def extract(subject,
session,
recording,
epoch,
signal_type='BB',
BEM='three_layer',
debug=False,
chunks=100,
njobs=4):
mne.set_log_level('WARNING')
lcmv.logging.getLogger().setLevel(logging.INFO)
set_n_threads(1)
logging.info('Reading stimulus data')
if epoch == 'stimulus':
data_cov, epochs, epochs_filename = get_stim_epoch(
subject, session, recording)
else:
data_cov, epochs, epochs_filename = get_response_epoch(
subject, session, recording)
raw_filename = glob('TODO' % (subject, session, recording))
trans_filename = glob('TODO' % (subject, session, recording))[0]
logging.info('Setting up source space and forward model')
forward, bem, source = sr.get_leadfield(subject,
raw_filename,
epochs_filename,
trans_filename,
bem_sub_path='bem_ft')
labels = sr.get_labels(subject)
labels = sr.labels_exclude(labels,
exclude_filters=[
'wang2015atlas.IPS4', 'wang2015atlas.IPS5',
'wang2015atlas.SPL', 'JWDG_lat_Unknown'
])
labels = sr.labels_remove_overlap(
labels,
priority_filters=['wang', 'JWDG'],
)
fois_h = np.arange(36, 162, 4)
fois_l = np.arange(2, 36, 1)
tfr_params = {
'HF': {
'foi': fois_h,
'cycles': fois_h * 0.25,
'time_bandwidth': 2 + 1,
'n_jobs': njobs,
'est_val': fois_h,
'est_key': 'HF',
'sf': 600,
'decim': 10
},
'LF': {
'foi': fois_l,
'cycles': fois_l * 0.4,
'time_bandwidth': 1 + 1,
'n_jobs': njobs,
'est_val': fois_l,
'est_key': 'LF',
'sf': 600,
'decim': 10
}
}
events = epochs.events[:, 2]
filters = lcmv.setup_filters(epochs.info, forward, data_cov, None, labels)
set_n_threads(1)
for i in range(0, len(events), chunks):
filename = lcmvfilename(subject,
session,
signal_type,
recording,
chunk=i)
if os.path.isfile(filename):
continue
if signal_type == 'BB':
logging.info('Starting reconstruction of BB signal')
M = lcmv.reconstruct_broadband(filters,
epochs.info,
epochs._data[i:i + chunks],
events[i:i + chunks],
epochs.times,
njobs=1)
else:
logging.info('Starting reconstruction of TFR signal')
M = lcmv.reconstruct_tfr(filters,
epochs.info,
epochs._data[i:i + chunks],
events[i:i + chunks],
epochs.times,
est_args=tfr_params[signal_type],
njobs=4)
M.to_hdf(filename, 'epochs')
set_n_threads(njobs)
def extract_reconstruct_tfr_block(subject,
session,
block,
epoch,
signal_type='BB',
BEM='three_layer',
debug=False,
chunks=50,
njobs=4):
#check if the block exists and recording
subject_int = int(subject[1:])
fname = path + '/filenames_sub%i.pickle' % (subject_int)
f = open(fname, 'rb')
data = pickle.load(f)
df = pd.DataFrame.from_dict(data)
blocks = np.array(
df[df.subject == subject_int][df.session == session].block)
if block in blocks:
recording = df[df.subject == subject_int][df.session == session][
df.block == block].trans_matrix.iloc[0]
else:
print('block does not exist')
sys.exit(0)
if epoch == 'stimulus':
fname = get_filenames_block(subject, session, block, recording)[0][0]
else:
fname = get_filenames_block(subject, session, block, recording)[1][0]
# fname_aux='filter_sub%i_SESS%i_recording%i_epoch%s.pickle'%( subject_int,session,recording,epoch)
# filename = join(path+'/extra', fname_aux)
# f=open(filename,'rb')
# filters=pickle.load(f)
# f.close()
filters = extract_filter(subject,
session_int,
recording,
epoch,
signal_type=signal_type)
fois_h = np.arange(36, 162, 4)
fois_l = np.arange(2, 36, 1)
tfr_params = {
'HF': {
'foi': fois_h,
'cycles': fois_h * 0.25,
'time_bandwidth': 2 + 1,
'n_jobs': njobs,
'est_val': fois_h,
'est_key': 'HF',
'sf': 600,
'decim': 10
},
'LF': {
'foi': fois_l,
'cycles': fois_l * 0.4,
'time_bandwidth': 1 + 1,
'n_jobs': njobs,
'est_val': fois_l,
'est_key': 'LF',
'sf': 600,
'decim': 10
}
}
print(fname)
print('loading data')
epochs = preprocessing.load_epochs([fname])
print('concataneiting data')
epochs = preprocessing.concatenate_epochs(epochs, None)
