def DLC_pos(filtered=True, force_reload=False, save=False):
"""
Load position from DLC files (*.h5) and returns it as a nts.TsdFrame
"""
files = os.listdir()
if ('positions.h5' in files) and (force_reload == False):
data = pd.read_hdf('positions.h5')
pos = nts.TsdFrame(data)
return pos
for f in files:
if filtered and f.endswith('filtered.h5'):
filename = f
break
if not filtered and not f.endswith('filtered.h5') and f.endswith(
'.h5'):
filename = f
break
data = pd.read_hdf(filename)
data = data[data.keys()[0][0]]
TTL = digitalin('digitalin.dat')[0, :]
tf = bk.compute.TTL_to_times(TTL)
if len(tf) > len(data):
tf = np.delete(tf, -1)
data.index = tf * 1_000_000
if save:
data.to_hdf('positions.h5', 'pos')
pos = nts.TsdFrame(data)
return pos
def getPeaksandTroughs(lfp, min_points):
"""
At 250Hz (1250/5), 2 troughs cannont be closer than 20 (min_points) points (if theta reaches 12Hz);
"""
import neuroseries as nts
import scipy.signal
if isinstance(lfp, nts.time_series.Tsd):
troughs = nts.Tsd(lfp.as_series().iloc[scipy.signal.argrelmin(
lfp.values, order=min_points)[0]],
time_units='us')
peaks = nts.Tsd(lfp.as_series().iloc[scipy.signal.argrelmax(
lfp.values, order=min_points)[0]],
time_units='us')
tmp = nts.Tsd(
troughs.realign(peaks, align='next').as_series().drop_duplicates(
'first')) # eliminate double peaks
peaks = peaks[tmp.index]
tmp = nts.Tsd(
peaks.realign(troughs, align='prev').as_series().drop_duplicates(
'first')) # eliminate double troughs
troughs = troughs[tmp.index]
return (peaks, troughs)
elif isinstance(lfp, nts.time_series.TsdFrame):
peaks = nts.TsdFrame(lfp.index.values, np.zeros(lfp.shape))
troughs = nts.TsdFrame(lfp.index.values, np.zeros(lfp.shape))
for i in lfp.keys():
peaks[i], troughs[i] = getPeaksandTroughs(lfp[i], min_points)
return (peaks, troughs)
def calc_Angular_velocity(decoded_angles): #Calculate angular_velocity as a function of time... ang_vector = [] for i in range(len(decoded_angles)): ang_vector.append(nts.TsdFrame((decoded_angles[i]).index.values[:-1],np.diff(decoded_angles[i]), time_units = 'us')) ts = [] for i in range(len(ang_vector)): ts.append((ang_vector[i]).index.values) ang_vec = [] for i in range(len(decoded_angles)): y1 = np.absolute(np.diff(decoded_angles[i])) ang_vec.append(nts.TsdFrame(np.arange(len(ts[i])), np.where(y1> np.pi, 2*np.pi - y1, y1)/np.diff(decoded_angles[i].index.values))) return ang_vec
def lfp(start,
stop,
n_channels=90,
channel=64,
frequency=1250.0,
precision='int16',
verbose=False):
p = session + ".lfp"
if verbose:
print('Load LFP from ' + p)
# From Guillaume viejo
import neuroseries as nts
bytes_size = 2
start_index = int(start * frequency * n_channels * bytes_size)
stop_index = int(stop * frequency * n_channels * bytes_size)
#In order not to read after the file
if stop_index > os.path.getsize(p): stop_index = os.path.getsize(p)
fp = np.memmap(p,
np.int16,
'r',
start_index,
shape=(stop_index - start_index) // bytes_size)
data = np.array(fp).reshape(len(fp) // n_channels, n_channels)
if type(channel) is not list:
timestep = np.arange(0, len(data)) / frequency + start
return nts.Tsd(timestep, data[:, channel], time_units='s')
elif type(channel) is list:
timestep = np.arange(0, len(data)) / frequency + start
return nts.TsdFrame(timestep, data[:, channel], time_units='s')
def speed(pos, value_gaussian_filter, columns_to_drop=None):
body = []
for i in pos:
body.append(i[0])
body = np.unique(body)
all_speed = np.empty((len(pos) - 1, 5))
i = 0
for b in body:
x_speed = np.diff(pos.as_units('s')[b]['x']) / np.diff(
pos.as_units('s').index)
y_speed = np.diff(pos.as_units('s')[b]['y']) / np.diff(
pos.as_units('s').index)
v = np.sqrt(x_speed**2 + y_speed**2)
all_speed[:, i] = v
i += 1
all_speed = scipy.ndimage.gaussian_filter1d(all_speed,
value_gaussian_filter,
axis=0)
all_speed = nts.TsdFrame(t=pos.index.values[:-1],
d=all_speed,
columns=body)
if columns_to_drop != None:
all_speed = all_speed.drop(columns=columns_to_drop)
return all_speed
def loadLFP(path,
n_channels=90,
channel=64,
frequency=1250.0,
precision='int16'):
import neuroseries as nts
if type(channel) is not list:
f = open(path, 'rb')
startoffile = f.seek(0, 0)
endoffile = f.seek(0, 2)
bytes_size = 2
n_samples = int((endoffile - startoffile) / n_channels / bytes_size)
duration = n_samples / frequency
interval = 1 / frequency
f.close()
with open(path, 'rb') as f:
data = np.fromfile(f, np.int16).reshape(
(n_samples, n_channels))[:, channel]
timestep = np.arange(0, len(data)) / frequency
return nts.Tsd(timestep, data, time_units='s')
elif type(channel) is list:
f = open(path, 'rb')
startoffile = f.seek(0, 0)
endoffile = f.seek(0, 2)
