def load_delta_theta_cfc(hdf5_path, rois, subjects, sessions, outfile=None):
bands_to_extract = [('BOLD bandpass', 'Delta'), ('Delta', 'Theta'),
('Theta', 'Gamma')]
bands = ['BOLD bandpass', 'Delta', 'Theta', 'Alpha', 'Gamma']
# Getting metadata from hdf5_path
# file = h5py.File(hdf5_path)
# sessions = list(file)
# layer = file.get(sessions[0] + '/' + subjects[0] + '/' + rois[0] + '/r_vals')
# data = layer5[...]
# file.close()
tables = {}
for band_pair in bands_to_extract:
phase_band = band_pair[0]
amp_band = band_pair[1]
for b, band in enumerate(bands):
# Getting indcies
if phase_band in band:
p = b
if amp_band in band:
a = b
for session in sessions:
table_name = "%s_%s_%s" % (phase_band, amp_band, session)
print('Creating table %s' % table_name)
cfc_output = np.ndarray(shape=(len(subjects), len(rois)))
for s, subject in enumerate(subjects):
for r, roi in enumerate(rois):
cfc_key = session + '/' + subject + '/' + roi + '/r_vals'
file = h5py.File(hdf5_path)
cfc_data = file.get(cfc_key)[...]
cfc_value = cfc_data[p, a]
cfc_output[s, r] = cfc_value
file.close()
cfc_table = pd.DataFrame(cfc_output, index=subjects, columns=rois)
tables[table_name] = cfc_table
if outfile is not None:
mf.save_xls(tables, outfile)
return tables
return session_dict
outfile = ddir + '/mPLSC_cfc.pkl'
check = input('Run mPLSC? y/n ')
if check == 'y':
check = input('Load cfc data from hdf5? y/n ')
if check == 'y':
hdf5_file = ddir + '/MEG_BOLD_phase_amp_coupling.hdf5'
cfc_tables = _load_cfc_tables_from_hdf5(hdf5_file, meg_sess, subjects,
rois, bands)
for session in list(cfc_tables):
session_dict = cfc_tables[session]
fname = ddir + '/MEG_BOLD_phase_amp_coupling_%s.xlsx' % session
mf.save_xls(session_dict, fname)
cfc_fnames = [
'MEG_BOLD_phase_amp_coupling_%s.xlsx' % sess for sess in meg_sess
]
cfc_filelist = [os.path.join(ddir, f) for f in cfc_fnames]
cfc_tables = _load_cfc_subtables(cfc_filelist)
x_tables = []
for sess in list(cfc_tables):
for t in list(cfc_tables[sess]):
x_tables.append(cfc_tables[sess][t])
# Creating list of y tables
behavior_metadata = pd.read_csv(pdir + '/data/b_variables_mPLSC.txt',
delimiter='\t',
header=None)
y_saliences_zscores_thresh = {}
for behavior_category in list(y_saliences_zscores):
unthresh_df = y_saliences_zscores[behavior_category]
unthresh_vals = unthresh_df.values
thresh_vals = np.ndarray(shape=unthresh_vals.shape)
for r in range(len(unthresh_df.index)):
for c in range(len(list(unthresh_df))):
if np.abs(unthresh_vals[r, c]) < 4:
thresh_vals[r, c] = 0
else:
thresh_vals[r, c] = np.abs(unthresh_vals[r, c])
thresh_df = pd.DataFrame(thresh_vals,
index=unthresh_df.index,
columns=list(unthresh_df))
y_saliences_zscores_thresh[behavior_category] = thresh_df
mf.save_xls(y_saliences_zscores_thresh,
fig_path + '/behavior_saliences_z.xlsx')
print('%s: Averaging saliences within behavior categories' % pu.ctime())
behavior_avg = mf.average_subtable_saliences(y_saliences_zscores_thresh)
behavior_avg.to_excel(fig_path + '/behavior_average_z.xlsx')
print('%s: Creating quick histograms, getting means' % pu.ctime())
def _quick_hist(data, fname=None):
import matplotlib.pyplot as plt
fig, ax = plt.subplots()
ax.hist(data, bins=20)
if fname is not None:
fig.savefig(fname, bbox_inches='tight')
with open(output_file, 'wb') as file:
pkl.dump(output, file)
else:
with open(output_file, 'rb') as file:
output = pkl.load(file)
res_perm = output['permutation_tests']
res_boot = output['bootstrap_tests']
y_saliences = output['y_saliences']
y_saliences_z = output['y_saliences_zscores']
x_saliences = output['x_saliences']
x_saliences_z = output['x_saliences_zscores']
fig_path = pdir + '/figures/mPLSC_power_all_sessions'
mf.save_xls(y_saliences, fig_path + '/behavior_saliences.xlsx')
mf.save_xls(x_saliences, fig_path + '/brain_saliences.xlsx')
y_saliences_zscores_thresh = {}
for behavior_category in list(y_saliences_z):
unthresh_df = y_saliences_z[behavior_category]
unthresh_vals = unthresh_df.values
thresh_vals = np.ndarray(shape=unthresh_vals.shape)
for r in range(len(unthresh_df.index)):
for c in range(len(list(unthresh_df))):
if np.abs(unthresh_vals[r, c]) < 4:
thresh_vals[r, c] = 0
else:
thresh_vals[r, c] = np.abs(unthresh_vals[r, c])
thresh_df = pd.DataFrame(thresh_vals,
index=unthresh_df.index,