def extract_filter(subject,
session,
recording,
epoch,
signal_type='BB',
BEM='three_layer',
debug=False,
chunks=50,
njobs=4):
mne.set_log_level('WARNING')
lcmv.logging.getLogger().setLevel(logging.INFO)
logging.getLogger().setLevel(logging.INFO)
set_n_threads(1)
subject_int = int(subject[1:])
print('reading stimulus data')
MRI_subjects = ["S1", "S5", "S6", "S8", "S11", "S12", "S16", "S17"]
if subject in MRI_subjects:
subject2 = subject
else:
subject2 = "fsaverage"
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)
fname = '/storage/genis/preprocessed_megdata/filenames_sub%i.pickle' % (
subject_int)
f = open(fname, 'rb')
data = pickle.load(f)
df = pd.DataFrame.from_dict(data)
raw_filename = df[df.subject == subject_int][df.session == session][
df.trans_matrix == recording].filename.iloc[0]
print('hola')
# trans_filename = '/home/gprat/cluster_home/pymeg/S%i-sess%i-%i-trans.fif' % (subject_int, session, recording)
trans_filename = '/home/genis/pymeg/S%i-sess%i-%i-trans.fif' % (
subject_int, session, recording)
print('data_cov_done')
# raw_filename = glob('TODO' % (subject, session, recording))
# trans_filename = glob('TODO' % (subject, session, recording))[0]
logging.info('Setting up source space and forward model')
epo_filename = get_filename_trans_matrix(subject, session, recording)
forward, bem, source = sr.get_leadfield(subject2,
raw_filename,
epo_filename,
trans_filename,
bem_sub_path='bem_ft')
labels = sr.get_labels(subject2)
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'],
)
filters = lcmv.setup_filters(epochs.info, forward, data_cov, None, labels)
subject_int = int(subject[1:])
fname = 'filter_sub%i_SESS%i_recording%i_epoch%s.pickle' % (
subject_int, session, recording, epoch)
filename = join(path, fname)
f = open(filename, 'wb')
pickle.dump(filters, f)
print('filter_done')
f.close()
return filters
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 do_source_recon(subj, session, njobs=4):
epochs_filename_stim = os.path.join(data_folder, "epochs", subj, session,
'{}-epo.fif.gz'.format('stimlock'))
epochs_filename_resp = os.path.join(data_folder, "epochs", subj, session,
'{}-epo.fif.gz'.format('resplock'))
trans_filename = os.path.join(data_folder, "transformation_matrix",
'{}_{}-trans.fif'.format(subj, session))
if os.path.isfile(epochs_filename_stim):
runs = sorted([
run.split('/')[-1] for run in glob.glob(
os.path.join(data_folder, "raw", subj, session, "meg", "*.ds"))
])
center = int(np.floor(len(runs) / 2.0))
raw_filename = os.path.join(data_folder, "raw", subj, session, "meg",
runs[center])
# # make transformation matrix:
# sr.make_trans(subj, raw_filename, epochs_filename, trans_filename)
# load labels:
labels = sr.get_labels(subject=subj,
filters=['*wang*.label', '*JWG*.label'],
annotations=['HCPMMP1'])
labels = sr.labels_exclude(labels=labels,
exclude_filters=[
'wang2015atlas.IPS4',
'wang2015atlas.IPS5',
'wang2015atlas.SPL', 'JWG_lat_Unknown'
])
labels = sr.labels_remove_overlap(labels=labels,
priority_filters=['wang', 'JWG'])
print(labels)
# load epochs:
epochs_stim = mne.read_epochs(epochs_filename_stim)
epochs_stim = epochs_stim.pick_channels(
[x for x in epochs_stim.ch_names if x.startswith('M')])
epochs_resp = mne.read_epochs(epochs_filename_resp)
epochs_resp = epochs_resp.pick_channels(
[x for x in epochs_resp.ch_names if x.startswith('M')])
# baseline stuff:
overlap = list(
set(epochs_stim.events[:, 2]).intersection(
set(epochs_resp.events[:, 2])))
epochs_stim = epochs_stim[[str(l) for l in overlap]]
epochs_resp = epochs_resp[[str(l) for l in overlap]]
id_time = (-0.3 <= epochs_stim.times) & (epochs_stim.times <= -0.2)
means = epochs_stim._data[:, :, id_time].mean(-1)
epochs_stim._data = epochs_stim._data - means[:, :, np.newaxis]
epochs_resp._data = epochs_resp._data - means[:, :, np.newaxis]
# TFR settings:
fois_h = np.arange(42, 162, 4)
fois_l = np.arange(2, 42, 2)
tfr_params = {
'HF': {
'foi': fois_h,
'cycles': fois_h * 0.4,
'time_bandwidth': 5 + 1,
'n_jobs': njobs,
'est_val': fois_h,
'est_key': 'HF'
},
'LF': {
'foi': fois_l,
'cycles': fois_l * 0.4,