print('Picking pick_channels')
epochs = epochs.pick_channels(
[x for x in epochs.ch_names if x.startswith('M')])
events = epochs.events[:, 2]
for i in range(0, len(events), chunks):
print('chunk:', i)
filename = lcmvfilename_block(subject,
session,
block,
signal_type,
recording,
epoch,
chunk=i)
if os.path.isfile(filename):
continue
if signal_type == 'BB':
logging.info('Starting reconstruction of BB signal')
M = lcmv.reconstruct_broadband(filters,
epochs.info,
epochs._data[i:i + chunks],
events[i:i + chunks],
epochs.times,
njobs=1)
else:
logging.info('Starting reconstruction of TFR signal')
M = lcmv.reconstruct_tfr(filters,
epochs.info,
epochs._data[i:i + chunks],
events[i:i + chunks],
epochs.times,
est_args=tfr_params[signal_type],
njobs=1)
M.to_hdf(filename, 'epochs')
del M
del epochs
print('done')
def extract(
subject,
session,
epoch_type="stimulus",
signal_type="BB",
only_glasser=False,
BEM="three_layer",
debug=False,
chunks=100,
njobs=4,
):
mne.set_log_level("WARNING")
pymeglcmv.logging.getLogger().setLevel(logging.INFO)
set_n_threads(1)
logging.info("Reading stimulus data")
if epoch_type == "stimulus":
data_cov, epochs = get_stim_epoch(subject, session)
elif epoch_type == "response":
data_cov, epochs = get_response_epoch(subject, session)
else:
raise RuntimeError("Did not recognize epoch")
logging.info("Setting up source space and forward model")
forward, bem, source = get_leadfield(subject, session, BEM)
if not only_glasser:
labels = pymegsr.get_labels(
subject="S%02i" % subject,
filters=["*wang*.label", "*JWDG*.label"],
annotations=["HCPMMP1"],
)
labels = pymegsr.labels_exclude(
labels=labels,
exclude_filters=[
"wang2015atlas.IPS4",
"wang2015atlas.IPS5",
"wang2015atlas.SPL",
"JWDG_lat_Unknown",
],
)
labels = pymegsr.labels_remove_overlap(
labels=labels, priority_filters=["wang", "JWDG"])
else:
labels = pymegsr.get_labels(
subject="S%02i" % subject,
filters=["select_nothing"],
annotations=["HCPMMP1"],
)
# Now chunk Reconstruction into blocks of ~100 trials to save Memory
fois = np.arange(10, 150, 5)
lfois = np.arange(1, 10, 1)
tfr_params = {
"F": {
"foi": fois,
"cycles": fois * 0.1,
"time_bandwidth": 2,
"n_jobs": 1,
"est_val": fois,
"est_key": "F",
},
"LF": {
"foi": lfois,
"cycles": lfois * 0.25,
"time_bandwidth": 2,
"n_jobs": 1,
"est_val": lfois,
"est_key": "LF",
},
}
events = epochs.events[:, 2]
data = []
filters = pymeglcmv.setup_filters(epochs.info, forward, data_cov, None,
labels)
set_n_threads(1)
for i in range(0, len(events), chunks):
filename = lcmvfilename(subject,
session,
signal_type,
epoch_type,
chunk=i,
only_glasser=only_glasser)
logging.info(filename)
# if os.path.isfile(filename):
# continue
if signal_type == "BB":
logging.info("Starting reconstruction of BB signal")
M = pymeglcmv.reconstruct_broadband(
filters,
epochs.info,
epochs._data[i:i + chunks],
events[i:i + chunks],
epochs.times,
njobs=1,
)
else:
logging.info("Starting reconstruction of TFR signal")
M = pymeglcmv.reconstruct_tfr(
filters,
epochs.info,
epochs._data[i:i + chunks],
events[i:i + chunks],
epochs.times,
est_args=tfr_params[signal_type],
njobs=4,
)
M.to_hdf(filename, "epochs", mode="w")
set_n_threads(njobs)
def extract_reconstruct_tfr(subject,
session,
recording,
epoch,
signal_type='BB',
BEM='three_layer',
debug=False,
chunks=50,
njobs=4):
if epoch == 'stimulus':
filenames = glob(get_filenames(subject, session, recording)[0])
else:
filenames = glob(get_filenames(subject, session, recording)[1])
subject_int = int(subject[1:])
fname = 'filter_sub%i_SESS%i_recording%i_epoch%s.pickle' % (
subject_int, session, recording, epoch)
filename = join(path + '/extra', fname)
f = open(filename, 'rb')
filters = pickle.load(f)
f.close()
fois_h = np.arange(36, 162, 4)
fois_l = np.arange(2, 36, 1)