bytes_size = 2
n_samples = int((endoffile - startoffile) / n_channels / bytes_size)
duration = n_samples / frequency
f.close()
with open(path, 'rb') as f:
data = np.fromfile(f, np.int16).reshape(
(n_samples, n_channels))[:, channel]
timestep = np.arange(0, len(data)) / frequency
return nts.TsdFrame(timestep, data, time_units='s')
def processed_files(animal_id, date):
'input to the function must be a strings'
data_dir = 'D:/EphysData/Experiments/' + date + '/' + animal_id + '-' + date + '/' + animal_id + '-' + date + '/Analysis'
dir = 'D:/EphysData/Experiments/' + date + '/' + animal_id + '-' + date + '/' + animal_id + '-' + date
epochs = nts.IntervalSet(pd.read_hdf(data_dir + '/BehavEpochs.H5'))
position = pd.read_hdf(data_dir + '/' + 'Position.H5')
position = nts.TsdFrame(t=position.index.values,
d=position.values,
columns=position.columns,
time_units='s')
spikes, shank = loadSpikeData(
dir) #shank tells the number of cells on each shank
tcurv = {}
for i in range(len(epochs)):
tcurv[i] = computeAngularTuningCurves(
spikes, position['ry'],
nts.IntervalSet(epochs.loc[i, 'start'], epochs.loc[i, 'end']), 60)
'tuning curves are computed based on entire epoch, to restrict it, just modify the end time above'
#load tuning curve
#np.load(dir+'/'+animal_id+'.npy').item()
np.save(os.path.join(data_dir, animal_id), tcurv)
return spikes, epochs, position, tcurv
def lfp(
channel,
start=0,
stop=1e8,
fs=1250.0,
n_channels_local=None,
precision=np.int16,
dat=False,
verbose=False,
memmap=False,
p=None,
volt_step=0.195,
):
if (np.isnan(channel)) or (channel is None):
return None
if p is None:
p = session + ".lfp"
if dat:
p = session + ".dat"
if n_channels_local is None:
n_channels = xml()["nChannels"]
else:
n_channels = n_channels_local
if verbose:
print("Load data from " + p)
print(f"File contains {n_channels} channels")
# From Guillaume viejo
import neuroseries as nts
bytes_size = 2
start_index = int(start * fs * n_channels * bytes_size)
stop_index = int(stop * fs * n_channels * bytes_size)
# In order not to read after the file
if stop_index > os.path.getsize(p):
stop_index = os.path.getsize(p)
fp = np.memmap(p,
precision,
"r",
start_index,
shape=(stop_index - start_index) // bytes_size)
if memmap == True:
print("/!\ memmap is not compatible with volt_step /!\ ")
return fp.reshape(-1, n_channels)[:, channel]
data = np.array(fp).reshape(len(fp) // n_channels, n_channels) * volt_step
if type(channel) is not list:
timestep = np.arange(0, len(data)) / fs + start
return nts.Tsd(timestep, data[:, channel], time_units="s")
elif type(channel) is list:
timestep = np.arange(0, len(data)) / fs + start
return nts.TsdFrame(timestep, data[:, channel], time_units="s")
def getPhase(lfp, fmin, fmax, nbins, fsamp, power=False):
""" Continuous Wavelets Transform
return phase of lfp in a Tsd array
"""
import neuroseries as nts
from Wavelets import MyMorlet as Morlet
if isinstance(lfp, nts.time_series.TsdFrame):
allphase = nts.TsdFrame(lfp.index.values, np.zeros(lfp.shape))
allpwr = nts.TsdFrame(lfp.index.values, np.zeros(lfp.shape))
for i in lfp.keys():
allphase[i], allpwr[i] = getPhase(lfp[i],
fmin,
fmax,
nbins,
fsamp,
power=True)
if power:
return allphase, allpwr
else:
return allphase
elif isinstance(lfp, nts.time_series.Tsd):
cw = Morlet(lfp.values, fmin, fmax, nbins, fsamp)
cwt = cw.getdata()
cwt = np.flip(cwt, axis=0)
wave = np.abs(cwt)**2.0
phases = np.arctan2(np.imag(cwt), np.real(cwt)).transpose()
cwt = None
index = np.argmax(wave, 0)
# memory problem here, need to loop
phase = np.zeros(len(index))
for i in range(len(index)):
phase[i] = phases[i, index[i]]
phases = None
if power:
pwrs = cw.getpower()
pwr = np.zeros(len(index))
for i in range(len(index)):
pwr[i] = pwrs[index[i], i]
return nts.Tsd(lfp.index.values,
phase), nts.Tsd(lfp.index.values, pwr)
else:
return nts.Tsd(lfp.index.values, phase)
def load_continuous_tsd(paths,
t_min=None,
t_max=None,
downsample=None,
columns=None):
"""
read data for a specific time interval from a list of files (or ContinuousFile objects)
Args:
paths: a list of pathnames or ContinuousFile objects
t_min: the low end of the time interval to read
t_max: the high end of the time interval to read
downsample: if not None, it should be an integer and acts as a downsampling factor (useful to get e.g. LFP)
columns: a list of column names for the resulting TsdFrame. If None, the labels from the ContinuousFile objects
are used
Returns:
a TsdFrame with the data
"""
import scipy.signal as ss
if isinstance(paths, str):
paths = (paths, )
elif not is_sequence(paths):
raise TypeError("paths must be a string or list of strings.")
