def setUpContainer(self):
self.timeseries = TimeSeries(name='dummy timeseries',
description='desc',
data=np.ones((3, 3)),
unit='flibs',
timestamps=np.ones((3, )))
bands = DynamicTable(name='bands',
description='band info for LFPSpectralAnalysis',
columns=[
VectorData(name='band_name',
description='name of bands',
data=['alpha', 'beta', 'gamma']),
VectorData(
name='band_limits',
description='low and high cutoffs in Hz',
data=np.ones((3, 2)))
])
spec_anal = DecompositionSeries(name='LFPSpectralAnalysis',
description='my description',
data=np.ones((3, 3, 3)),
timestamps=np.ones((3, )),
source_timeseries=self.timeseries,
metric='amplitude',
bands=bands)
return spec_anal
def test_init(self):
timeseries = TimeSeries(name='dummy timeseries',
description='desc',
data=np.ones((3, 3)),
unit='Volts',
timestamps=np.ones((3, )))
bands = DynamicTable(name='bands',
description='band info for LFPSpectralAnalysis',
columns=[
VectorData(name='band_name',
description='name of bands',
data=['alpha', 'beta', 'gamma']),
VectorData(
name='band_limits',
description='low and high cutoffs in Hz',
data=np.ones((3, 2)))
])
spec_anal = DecompositionSeries(name='LFPSpectralAnalysis',
description='my description',
data=np.ones((3, 3, 3)),
timestamps=np.ones((3, )),
source_timeseries=timeseries,
metric='amplitude',
bands=bands)
self.assertEqual(spec_anal.name, 'LFPSpectralAnalysis')
self.assertEqual(spec_anal.description, 'my description')
np.testing.assert_equal(spec_anal.data, np.ones((3, 3, 3)))
np.testing.assert_equal(spec_anal.timestamps, np.ones((3, )))
self.assertEqual(spec_anal.bands['band_name'].data,
['alpha', 'beta', 'gamma'])
np.testing.assert_equal(spec_anal.bands['band_limits'].data,
np.ones((3, 2)))
self.assertEqual(spec_anal.source_timeseries, timeseries)
self.assertEqual(spec_anal.metric, 'amplitude')
def test_constructor_ids_default(self):
columns = [
VectorData(name=s['name'], description=s['description'], data=d)
for s, d in zip(self.spec, self.data)
]
table = DynamicTable("with_spec", 'a test table', columns=columns)
self.check_table(table)
def test_nd_array_to_df(self):
data = np.array([[1, 1, 1], [2, 2, 2], [3, 3, 3]])
col = VectorData(name='name', description='desc', data=data)
df = DynamicTable('test', 'desc', np.arange(3, dtype='int'), (col, )).to_dataframe()
df2 = pd.DataFrame({'name': [x for x in data]},
index=pd.Index(name='id', data=[0, 1, 2]))
assert_frame_equal(df, df2)
def with_columns_and_data(self):
columns = [
VectorData(name=s['name'], description=s['description'], data=d)
for s, d in zip(self.spec, self.data)
]
return DynamicTable("with_columns_and_data",
'a test table',
columns=columns)
def test_constructor_ElementIdentifier_ids(self):
columns = [
VectorData(name=s['name'], description=s['description'], data=d)
for s, d in zip(self.spec, self.data)
]
ids = ElementIdentifiers('ids', [0, 1, 2, 3, 4])
table = DynamicTable("with_columns",
'a test table',
id=ids,
columns=columns)
self.check_table(table)
def test_constructor_ids_bad_ids(self):
columns = [
VectorData(name=s['name'], description=s['description'], data=d)
for s, d in zip(self.spec, self.data)
]
msg = "must provide same number of ids as length of columns"
with self.assertRaisesRegex(ValueError, msg):
DynamicTable("with_columns",
'a test table',
id=[0, 1],
columns=columns)
def create_ragged_array(name, values):
"""
:param values: list of lists
:return:
"""
vector_data = VectorData(
name, 'indicates which compartments the data refers to',
[item for sublist in values for item in sublist])
vector_index = VectorIndex(
name + '_index', np.cumsum([len(x) for x in values]), target=vector_data)
return vector_data, vector_index
def spectral_decomposition(block_path, bands_vals):
"""
Takes preprocessed LFP data and does the standard Hilbert transform on
different bands. Takes about 20 minutes to run on 1 10-min block.