tfr_params = {
'HF': {
'foi': fois_h,
'cycles': fois_h * 0.25,
'time_bandwidth': 2 + 1,
'n_jobs': njobs,
'est_val': fois_h,
'est_key': 'HF',
'sf': 600,
'decim': 10
},
'LF': {
'foi': fois_l,
'cycles': fois_l * 0.4,
'time_bandwidth': 1 + 1,
'n_jobs': njobs,
'est_val': fois_l,
'est_key': 'LF',
'sf': 600,
'decim': 10
}
}
print('filters done')
for ifname, fname in enumerate(filenames):
print(fname)
epochs = preprocessing.load_epochs([fname])
epochs = preprocessing.concatenate_epochs(epochs, None)
epochs = epochs.pick_channels(
[x for x in epochs.ch_names if x.startswith('M')])
events = epochs.events[:, 2]
for i in range(0, len(events), chunks):
print('chunk:', i)
filename = lcmvfilename(subject,
session,
signal_type,
recording,
epoch + str(ifname),
chunk=i)
if os.path.isfile(filename):
continue
if signal_type == 'BB':
logging.info('Starting reconstruction of BB signal')
M = lcmv.reconstruct_broadband(filters,
epochs.info,
epochs._data[i:i + chunks],
events[i:i + chunks],
epochs.times,
njobs=1)
else:
logging.info('Starting reconstruction of TFR signal')
M = lcmv.reconstruct_tfr(filters,
epochs.info,
epochs._data[i:i + chunks],
events[i:i + chunks],
epochs.times,
est_args=tfr_params[signal_type],
njobs=1)
M.to_hdf(filename, 'epochs')
del M
del epochs
print('done')
def extract(
recording_number,
epoch,
signal_type="BB",
BEM="three_layer",
chunks=100,
njobs=4,
glasser_only=True,
):
recording = ps.recordings[recording_number]
mne.set_log_level("ERROR")
lcmv.logging.getLogger().setLevel(logging.INFO)
set_n_threads(1)
logging.info("Reading stimulus data")
data_cov, epochs = get_epoch(epoch, recording)
raw_filename = raw_path / recording.filename
trans_filename = trans_path / (
"SQC_S%02i-SESS%i_B%i_trans.fif" %
(recording.subject, recording.session, recording.block[1]))
epoch_filename = ps.filenames(recording.subject, epoch, recording.session,
recording.block[1])[0]
logging.info("Setting up source space and forward model")
forward, bem, source = sr.get_leadfield(
"SQC_S%02i" % recording.subject,
str(raw_filename),
str(epoch_filename),
str(trans_filename),
bem_sub_path="bem",
sdir="/home/nwilming/seqconf/fsdir/",
)
if glasser_only:
labels = sr.get_labels(
"SQC_S%02i" % recording.subject,
filters=["*wang2015atlas*"],
sdir="/home/nwilming/seqconf/fsdir/",
)
else:
labels = sr.get_labels("SQC_S%02i" % recording.subject,
sdir="/home/nwilming/seqconf/fsdir/")
labels = sr.labels_exclude(
labels,
exclude_filters=[
"wang2015atlas.IPS4",
"wang2015atlas.IPS5",
"wang2015atlas.SPL",
"JWDG_lat_Unknown",
],
)
labels = sr.labels_remove_overlap(labels,
priority_filters=["wang", "JWDG"])
fois_h = np.arange(36, 162, 4)
fois_l = np.arange(2, 36, 1)
tfr_params = {
"HF": {
"foi": fois_h,
"cycles": fois_h * 0.25,
"time_bandwidth": 2 + 1,
"n_jobs": njobs,
"est_val": fois_h,
"est_key": "HF",
"sf": 600,
"decim": 10,
},
"LF": {
"foi": fois_l,
"cycles": fois_l * 0.4,
"time_bandwidth": 1 + 1,
"n_jobs": njobs,
"est_val": fois_l,
"est_key": "LF",
"sf": 600,
"decim": 10,
},
}
events = epochs.events[:, 2]
filters = lcmv.setup_filters(epochs.info, forward, data_cov, None, labels)
set_n_threads(1)
for i in range(0, len(events), chunks):
filename = lcmvfilename(
recording,
signal_type,
epoch,
chunk=i,
)
if os.path.isfile(filename):
continue
if signal_type == "BB":
logging.info("Starting reconstruction of BB signal")
M = lcmv.reconstruct_broadband(
filters,
epochs.info,
epochs._data[i:i + chunks],
events[i:i + chunks],
epochs.times,
njobs=1,
)
else:
logging.info("Starting reconstruction of TFR signal")
M = lcmv.reconstruct_tfr(
filters,
epochs.info,
epochs._data[i:i + chunks],
events[i:i + chunks],
epochs.times,
est_args=tfr_params[signal_type],
njobs=4,
)
M.to_hdf(str(filename), "epochs")
set_n_threads(njobs)