if isinstance(paths[0], str):
cf = [ContinuousFile(p) for p in paths]
else:
cf = paths
data, tstamps = cf[0].read_interval(t_min, t_max)
if downsample:
data = ss.decimate(data, downsample, zero_phase=True)
data = data.reshape((-1, 1))
columns_from_files = False
if columns is None:
columns = [cf[0].label]
columns_from_files = True
if isinstance(columns, tuple):
columns = list(columns)
for f in cf[1:]:
d, ts1 = f.read_interval(t_min, t_max)
assert len(ts1) == len(tstamps)
if downsample:
d = ss.decimate(d, downsample, zero_phase=True)
data = np.hstack((data, d.reshape((-1, 1))))
if columns_from_files:
columns.append(f.label)
if downsample:
tstamps = tstamps[::downsample]
data = data[:, :len(tstamps)]
cont_tsd = nts.TsdFrame(tstamps, data, columns=columns)
return cont_tsd
def downsample(tsd, up, down):
import scipy.signal
import neuroseries as nts
dtsd = scipy.signal.resample_poly(tsd.values, up, down)
dt = tsd.as_units('s').index.values[np.arange(0, tsd.shape[0], down)]
if len(tsd.shape) == 1:
return nts.Tsd(dt, dtsd, time_units='s')
elif len(tsd.shape) == 2:
return nts.TsdFrame(dt,
dtsd,
time_units='s',
columns=list(tsd.columns))
def pos(save=False):
# BK : 04/08/2020
# Return a NeuroSeries DataFrame of position whith the time as index
# session_path = get_session_path(session_name)
import csv
pos_clean = scipy.io.loadmat(path + "/posClean.mat")["posClean"]
# if save == True :
# with open('position'+'.csv', 'w') as csvfile:
# filewriter=csv.writer(csvfile)
return nts.TsdFrame(t=pos_clean[:, 0],
d=pos_clean[:, 1:],
columns=["x", "y"],
time_units="s")
def load_Positions(path, ep):
#Load the position HDF file as TsDFrame
#Epoch to be selected from all the recording time. Available options : 'wake', 'presleep', 'postsleep'
import os
import neuroseries as nts
if not os.path.exists(path):
print("The path "+path+" doesn't exist; Exiting ...")
sys.exit()
new_path = os.path.join(path, 'Analysis/')
if os.path.exists(new_path):
new_path = os.path.join(path, 'Analysis/')
store = pd.HDFStore(new_path + 'Positions.h5')
pos = store['positions']
try:
presl = (pos.loc[ep])
positions_vs_t = nts.TsdFrame(t = (presl['timestamp']).values, d = presl[['x','y','angle']].values, columns = presl[['x','y','angle']].columns, time_units = 's')
return positions_vs_t
except:
print("Specific epoch name didn't exist is the position data. Exiting.....")
sys.exit()
def loadLFP(path,
n_channels=90,
channel=64,
frequency=1250.0,
precision="int16"):
"""
LEGACY
"""
# From Guillaume Viejo
import neuroseries as nts
if type(channel) is not list:
f = open(path, "rb")
startoffile = f.seek(0, 0)
endoffile = f.seek(0, 2)
bytes_size = 2
n_samples = int((endoffile - startoffile) / n_channels / bytes_size)
duration = n_samples / frequency
interval = 1 / frequency
f.close()
with open(path, "rb") as f:
print("opening")
data = np.fromfile(f, np.int16).reshape(
(n_samples, n_channels))[:, channel]
timestep = np.arange(0, len(data)) / frequency
return nts.Tsd(timestep, data, time_units="s")
elif type(channel) is list:
f = open(path, "rb")
startoffile = f.seek(0, 0)
endoffile = f.seek(0, 2)
bytes_size = 2
n_samples = int((endoffile - startoffile) / n_channels / bytes_size)
duration = n_samples / frequency
f.close()
with open(path, "rb") as f:
data = np.fromfile(f, np.int16).reshape(
(n_samples, n_channels))[:, channel]
timestep = np.arange(0, len(data)) / frequency
return nts.TsdFrame(timestep, data, time_units="s")
def loadPosition(path,
events=None,
episodes=None,
n_ttl_channels=1,
optitrack_ch=None,
names=['ry', 'rx', 'rz', 'x', 'y', 'z'],
update_wake_epoch=True):
"""
load the position contained in /Analysis/Position.h5
Notes:
The order of the columns is assumed to be
['ry', 'rx', 'rz', 'x', 'y', 'z']
Args:
path: string
Returns:
neuroseries.TsdFrame
"""
if not os.path.exists(path): # Checking for path
print("The path " + path + " doesn't exist; Exiting ...")
sys.exit()
new_path = os.path.join(path, 'Analysis')
if not os.path.exists(new_path): os.mkdir(new_path)
file = os.path.join(path, 'Analysis', 'Position.h5')
if not os.path.exists(file):
makePositions(path, events, episodes, n_ttl_channels, optitrack_ch,
names, update_wake_epoch)
if os.path.exists(file):
store = pd.HDFStore(file, 'r')
position = store['position']
store.close()
position = nts.TsdFrame(t=position.index.values,
d=position.values,
columns=position.columns,
time_units='s')
return position
else:
print("Cannot find " + file + " for loading position")
sys.exit()
def det_pos(data_directory, ID, color, path2save):
#load the angular value at each time steps and make a nts frame of it
data = np.genfromtxt(data_directory + ID + '_PosHD.txt')
mouse_position = nts.TsdFrame(d=data[:, [1, 2, 3]],
t=data[:, 0],
time_units='s')
# But TsdFrame is a wrapper of pandas and you can change name of columns in pandas
# So let's change the columns name
mouse_position.columns = ['x', 'y', 'ang']
#Plot position
plt.figure()
plt.plot(mouse_position['x'].values, mouse_position['y'].values, color)
plt.xlabel("x position (cm)")
plt.ylabel("y position (cm)")
plt.title("Position of the mouse in the arena")
if path2save == 'a': plot_curve = './plots/' + 'position_' + '.pdf'
elif path2save == 'b':
plot_curve = r'cd /home/grvite/Dropbox (Peyrache Lab)/Peyrache Lab Team Folder/Projects/DreamSpeed - Gilberto/figs/' + 'position_' + '.pdf'
