sampling_frequency : float
Returns
-------
ripple_band_power : ndarray (n_time, n_signals)
"""
n_time = lfps.shape[0]
m = Multitaper(lfps,
sampling_frequency=sampling_frequency,
time_halfbandwidth_product=1,
time_window_duration=0.020,
time_window_step=1 / sampling_frequency)
c = Connectivity.from_multitaper(m)
closest_200Hz_freq_ind = np.argmin(np.abs(c.frequencies - 200))
power = c.power()[..., closest_200Hz_freq_ind, :].squeeze()
n_power_time = power.shape[0]
unobserved = np.full((n_time - n_power_time, *power.shape[1:]), np.nan)
return np.concatenate((power, unobserved))
def estimate_ripple_band_power(lfps, sampling_frequency):
m = Multitaper(lfps.values, sampling_frequency=sampling_frequency,
time_halfbandwidth_product=1,
time_window_duration=0.020,
time_window_step=0.020,
start_time=lfps.index[0].total_seconds())
c = Connectivity.from_multitaper(m)
closest_200Hz_freq_ind = np.argmin(np.abs(c.frequencies - 200))
power = c.power()[..., closest_200Hz_freq_ind, :].squeeze() + np.spacing(1)
n_samples = int(0.020 * sampling_frequency)
index = lfps.index[np.arange(1, power.shape[0] * n_samples + 1, n_samples)]
power = pd.DataFrame(power, index=index)
return power.reindex(lfps.index)
# Granger causality from v1 to v4
granger_v1_v4 = np.empty((1, 203, 0))
for idx_v1 in range(len(stc_label_v1[1].vertices)):
granger_v4 = np.empty((1, 203, 0))
for idx_v4 in range(len(stc_label_v4[1].vertices)):
signal = np.append(signal_v1[:, :, idx_v1, np.newaxis],
signal_v4[:, :, idx_v4, np.newaxis],
axis=2)
# Compute granger causality
m = Multitaper(signal,
sfreq,
time_halfbandwidth_product=2,
start_time=-0.8,
n_tapers=1)
c = Connectivity(fourier_coefficients=m.fft(),
frequencies=m.frequencies)
granger = c.pairwise_spectral_granger_prediction()
granger_v4 = np.append(granger_v4,
granger[..., 0, 1, np.newaxis],
axis=2)
granger_v1_v4 = np.append(granger_v1_v4,
granger_v4.mean(2)[..., np.newaxis],
axis=2)
granger_all_v1_v4 = granger_v1_v4.mean(2)
# Granger causality from v4 to v1
granger_v1_v4 = np.empty((1, 203, 0))
for idx_v1 in range(len(stc_label_v1[1].vertices)):
granger_v4 = np.empty((1, 203, 0))
for idx_v4 in range(len(stc_label_v4[1].vertices)):
signal = np.append(signal_v4[:, :, idx_v4, np.newaxis],
# Create signals input
datarray = np.asarray(tcs_v1)
signal_v1 = np.transpose(datarray.mean(2),
(2, 1, 0)) #(times,epochs,signals))
datarray = np.asarray(tcs_v4)
signal_v4 = np.transpose(datarray.mean(2), (2, 1, 0))
signal = np.append(signal_v1, signal_v4, axis=2)
# Compute granger causality
m = Multitaper(signal,
sfreq,
time_halfbandwidth_product=2,
start_time=-0.8,
n_tapers=1)
c = Connectivity(fourier_coefficients=m.fft()[:, :, :, 0:130, :],
frequencies=m.frequencies[0:130])
cross_spectral_matrix = c._expectation(c._cross_spectral_matrix)
n_signals = cross_spectral_matrix.shape[-1]
total_power = c.power()
n_frequencies = cross_spectral_matrix.shape[-3]
non_neg_index = np.arange(0, (n_frequencies + 1) // 2)
new_shape = list(cross_spectral_matrix.shape)
new_shape[-3] = non_neg_index.size
predictive_power = np.empty(new_shape)
nvert_v1 = len(stc_label_v1[1].vertices)
nvert_v4 = len(stc_label_v4[1].vertices)
n_sign_v1 = np.arange(nvert_v1)
n_sign_v4 = np.arange(nvert_v1, nvert_v1 + nvert_v4)
signal_v1[:,0:idx_slow,idx_v1,np.newaxis], axis=2)
signal_mt_v1_med = np.append(signal_mt[:,idx_slow:idx_slow+idx_med,idx_mt,np.newaxis],
signal_v1[:,idx_slow:idx_slow+idx_med,idx_v1,np.newaxis], axis=2)
signal_mt_v1_fast = np.append(signal_mt[:,idx_slow+idx_med:idx_all,idx_mt,np.newaxis],
signal_v1[:,idx_slow+idx_med:idx_all,idx_v1,np.newaxis], axis=2)
# Compute granger causality
m_v1_mt_slow = Multitaper(signal_v1_mt_slow, sfreq, time_halfbandwidth_product=2, start_time=-0.8, n_tapers=1)
m_v1_mt_med = Multitaper(signal_v1_mt_med, sfreq, time_halfbandwidth_product=2, start_time=-0.8, n_tapers=1)
m_v1_mt_fast = Multitaper(signal_v1_mt_fast, sfreq, time_halfbandwidth_product=2, start_time=-0.8, n_tapers=1)
m_mt_v1_slow = Multitaper(signal_mt_v1_slow, sfreq, time_halfbandwidth_product=2, start_time=-0.8, n_tapers=1)
m_mt_v1_med = Multitaper(signal_mt_v1_med, sfreq, time_halfbandwidth_product=2, start_time=-0.8, n_tapers=1)
m_mt_v1_fast = Multitaper(signal_mt_v1_fast, sfreq, time_halfbandwidth_product=2, start_time=-0.8, n_tapers=1)
c_v1_mt_slow = Connectivity(fourier_coefficients=m_v1_mt_slow.fft(), frequencies=m_v1_mt_slow.frequencies)
c_v1_mt_med = Connectivity(fourier_coefficients=m_v1_mt_med.fft(), frequencies=m_v1_mt_med.frequencies)
c_v1_mt_fast = Connectivity(fourier_coefficients=m_v1_mt_fast.fft(), frequencies=m_v1_mt_fast.frequencies)
c_mt_v1_slow = Connectivity(fourier_coefficients=m_mt_v1_slow.fft(), frequencies=m_mt_v1_slow.frequencies)
c_mt_v1_med = Connectivity(fourier_coefficients=m_mt_v1_med.fft(), frequencies=m_mt_v1_med.frequencies)
c_mt_v1_fast = Connectivity(fourier_coefficients=m_mt_v1_fast.fft(), frequencies=m_mt_v1_fast.frequencies)
granger_v1_mt_slow[:,:,idx_mt] = c_v1_mt_slow.pairwise_spectral_granger_prediction()[...,0,1]
granger_v1_mt_med[:,:,idx_mt] = c_v1_mt_med.pairwise_spectral_granger_prediction()[...,0,1]
granger_v1_mt_fast[:,:,idx_mt] = c_v1_mt_fast.pairwise_spectral_granger_prediction()[...,0,1]
granger_mt_v1_slow[:,:,idx_mt] = c_mt_v1_slow.pairwise_spectral_granger_prediction()[...,0,1]
granger_mt_v1_med[:,:,idx_mt] = c_mt_v1_med.pairwise_spectral_granger_prediction()[...,0,1]
granger_mt_v1_fast[:,:,idx_mt] = c_mt_v1_fast.pairwise_spectral_granger_prediction()[...,0,1]