Parameters
----------
block_path : str
subject file path
bands_vals : [2,nBands] numpy array with Gaussian filter parameters, where:
bands_vals[0,:] = filter centers [Hz]
bands_vals[1,:] = filter sigmas [Hz]
Returns
-------
Saves spectral power (DecompositionSeries) in the current NWB file.
Only if container for this data do not exist in the file.
"""
# Get filter parameters
band_param_0 = bands_vals[0, :]
band_param_1 = bands_vals[1, :]
with NWBHDF5IO(block_path, 'r+', load_namespaces=True) as io:
nwb = io.read()
lfp = nwb.processing['ecephys'].data_interfaces[
'LFP'].electrical_series['preprocessed']
rate = lfp.rate
nBands = len(band_param_0)
nSamples = lfp.data.shape[0]
nChannels = lfp.data.shape[1]
Xp = np.zeros(
(nBands, nChannels, nSamples)) #power (nBands,nChannels,nSamples)
# Apply Hilbert transform ----------------------------------------------
print('Running Spectral Decomposition...')
start = time.time()
for ch in np.arange(nChannels):
Xch = lfp.data[:,
ch] * 1e6 # 1e6 scaling helps with numerical accuracy
Xch = Xch.reshape(1, -1)
Xch = Xch.astype('float32') # signal (nChannels,nSamples)
X_fft_h = None
for ii, (bp0, bp1) in enumerate(zip(band_param_0, band_param_1)):
kernel = gaussian(Xch, rate, bp0, bp1)
X_analytic, X_fft_h = hilbert_transform(Xch,
rate,
kernel,
phase=None,
X_fft_h=X_fft_h)
Xp[ii, ch, :] = abs(X_analytic).astype('float32')
print('Spectral Decomposition finished in {} seconds'.format(
time.time() - start))
# data: (ndarray) dims: num_times * num_channels * num_bands
Xp = np.swapaxes(Xp, 0, 2)
# Spectral band power
# bands: (DynamicTable) frequency bands that signal was decomposed into
band_param_0V = VectorData(
name='filter_param_0',
description='frequencies for bandpass filters',
data=band_param_0)
band_param_1V = VectorData(
name='filter_param_1',
description='frequencies for bandpass filters',
data=band_param_1)
bandsTable = DynamicTable(
name='bands',
description='Series of filters used for Hilbert transform.',
columns=[band_param_0V, band_param_1V],
colnames=['filter_param_0', 'filter_param_1'])
decs = DecompositionSeries(
name='DecompositionSeries',
data=Xp,
description='Analytic amplitude estimated with Hilbert transform.',
metric='amplitude',
unit='V',
bands=bandsTable,
rate=rate,
source_timeseries=lfp)
# Storage of spectral decomposition on NWB file ------------------------
ecephys_module = nwb.processing['ecephys']
ecephys_module.add_data_interface(decs)
io.write(nwb)
print('Spectral decomposition saved in ' + block_path)
def with_table_columns(self):
cols = [VectorData(**d) for d in self.spec]
table = DynamicTable("with_table_columns",
'a test table',
columns=cols)
return table
data=['A1', 'A2', 'A3'],
description='String with a recording tag')