plt.savefig
plt.show()
return (mouse_position)
def loadLFP(path,
n_channels=90,
channel=64,
frequency=1250.0,
precision='int16'):
import neuroseries as nts
f = open(path, 'rb')
startoffile = f.seek(0, 0)
endoffile = f.seek(0, 2)
bytes_size = 2
n_samples = int((endoffile - startoffile) / n_channels / bytes_size)
duration = n_samples / frequency
interval = 1 / frequency
f.close()
fp = np.memmap(path, np.int16, 'r', shape=(n_samples, n_channels))
timestep = np.arange(0, n_samples) / frequency
if type(channel) is not list:
timestep = np.arange(0, n_samples) / frequency
return nts.Tsd(timestep, fp[:, channel], time_units='s')
elif type(channel) is list:
timestep = np.arange(0, n_samples) / frequency
return nts.TsdFrame(timestep, fp[:, channel], time_units='s')
def loadBunch_Of_LFP(path,
start,
stop,
n_channels=90,
channel=64,
frequency=1250.0,
precision='int16'):
import neuroseries as nts
bytes_size = 2
start_index = int(start * frequency * n_channels * bytes_size)
stop_index = int(stop * frequency * n_channels * bytes_size)
fp = np.memmap(path,
np.int16,
'r',
start_index,
shape=(stop_index - start_index) // bytes_size)
data = np.array(fp).reshape(len(fp) // n_channels, n_channels)
if type(channel) is not list:
timestep = np.arange(0, len(data)) / frequency
return nts.Tsd(timestep, data[:, channel], time_units='s')
elif type(channel) is list:
timestep = np.arange(0, len(data)) / frequency
return nts.TsdFrame(timestep, data[:, channel], time_units='s')
# pos_sws_ep = pos_sws_ep.intersect(nts.IntervalSet(pos_sws_ep['start'].iloc[0], pos_sws_ep['end'].iloc[0] + 60*60*1000*1000))
# if pre_sws_ep.tot_length()/1000/1000/60 > 30.0 and pos_sws_ep.tot_length()/1000/1000/60 > 30.0:
if pre_ep.tot_length() / 1000 / 1000 / 60 > 3.0 and post_ep.tot_length(
) / 1000 / 1000 / 60 > 3.0:
for hd in range(2):
index = np.where(hd_info_neuron == hd)[0]
allpop = all_pop[index].copy()
if allpop.shape[1] and allpop.shape[1] > 2:
eigen = compute_eigen(allpop)
###############################################################################################################
# SHANK LOOP
###############################################################################################################
for shank in np.unique(shankIndex):
index2 = np.where((shankIndex[index] == shank))[0]
prepop = nts.TsdFrame(pre_pop[index2].copy())
pospop = nts.TsdFrame(pos_pop[index2].copy())
pre_score = compute_score(prepop.copy(), eigen[:,
index2])
pos_score = compute_score(pospop.copy(), eigen[:,
index2])
prerip_score = compute_rip_score(
rip_tsd.restrict(pre_ep), pre_score, bins2)
posrip_score = compute_rip_score(
rip_tsd.restrict(post_ep), pos_score, bins2)
a = pd.DataFrame(index=prerip_score.index.values,
data=gaussFilt(
prerip_score.mean(1).values,
sws_ep = loadEpoch(data_directory, 'sws')
rem_ep = loadEpoch(data_directory, 'rem')
# Next step is to load the angular value at each time steps
# We need to load Mouse12-120806_PosHD.txt which is a text file
# We can use the function genfromtxt of numpy that load a simple text file
data = np.genfromtxt(data_directory + 'Mouse12-120806_PosHD.txt')
# Check your variable by typing
data
# It's an array, we can check the dimension by typing :
data.shape
# It has 40858 lines and 4 columns
# The columns are respectively [times | x position in the arena | y position in the arena | angular value of the head]
# So we can use the TsdFrame object of neuroseries as seen in main2.py
mouse_position = nts.TsdFrame(d=data[:, [1, 2, 3]],
t=data[:, 0],
time_units='s')
# Check your variable
mouse_position
# By defaut, TsdFrame does not take column name as input
mouse_position.columns
# But TsdFrame is a wrapper of pandas and you can change name of columns in pandas
# So let's change the columns name
mouse_position.columns = ['x', 'y', 'ang']
# It's good to always check the data by plotting them
# To see the position of the mouse in the arena during the session
# you can plot the x position versus the y position
import matplotlib.pyplot as plt
plt.figure()
plt.plot(mouse_position['x'].values, mouse_position['y'].values)
def compute_population_correlation(session):
# for session in sessions:
start_time = time.clock()
print(session)
store = pd.HDFStore("/mnt/DataGuillaume/population_activity_hd/" + session)
rip_pop = store['rip']
rem_pop = store['rem']
wak_pop = store['wake']
store.close()
###############################################################################################################
# POPULATION CORRELATION FOR EACH RIPPLES
###############################################################################################################
#matrix of distance between ripples in second
interval_mat = np.vstack(nts.TsdFrame(rip_pop).as_units(
's').index.values) - nts.TsdFrame(rip_pop).as_units('s').index.values
rip_corr = np.ones(interval_mat.shape) * np.nan
# doing the upper part of the diagonal
# rip_corr = np.eye(interval_mat.shape[0])
# bad
tmp = np.zeros_like(rip_corr)
tmp[np.triu_indices(interval_mat.shape[0], 1)] += 1
tmp[np.tril_indices(interval_mat.shape[0], 300)] += 1
index = np.where(tmp == 2)
for i, j in zip(index[0], index[1]):
rip_corr[i, j] = scipy.stats.pearsonr(rip_pop.iloc[i].values,