'time_bandwidth': 1 + 1,
'n_jobs': njobs,
'est_val': fois_l,
'est_key': 'LF'
}
}
# get cov:
data_cov = lcmv.get_cov(epochs_stim, tmin=0, tmax=1)
noise_cov = None
# get lead field:
forward, bem, source = sr.get_leadfield(
subject=subj,
raw_filename=raw_filename,
epochs_filename=epochs_filename_stim,
trans_filename=trans_filename,
conductivity=(0.3, 0.006, 0.3),
njobs=njobs)
# do source level analysis:
for tl, epochs in zip(['stimlock', 'resplock'],
[epochs_stim, epochs_resp]):
for signal_type in ['LF', 'HF']:
print(signal_type)
# events:
events = epochs.events[:, 2]
data = []
filters = lcmv.setup_filters(epochs.info,
forward,
data_cov,
None,
labels,
njobs=njobs)
# in chunks:
chunks = 100
for i in range(0, len(events), chunks):
filename = os.path.join(data_folder, "source_level", 'lcmv_{}_{}_{}_{}_{}-source.hdf'.\
format(subj, session, tl, signal_type, i))
# if os.path.isfile(filename):
# continue
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=njobs)
M.to_hdf(filename, 'epochs')
del M
def decode(
subject,
area,
epoch_type="stimulus",
only_glasser=False,
BEM="three_layer",
debug=False,
target="response",
):
mne.set_log_level("WARNING")
pymeglcmv.logging.getLogger().setLevel(logging.INFO)
set_n_threads(1)
labels = get_labels(subject, only_glasser)
labels = [x for x in labels if any([cl for cl in areas_to_labels[area] if cl in x.name])]
print(labels)
if len(labels) < 1:
raise RuntimeError('Expecting at least two labels')
label = labels.pop()
for l in labels:
label += l
#label = labels[0] + labels[1]
print('Selecting this label for area %s:'%area, label)
#return
logging.info("Reading stimulus data")
if epoch_type == "stimulus":
# data_cov, epochs = get_stim_epoch(subject, session)
data = [get_stim_epoch(subject, i) for i in range(4)]
elif epoch_type == "response":
# data_cov, epochs = get_response_epoch(subject, session)
data = [get_response_epoch(subject, i) for i in range(4)]
else:
raise RuntimeError("Did not recognize epoch")
logging.info("Setting up source space and forward model")
fwds = [get_leadfield(subject, session, BEM)[0] for session in range(4)]
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 = [d[1].events[:, 2] for d in data]
filters = []
for (data_cov, epochs), forward in zip(data, fwds):
filters.append(
pymeglcmv.setup_filters(epochs.info, forward, data_cov, None, [label])
)
set_n_threads(1)
F_tfrdata, events, F_freq, times = decoding.get_lcmv(
tfr_params["F"], [d[1] for d in data], filters, njobs=6
)
LF_tfrdata, events, LF_freq, times = decoding.get_lcmv(
tfr_params["LF"], [d[1] for d in data], filters, njobs=6
)
tfrdata = np.hstack([F_tfrdata, LF_tfrdata])
del LF_tfrdata, F_tfrdata
freq = np.concatenate([F_freq, LF_freq])
meta = augment_meta(
preprocessing.get_meta_for_subject(
subject, epoch_type, sessions=range(4)
).set_index("hash")
)
# Kick out trials with RT < 0.225
choice_rt = meta.choice_rt
valid_trials = choice_rt[choice_rt>=0.225].index.values
valid_trials = np.isin(events, valid_trials)
tfrdata = tfrdata[valid_trials]
events = events[valid_trials]
# How many kicked out?
n_out = (~valid_trials).sum()
n_all = len(events)
print('Kicking out %i/%i (%0.2f percent) trials due to RT'%(n_out, n_all, n_out/n_all))
all_s = []
for target in ["response", "unsigned_confidence", "signed_confidence"]:
fname = "/home/nwilming/S%i_trg%s_ROI_%s.hdf" % (subject, target, area)
try:
k = pd.read_hdf(fname)
except FileNotFoundError:
target_vals = meta.loc[:, target]
dcd = decoding.Decoder(target_vals)
k = dcd.classify(
tfrdata, times, freq, events, area,
average_vertices=False, use_phase=True
)
k.loc[:, "target"] = target
k.to_hdf(fname, "df")
all_s.append(k)
all_s = pd.concat(all_s)
all_s.loc[:, 'ROI'] = area
all_s.to_hdf("/home/nwilming/S%i_ROI_%s.hdf" % (subject, area), "df")
return k
def get_filter(
subject,
session,
epoch_type="stimulus",
only_glasser=False,
BEM="three_layer",
):
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)
return filters
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(
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)