#####################################################################
# The :py:class:`~pynwb.icephys.IntracellularRecordingsTable` table is not just a ``DynamicTable``
# but an ``AlignedDynamicTable``. The ``AlignedDynamicTable`` type is itself a ``DynamicTable``
# that may contain an arbitrary number of additional ``DynamicTable``, each of which defines
# a "category". This is similar to a table with "sub-headings". In the case of the
# :py:class:`~pynwb.icephys.IntracellularRecordingsTable`, we have three predefined categories,
# i.e., electrodes, stimuli, and responses. We can also dynamically add new categories to
# the table. As each category corresponds to a ``DynamicTable``, this means we have to create a
# new ``DynamicTable`` and add it to our table.
# Create a new DynamicTable for our category that contains a location column of type VectorData
location_column = VectorData(name='location',
data=['Mordor', 'Gondor', 'Rohan'],
description='Recording location in Middle Earth')
lab_category = DynamicTable(
name='recording_lab_data',
description='category table for lab-specific recording metadata',
colnames=[
'location',
],
columns=[
location_column,
])
# Add the table as a new category to our intracellular_recordings
nwbfile.intracellular_recordings.add_category(category=lab_category)
# Note, the name of the category is name of the table, i.e., 'recording_lab_data'
def get_bipolar_referenced_electrodes(X,
electrodes,
rate,
grid_size=None,
grid_step=1):
'''
Bipolar referencing of electrodes according to the scheme of Dr. John Burke
Each electrode (with obvious exceptions at the edges) yields two bipolar-
referenced channels, one for the "right" and one for the "below" neighbors.
These are returned as an ElectricalSeries.
Input arguments:
--------
X:
numpy array containing (raw) electrode traces (Nelectrodes x T)
electrodes:
DynamicTableRegion containing the metadata for the electrodes whose
traces are in X
rate:
sampling rate of X; for storage in ElectricalSeries
grid_size:
numpy array with the two dimensions of the grid (2, )
Returns:
--------
bipolarElectrodes:
ElectricalSeries containing the bipolar-referenced (pseudo) electrodes
and associated metadata. (NB that a, e.g., 16x16 grid yields 960 of
these pseudo-electrodes.)
'''
# set mutable default argument(s)
if grid_size is None:
grid_size = np.array([16, 16])
# malloc
elec_layout = np.arange(np.prod(grid_size) - 1, -1,
-1).reshape(grid_size).T
elec_layout = elec_layout[::grid_step, ::grid_step]
grid_size = elec_layout.T.shape # in case grid_step > 1
Nchannels = 2 * np.prod(grid_size) - np.sum(grid_size)
XX = np.zeros((Nchannels, X.shape[1]))
# create a new dynamic table to hold the metadata
column_names = ['x', 'y', 'z', 'imp', 'location', 'label', 'bad']
columns = [
VectorData(name=name, description=electrodes.table[name].description)
for name in column_names
]
bipolarTable = DynamicTable(
name='bipolar-referenced metadata',
description=('pseudo-channels derived via John Burke style'
' bipolar referencing'),
colnames=column_names,
columns=columns,
)
# compute bipolar-ref'd channel and add a new row of metadata to table
def add_new_channel(iChannel, iElectrode, jElectrode):
jElectrode = int(jElectrode)
# "bipolar referencing": the difference of neighboring electrodes
XX[iChannel, :] = X[iElectrode, :] - X[jElectrode, :]
# add a row to the table for this new pseudo-electrode
bipolarTable.add_row({
'location':
'_'.join({
electrodes.table['location'][iElectrode],
electrodes.table['location'][jElectrode]
}),
'label':
'-'.join([
electrodes.table['label'][iElectrode],
electrodes.table['label'][jElectrode]
]),
'bad': (electrodes.table['bad'][iElectrode]
or electrodes.table['bad'][jElectrode]),
**{
name: electrodes.table[name][iElectrode]
for name in ['x', 'y', 'z', 'imp']
},
})
return iChannel + 1
iChannel = 0
# loop across columns and rows (remembering that grid is transposed)
for i in range(grid_size[1]):
for j in range(grid_size[0]):
if j < grid_size[0] - 1:
iChannel = add_new_channel(iChannel, elec_layout[i, j],
elec_layout[i, j + 1])
if i < grid_size[1] - 1:
iChannel = add_new_channel(iChannel, elec_layout[i, j],
elec_layout[i + 1, j])
# create one big region for the entire table
bipolarTableRegion = bipolarTable.create_region(
'electrodes', [i for i in range(Nchannels)], 'all bipolar electrodes')
return XX, bipolarTable, bipolarTableRegion
def transform(block_path, filter='default', bands_vals=None):
"""
Takes raw LFP data and does the standard Hilbert algorithm:
1) CAR
2) notch filters
3) Hilbert transform on different bands
Takes about 20 minutes to run on 1 10-min block.