rip_pop.iloc[j].values)[0]
rip_corr[j, i] = rip_corr[i, j]
# print(rip_corr[i,j])
allrip_corr = pd.DataFrame(index=interval_mat[index], data=rip_corr[index])
rip_corr = pd.DataFrame(index=rip_pop.index.values,
data=rip_corr,
columns=rip_pop.index.values)
np.fill_diagonal(rip_corr.values, 1.0)
rip_corr = rip_corr.fillna(0)
###############################################################################################################
# POPULATION CORRELATION FOR EACH THETA CYCLE OF REM
###############################################################################################################
# compute all time interval for each ep of theta
interval_mat = np.vstack(nts.TsdFrame(rem_pop).as_units(
's').index.values) - nts.TsdFrame(rem_pop).as_units('s').index.values
rem_corr = np.ones(interval_mat.shape) * np.nan
# index = np.where(np.logical_and(interval_mat < 3.0, interval_mat >= 0.0))
# rem_corr = np.eye(interval_mat.shape[0])
# bad
tmp = np.zeros_like(rem_corr)
tmp[np.triu_indices(interval_mat.shape[0], 1)] += 1
tmp[np.tril_indices(interval_mat.shape[0], 300)] += 1
index = np.where(tmp == 2)
for i, j in zip(index[0], index[1]):
rem_corr[i, j] = scipy.stats.pearsonr(rem_pop.iloc[i].values,
rem_pop.iloc[j].values)[0]
rem_corr[j, i] = rem_corr[i, j]
allrem_corr = pd.DataFrame(index=interval_mat[index], data=rem_corr[index])
rem_corr = pd.DataFrame(index=rem_pop.index.values,
data=rem_corr,
columns=rem_pop.index.values)
np.fill_diagonal(rem_corr.values, 1.0)
rem_corr = rem_corr.fillna(0)
###############################################################################################################
# POPULATION CORRELATION FOR EACH THETA CYCLE OF WAKE
###############################################################################################################
# compute all time interval for each ep of theta
interval_mat = np.vstack(nts.TsdFrame(wak_pop).as_units(
's').index.values) - nts.TsdFrame(wak_pop).as_units('s').index.values
wak_corr = np.ones(interval_mat.shape) * np.nan
# index = np.where(np.logical_and(interval_mat < 3.0, interval_mat >= 0.0))
# wak_corr = np.eye(interval_mat.shape[0])
# bad
tmp = np.zeros_like(wak_corr)
tmp[np.triu_indices(interval_mat.shape[0], 1)] += 1
tmp[np.tril_indices(interval_mat.shape[0], 300)] += 1
index = np.where(tmp == 2)
for i, j in zip(index[0], index[1]):
wak_corr[i, j] = scipy.stats.pearsonr(wak_pop.iloc[i].values,
wak_pop.iloc[j].values)[0]
wak_corr[j, i] = wak_corr[i, j]
allwak_corr = pd.DataFrame(index=interval_mat[index], data=wak_corr[index])
wak_corr = pd.DataFrame(index=wak_pop.index.values,
data=wak_corr,
columns=wak_pop.index.values)
np.fill_diagonal(wak_corr.values, 1.0)
wak_corr = wak_corr.fillna(0)
###############################################################################################################
# STORING
###############################################################################################################
store = pd.HDFStore("/mnt/DataGuillaume/corr_pop_hd/" + session)
store.put('rip_corr', rip_corr)
store.put('allrip_corr', allrip_corr)
store.put('wak_corr', wak_corr)
store.put('allwak_corr', allwak_corr)
store.put('rem_corr', rem_corr)
store.put('allrem_corr', allrem_corr)
store.close()
print(time.clock() - start_time, "seconds")
return time.clock() - start_time
end = start + binsize
if ((start.values[0] >= first_spike)
and (last_spike >= end.values[0])):
spikes_in_interval = my_neuron.as_units(
's').loc[start.values[0]:end.values[0]]
firing_rate[i] = len(spikes_in_interval)
firing_rate = nts.Tsd(t=np.arange(rem_start, rem_end, binsize),
d=firing_rate,
time_units='s')
if j == HD_index[0]:
sleep_spikes = firing_rate
else:
sleep_spikes = np.vstack([sleep_spikes, firing_rate])
sleep_spikes = sleep_spikes.transpose()
time_bins = np.arange(rem_start, rem_end, binsize)
sleep_spikes = nts.TsdFrame(t=time_bins, d=sleep_spikes, time_units='s')
sleep_spikes.columns = HD_index
#Compute tuning curves for each neuron using the wake epoch
mouse_position = np.genfromtxt(data_directory + 'Mouse12-120806_PosHD.txt')
mouse_HD = nts.TsdFrame(t=mouse_position[:, 0],
d=mouse_position[:, 3],
time_units='s')
wake_ep = loadEpoch(data_directory, 'wake')
wake_start = wake_ep.as_units('s')['start'].values[0]
wake_end = wake_ep.as_units('s')['end'].values[0]
duration = wake_end - wake_start
duration = duration + 0.1
num_points = duration / binsize
num_points = int(num_points)
head_direction = np.zeros(num_points)
wake_ep = loadEpoch(data_directory, 'wake')
# The function will automaticaly search for the rigth file
# You can check your variables by typing them
# Next step is to load the angular value at each time steps
# We need to load Mouse12-120806_PosHD.txt which is a text file
# We can use the function genfromtxt of numpy that load a simple text file
data = np.genfromtxt('../data_matlab/Mouse12-120806/Mouse12-120806_PosHD.txt')
# Check your variable by typing it
# It's an array, we can check the dimension by typing :
data.shape
# It has 40858 lines and 4 columns
# The columns are respectively [times | x position in the arena | y position in the arena | angular value of the head]
# So we can use the TsdFrame object of neuroseries as seen in main3.py
mouse_position = nts.TsdFrame(d=data[:, [1, 2, 3]],
t=data[:, 0],
time_units='s',
columns=['x', 'y', 'ang'])
# It's good to always check the data by plotting them
# To see the position of the mouse in the arena during the session
# you can plot the x position versus the y position
figure()
plot(mouse_position['x'].values, mouse_position['y'].values)
xlabel("x position (cm)")
ylabel("y position (cm)")
show()
# Now we are going to compute the tuning curve for all neurons during exploration
# The process of making a tuning curve has been covered in main3_tuningcurves.py
# So here we are gonna define a function that will be looped over each HD neurons
# spikeshd = {k:spikes[k] for k in np.where(hd_info_neuron==1)[0] if k not in []}
# position = pd.read_csv(data_directory+session+"/"+session.split("/")[1] + ".csv", delimiter = ',', header = None, index_col = [0])
# angle = nts.Tsd(t = position.index.values, d = position[1].values, time_units = 's')
# tcurves = computeAngularTuningCurves(spikeshd, angle, wake_ep, nb_bins = 60, frequency = 1/0.0256)
# neurons = tcurves.idxmax().sort_values().index.values
####################################################################################################################
# POSITION X Y
####################################################################################################################
position = pd.read_csv(data_directory + session + "/" + session.split("/")[1] +
"_XY.csv",
delimiter=',',
header=None,
index_col=[0])
position = nts.TsdFrame(t=position.index.values,
d=position.values,
time_units='s')
spikesnohd = {
k: spikes[k]
for k in np.where(hd_info_neuron == 0)[0] if k not in []
}
placefield, extent = computePlaceFields(spikesnohd,
position,
wake_ep,
nb_bins=40,
frequency=1 / 0.0256)
####################################################################################################################
# PHASE SPIKE NO HD
session.split("/")[1] + ".csv",
delimiter=',',
header=None,
index_col=[0])
angle = nts.Tsd(t=position.index.values,
d=position[1].values,
time_units='s')
wakangle = pd.Series(index=np.arange(len(bins) - 1))
tmp = angle.groupby(
np.digitize(angle.as_units('ms').index.values, bins) - 1).mean()
wakangle.loc[tmp.index] = tmp
wakangle.index = data.index
wakangle = wakangle.interpolate(method='nearest')
data = nts.TsdFrame(t=data.index.values,
d=data.values,
time_units='ms')
abins = np.linspace(0, 2 * np.pi, 61)
wakangle = wakangle.dropna()
data = data.loc[wakangle.index]
index = np.digitize(wakangle.values, abins) - 1
a = data.groupby(index).mean()
data = data.restrict(theta_ep)
imap = Isomap(n_neighbors=100,
n_components=2).fit_transform(data.values[0:20000])
def compute_population_correlation(nuc, session):
start_time = time.clock()
print(session)
store = pd.HDFStore("/mnt/DataGuillaume/population_activity/"+session+".h5")
rip_pop = store['rip']
rem_pop = store['rem']
wak_pop = store['wake']
store.close()
# WHICH columns to keep
mappings = pd.read_hdf("/mnt/DataGuillaume/MergedData/MAPPING_NUCLEUS.h5")
tmp = mappings[mappings.index.str.contains(session)]['nucleus'] == nuc
neurons = tmp.index.values[np.where(tmp)[0]]
idx = np.array([int(n.split("_")[1]) for n in neurons])
rip_pop = rip_pop[idx]
rem_pop = rem_pop[idx]
wak_pop = wak_pop[idx]
###############################################################################################################
# POPULATION CORRELATION FOR EACH RIPPLES
###############################################################################################################
#matrix of distance between ripples in second
interval_mat = np.vstack(nts.TsdFrame(rip_pop).as_units('s').index.values) - nts.TsdFrame(rip_pop).as_units('s').index.values
rip_corr = np.ones(interval_mat.shape)*np.nan
# doing the upper part of the diagonal
# rip_corr = np.eye(interval_mat.shape[0])
# bad
tmp = np.zeros_like(rip_corr)
tmp[np.triu_indices(interval_mat.shape[0], 1)] += 1
tmp[np.tril_indices(interval_mat.shape[0], 300)] += 1
index = np.where(tmp == 2)
for i, j in zip(index[0], index[1]):
rip_corr[i,j] = scipy.stats.pearsonr(rip_pop.iloc[i].values, rip_pop.iloc[j].values)[0]
rip_corr[j,i] = rip_corr[i,j]
# print(rip_corr[i,j])
allrip_corr = pd.DataFrame(index = interval_mat[index], data = rip_corr[index])
rip_corr = pd.DataFrame(index = rip_pop.index.values, data = rip_corr, columns = rip_pop.index.values)
np.fill_diagonal(rip_corr.values, 1.0)
rip_corr = rip_corr.fillna(0)
###############################################################################################################
# POPULATION CORRELATION FOR EACH THETA CYCLE OF REM
###############################################################################################################
# compute all time interval for each ep of theta
interval_mat = np.vstack(nts.TsdFrame(rem_pop).as_units('s').index.values) - nts.TsdFrame(rem_pop).as_units('s').index.values
rem_corr = np.ones(interval_mat.shape)*np.nan
# index = np.where(np.logical_and(interval_mat < 3.0, interval_mat >= 0.0))
# rem_corr = np.eye(interval_mat.shape[0])
# bad
tmp = np.zeros_like(rem_corr)
tmp[np.triu_indices(interval_mat.shape[0], 1)] += 1
tmp[np.tril_indices(interval_mat.shape[0], 300)] += 1
index = np.where(tmp == 2)
for i, j in zip(index[0], index[1]):