Parameters
----------
block_path : str
subject file path
filter: str, optional
Frequency bands to filter the signal.
'default' for Chang lab default values (Gaussian filters)
'custom' for user defined (Gaussian filters)
bands_vals: 2D array, necessary only if filter='custom'
[2,nBands] numpy array with Gaussian filter parameters, where:
bands_vals[0,:] = filter centers [Hz]
bands_vals[1,:] = filter sigmas [Hz]
Returns
-------
Saves preprocessed signals (LFP) and spectral power (DecompositionSeries) in
the current NWB file. Only if containers for these data do not exist in the
file.
"""
write_file = 1
rate = 400.
# Define filter parameters
if filter == 'default':
band_param_0 = bands.chang_lab['cfs']
band_param_1 = bands.chang_lab['sds']
elif filter == 'high_gamma':
band_param_0 = bands.chang_lab['cfs'][(bands.chang_lab['cfs'] > 70)
& (bands.chang_lab['cfs'] < 150)]
band_param_1 = bands.chang_lab['sds'][(bands.chang_lab['cfs'] > 70)
& (bands.chang_lab['cfs'] < 150)]
#band_param_0 = [ bands.neuro['min_freqs'][-1] ] #for hamming window filter
#band_param_1 = [ bands.neuro['max_freqs'][-1] ]
#band_param_0 = bands.chang_lab['cfs'][29:37] #for average of gaussian filters
#band_param_1 = bands.chang_lab['sds'][29:37]
elif filter == 'custom':
band_param_0 = bands_vals[0, :]
band_param_1 = bands_vals[1, :]
block_name = os.path.splitext(block_path)[0]
start = time.time()
with NWBHDF5IO(block_path, 'a') as io:
nwb = io.read()
# Storage of processed signals on NWB file -----------------------------
if 'ecephys' not in nwb.modules:
# Add module to NWB file
nwb.create_processing_module(
name='ecephys',
description='Extracellular electrophysiology data.')
ecephys_module = nwb.modules['ecephys']
# LFP: Downsampled and power line signal removed
if 'LFP' in nwb.modules['ecephys'].data_interfaces:
lfp_ts = nwb.modules['ecephys'].data_interfaces[
'LFP'].electrical_series['preprocessed']
X = lfp_ts.data[:].T
rate = lfp_ts.rate
else:
# 1e6 scaling helps with numerical accuracy
X = nwb.acquisition['ECoG'].data[:].T * 1e6
fs = nwb.acquisition['ECoG'].rate
bad_elects = load_bad_electrodes(nwb)
print('Load time for h5 {}: {} seconds'.format(
block_name,
time.time() - start))
print('rates {}: {} {}'.format(block_name, rate, fs))
if not np.allclose(rate, fs):
assert rate < fs
start = time.time()
X = resample(X, rate, fs)
print('resample time for {}: {} seconds'.format(
block_name,
time.time() - start))
if bad_elects.sum() > 0:
X[bad_elects] = np.nan
# Subtract CAR
start = time.time()
X = subtract_CAR(X)
print('CAR subtract time for {}: {} seconds'.format(
block_name,
time.time() - start))
# Apply Notch filters
start = time.time()
X = linenoise_notch(X, rate)
print('Notch filter time for {}: {} seconds'.format(
block_name,
time.time() - start))
lfp = LFP()
# Add preprocessed downsampled signals as an electrical_series
lfp_ts = lfp.create_electrical_series(
name='preprocessed',
data=X.T,
electrodes=nwb.acquisition['ECoG'].electrodes,
rate=rate,
description='')
ecephys_module.add_data_interface(lfp)
# Spectral band power
if 'Bandpower_' + filter not in nwb.modules['ecephys'].data_interfaces:
# Apply Hilbert transform
X = X.astype('float32') # signal (nChannels,nSamples)
nChannels = X.shape[0]
nSamples = X.shape[1]
nBands = len(band_param_0)
Xp = np.zeros((nBands, nChannels,
nSamples)) # power (nBands,nChannels,nSamples)
X_fft_h = None
for ii, (bp0, bp1) in enumerate(zip(band_param_0, band_param_1)):
# if filter=='high_gamma':
# kernel = hamming(X, rate, bp0, bp1)
# else:
kernel = gaussian(X, rate, bp0, bp1)