rem_corr[i,j] = scipy.stats.pearsonr(rem_pop.iloc[i].values, rem_pop.iloc[j].values)[0]
rem_corr[j,i] = rem_corr[i,j]
allrem_corr = pd.DataFrame(index = interval_mat[index], data = rem_corr[index])
rem_corr = pd.DataFrame(index = rem_pop.index.values, data = rem_corr, columns = rem_pop.index.values)
np.fill_diagonal(rem_corr.values, 1.0)
rem_corr = rem_corr.fillna(0)
###############################################################################################################
# POPULATION CORRELATION FOR EACH THETA CYCLE OF WAKE
###############################################################################################################
# compute all time interval for each ep of theta
interval_mat = np.vstack(nts.TsdFrame(wak_pop).as_units('s').index.values) - nts.TsdFrame(wak_pop).as_units('s').index.values
wak_corr = np.ones(interval_mat.shape)*np.nan
# index = np.where(np.logical_and(interval_mat < 3.0, interval_mat >= 0.0))
# wak_corr = np.eye(interval_mat.shape[0])
# bad
tmp = np.zeros_like(wak_corr)
tmp[np.triu_indices(interval_mat.shape[0], 1)] += 1
tmp[np.tril_indices(interval_mat.shape[0], 300)] += 1
index = np.where(tmp == 2)
for i, j in zip(index[0], index[1]):
wak_corr[i,j] = scipy.stats.pearsonr(wak_pop.iloc[i].values, wak_pop.iloc[j].values)[0]
wak_corr[j,i] = wak_corr[i,j]
allwak_corr = pd.DataFrame(index = interval_mat[index], data = wak_corr[index])
wak_corr = pd.DataFrame(index = wak_pop.index.values, data = wak_corr, columns = wak_pop.index.values)
np.fill_diagonal(wak_corr.values, 1.0)
wak_corr = wak_corr.fillna(0)
###############################################################################################################
# STORING
###############################################################################################################
store = pd.HDFStore("/mnt/DataGuillaume/corr_pop_nucleus/"+nuc+"/"+session+".h5")
store.put('rip_corr', rip_corr)
store.put('allrip_corr', allrip_corr)
store.put('wak_corr', wak_corr)
store.put('allwak_corr', allwak_corr)
store.put('rem_corr', rem_corr)
store.put('allrem_corr', allrem_corr)
store.close()
print(time.clock() - start_time, "seconds")
return time.clock() - start_time
spikes = {
n: spikes[n]
for n in spikes.keys() if len(spikes[n].restrict(sws_ep))
}
n_neuron = len(spikes)
n_channel, fs, shank_to_channel = loadXML(data_directory + session + "/" +
session.split("/")[1] + '.xml')
# lfp_hpc = loadLFP(data_directory+session+"/"+session.split("/")[1]+'.eeg', n_channel, hpc_channel, float(fs), 'int16')
# lfp_hpc = downsample(lfp_hpc, 1, 5)
store = pd.HDFStore("../data/phase_spindles/" + session.split("/")[1] +
".lfp")
phase_hpc = nts.Tsd(store['phase_hpc_spindles'])
phase_thl = nts.Tsd(store['phase_thl_spindles'])
lfp_hpc = nts.Tsd(store['lfp_hpc'])
lfp_thl = nts.TsdFrame(store['lfp_thl'])
store.close()
##################################################################################################
# DETECTION UP/DOWN States
##################################################################################################
# print("up/down states")
# # bins of 5000 us
# bins = np.floor(np.arange(lfp_hpc.start_time(), lfp_hpc.end_time()+5000, 5000))
# total_value = nts.Tsd(bins[0:-1]+(bins[1]-bins[0])/2, np.zeros(len(bins)-1)).restrict(sws_ep)
# # each shank
# for s in shankStructure['thalamus']:
# neuron_index = np.where(shank == s)[0]
# if len(neuron_index):
# tmp = {i:spikes[i] for i in neuron_index}
# frate = getFiringRate(tmp, bins)
subplot(212)
plot(tuning_curve)
show()
# Ok it's ugly but who cares
# It's just random data
# This second example is about constructing a place field
# so this time, the spikes times are realigned to a 2d position
# Let's imagine the animal is in a circular environment this time
xpos = np.cos(angle.values) + np.random.randn(len(angle)) * 0.05
ypos = np.sin(angle.values) + np.random.randn(len(angle)) * 0.05
# We can stack the x,y position in a TsdFrame
position = np.vstack((xpos, ypos)).T
position = nts.TsdFrame(t=angle.times(), d=position, columns=['x', 'y'])
# and we can plot it
figure()
plot(position['x'], position['y'])
show()
# Now it's the same as before
# except the histogram is in 2d
position_spike = position.realign(spikes)
xbins = np.linspace(xpos.min(), xpos.max() + 0.01, 10)
ybins = np.linspace(ypos.min(), ypos.max() + 0.01, 10)
spike_count2, _, _ = np.histogram2d(position_spike['y'], position_spike['x'],
[ybins, xbins])
occupancy2, _, _ = np.histogram2d(position['y'], position['x'], [ybins, xbins])
spike_count2 = spike_count2 / (occupancy2 + 1)
# If you have timestamps associated with a value for example 15 points of EEG during 15seconds
my_eeg = np.sin(np.arange(0, 15))
# You use a Tsd (Time series data)
my_eeg = nts.Tsd(t=np.arange(15), d=my_eeg, time_units='s')
# Observe your variable
my_eeg
# And how the software transform you timetamps in second in timestamps in microsecond
# You can plot your data
plot(my_eeg, 'o-')
show()
# Now if you are using a fancy probe and recording for example 3 channel at the same times
# You use a TsdFrame
my_channels = np.random.rand(15, 3)
my_channels = nts.TsdFrame(t=np.arange(15), d=my_channels, time_units='s')
# You can plot your data
# It's always important to look at your data in the eyes
plot(my_channels, 'o-')
show()