X_analytic, X_fft_h = hilbert_transform(X,
rate,
kernel,
phase=None,
X_fft_h=X_fft_h)
Xp[ii] = abs(X_analytic).astype('float32')
# Scales signals back to Volt
X /= 1e6
band_param_0V = VectorData(
name='filter_param_0',
description='frequencies for bandpass filters',
data=band_param_0)
band_param_1V = VectorData(
name='filter_param_1',
description='frequencies for bandpass filters',
data=band_param_1)
bandsTable = DynamicTable(
name='bands',
description='Series of filters used for Hilbert transform.',
columns=[band_param_0V, band_param_1V],
colnames=['filter_param_0', 'filter_param_1'])
# data: (ndarray) dims: num_times * num_channels * num_bands
Xp = np.swapaxes(Xp, 0, 2)
decs = DecompositionSeries(
name='Bandpower_' + filter,
data=Xp,
description='Band power estimated with Hilbert transform.',
metric='power',
unit='V**2/Hz',
bands=bandsTable,
rate=rate,
source_timeseries=lfp_ts)
ecephys_module.add_data_interface(decs)
io.write(nwb)
print('done', flush=True)
def copy_obj(obj_old, nwb_old, nwb_new):
""" Creates a copy of obj_old. """
obj = None
obj_type = type(obj_old).__name__
#ElectricalSeries ----------------------------------------------------------
if obj_type == 'ElectricalSeries':
nChannels = obj_old.electrodes.table['x'].data.shape[0]
elecs_region = nwb_new.electrodes.create_region(
name='electrodes',
region=np.arange(nChannels).tolist(),
description='')
obj = ElectricalSeries(name=obj_old.name,
data=obj_old.data[:],
electrodes=elecs_region,
rate=obj_old.rate,
description=obj_old.description)
#LFP -----------------------------------------------------------------------
if obj_type == 'LFP':
obj = LFP(name=obj_old.name)
els_name = list(obj_old.electrical_series.keys())[0]
els = obj_old.electrical_series[els_name]
nChannels = els.data.shape[1]
elecs_region = nwb_new.electrodes.create_region(
name='electrodes',
region=np.arange(nChannels).tolist(),
description='')
obj_ts = obj.create_electrical_series(name=els.name,
comments=els.comments,
conversion=els.conversion,
data=els.data[:],
description=els.description,
electrodes=elecs_region,
rate=els.rate,
resolution=els.resolution,
starting_time=els.starting_time)
#TimeSeries ----------------------------------------------------------------
elif obj_type == 'TimeSeries':
obj = TimeSeries(name=obj_old.name,
description=obj_old.description,
data=obj_old.data[:],
rate=obj_old.rate,
resolution=obj_old.resolution,
conversion=obj_old.conversion,
starting_time=obj_old.starting_time,
unit=obj_old.unit)
#DecompositionSeries -------------------------------------------------------
elif obj_type == 'DecompositionSeries':
list_columns = []
for item in obj_old.bands.columns:
bp = VectorData(name=item.name,
description=item.description,
data=item.data[:])
list_columns.append(bp)
bandsTable = DynamicTable(name=obj_old.bands.name,
description=obj_old.bands.description,
columns=list_columns,
colnames=obj_old.bands.colnames)
obj = DecompositionSeries(
name=obj_old.name,
data=obj_old.data[:],
description=obj_old.description,
metric=obj_old.metric,
unit=obj_old.unit,
rate=obj_old.rate,
#source_timeseries=lfp,
bands=bandsTable,
)
#Spectrum ------------------------------------------------------------------
elif obj_type == 'Spectrum':
file_elecs = nwb_new.electrodes
nChannels = len(file_elecs['x'].data[:])
elecs_region = file_elecs.create_region(
name='electrodes',
region=np.arange(nChannels).tolist(),
description='')
obj = Spectrum(name=obj_old.name,
frequencies=obj_old.frequencies[:],
power=obj_old.power,
electrodes=elecs_region)
return obj