# Yes it's random...
# If I want the data of my recording between second 5 to second 12
my_spike.as_units('s').loc[5:12]
my_eeg.as_units('s').loc[5:12]
my_channels.as_units('s').loc[5:12]
# Shoud be the same in millisecond
my_spike.as_units('ms').loc[5000:12000]
my_eeg.as_units('ms').loc[5000:12000]
my_channels.as_units('ms').loc[5000:12000]
# And in microseconds which is the default mode
def compute_pop_pca(session):
data_directory = '/mnt/DataGuillaume/MergedData/'
import numpy as np
import scipy.io
import scipy.stats
import _pickle as cPickle
import time
import os, sys
import neuroseries as nts
from functions import loadShankStructure, loadSpikeData, loadEpoch, loadThetaMod, loadSpeed, loadXML, loadRipples, loadLFP, downsample, getPeaksandTroughs, butter_bandpass_filter
import pandas as pd
# to know which neurons to keep
data_directory = '/mnt/DataGuillaume/MergedData/'
datasets = np.loadtxt(data_directory + 'datasets_ThalHpc.list',
delimiter='\n',
dtype=str,
comments='#')
theta_mod, theta_ses = loadThetaMod(
'/mnt/DataGuillaume/MergedData/THETA_THAL_mod.pickle',
datasets,
return_index=True)
theta = pd.DataFrame(index=theta_ses['rem'],
columns=['phase', 'pvalue', 'kappa'],
data=theta_mod['rem'])
tmp2 = theta.index[theta.isnull().any(1)].values
tmp3 = theta.index[(theta['pvalue'] > 0.01).values].values
tmp = np.unique(np.concatenate([tmp2, tmp3]))
theta_modth = theta.drop(tmp, axis=0)
neurons_index = theta_modth.index.values
bins1 = np.arange(-1005, 1010, 25) * 1000
times = np.floor(
((bins1[0:-1] + (bins1[1] - bins1[0]) / 2) / 1000)).astype('int')
premeanscore = {
i: {
'rem': pd.DataFrame(index=[], columns=['mean', 'std']),
'rip': pd.DataFrame(index=times, columns=[])
}
for i in range(3)
}
posmeanscore = {
i: {
'rem': pd.DataFrame(index=[], columns=['mean', 'std']),
'rip': pd.DataFrame(index=times, columns=[])
}
for i in range(3)
}
bins2 = np.arange(-1012.5, 1025, 25) * 1000
tsmax = {i: pd.DataFrame(columns=['pre', 'pos']) for i in range(3)}
# for session in datasets:
# for session in datasets[0:15]:
# for session in ['Mouse12/Mouse12-120815']:
start_time = time.clock()
print(session)
generalinfo = scipy.io.loadmat(data_directory + session +
'/Analysis/GeneralInfo.mat')
shankStructure = loadShankStructure(generalinfo)
if len(generalinfo['channelStructure'][0][0][1][0]) == 2:
hpc_channel = generalinfo['channelStructure'][0][0][1][0][1][0][0] - 1
else:
hpc_channel = generalinfo['channelStructure'][0][0][1][0][0][0][0] - 1
spikes, shank = loadSpikeData(
data_directory + session + '/Analysis/SpikeData.mat',
shankStructure['thalamus'])
wake_ep = loadEpoch(data_directory + session, 'wake')
sleep_ep = loadEpoch(data_directory + session, 'sleep')
sws_ep = loadEpoch(data_directory + session, 'sws')
rem_ep = loadEpoch(data_directory + session, 'rem')
sleep_ep = sleep_ep.merge_close_intervals(threshold=1.e3)
sws_ep = sleep_ep.intersect(sws_ep)
rem_ep = sleep_ep.intersect(rem_ep)
speed = loadSpeed(data_directory + session +
'/Analysis/linspeed.mat').restrict(wake_ep)
speed_ep = nts.IntervalSet(
speed[speed > 2.5].index.values[0:-1],
speed[speed > 2.5].index.values[1:]).drop_long_intervals(
26000).merge_close_intervals(50000)
wake_ep = wake_ep.intersect(speed_ep).drop_short_intervals(3000000)
n_channel, fs, shank_to_channel = loadXML(data_directory + session + "/" +
session.split("/")[1] + '.xml')
rip_ep, rip_tsd = loadRipples(data_directory + session)
hd_info = scipy.io.loadmat(data_directory + session +
'/Analysis/HDCells.mat')['hdCellStats'][:, -1]
hd_info_neuron = np.array([hd_info[n] for n in spikes.keys()])
all_neurons = np.array(list(spikes.keys()))
mod_neurons = np.array([
int(n.split("_")[1]) for n in neurons_index
if session.split("/")[1] in n
])
if len(sleep_ep) > 1:
store = pd.HDFStore("/mnt/DataGuillaume/population_activity_25ms/" +
session.split("/")[1] + ".h5")
# all_pop = store['allwake']
pre_pop = store['presleep']
pos_pop = store['postsleep']
store.close()
store = pd.HDFStore("/mnt/DataGuillaume/population_activity_100ms/" +
session.split("/")[1] + ".h5")
all_pop = store['allwake']
# pre_pop = store['presleep']
# pos_pop = store['postsleep']
store.close()
def compute_eigen(popwak):
popwak = popwak - popwak.mean(0)
popwak = popwak / (popwak.std(0) + 1e-8)
from sklearn.decomposition import PCA
pca = PCA(n_components=popwak.shape[1])
xy = pca.fit_transform(popwak.values)
pc = pca.explained_variance_ > (
1 + np.sqrt(1 / (popwak.shape[0] / popwak.shape[1])))**2.0
eigen = pca.components_[pc]
lambdaa = pca.explained_variance_[pc]
return eigen, lambdaa
def compute_score(ep_pop, eigen, lambdaa, thr):
ep_pop = ep_pop - ep_pop.mean(0)
ep_pop = ep_pop / (ep_pop.std(0) + 1e-8)
a = ep_pop.values
score = np.zeros(len(ep_pop))
for i in range(len(eigen)):
if lambdaa[i] >= thr:
score += (np.dot(a, eigen[i])**2.0 -
np.dot(a**2.0, eigen[i]**2.0))
score = nts.Tsd(t=ep_pop.index.values, d=score)
return score
def compute_rip_score(tsd, score, bins):
times = np.floor(
((bins[0:-1] + (bins[1] - bins[0]) / 2) / 1000)).astype('int')
rip_score = pd.DataFrame(index=times, columns=[])
for r, i in zip(tsd.index.values, range(len(tsd))):
xbins = (bins + r).astype('int')
y = score.groupby(
pd.cut(score.index.values, bins=xbins,
labels=times)).mean()
if ~y.isnull().any():
rip_score[r] = y
return rip_score
def get_xmin(ep, minutes):
duree = (ep['end'] - ep['start']) / 1000 / 1000 / 60
tmp = ep.iloc[np.where(np.ceil(duree.cumsum()) <= minutes + 1)[0]]
return nts.IntervalSet(tmp['start'], tmp['end'])
pre_ep = nts.IntervalSet(sleep_ep['start'][0], sleep_ep['end'][0])
post_ep = nts.IntervalSet(sleep_ep['start'][1], sleep_ep['end'][1])
pre_sws_ep = sws_ep.intersect(pre_ep)
pos_sws_ep = sws_ep.intersect(post_ep)
pre_sws_ep = get_xmin(pre_sws_ep.iloc[::-1], 30)
pos_sws_ep = get_xmin(pos_sws_ep, 30)
if pre_sws_ep.tot_length('s') / 60 > 5.0 and pos_sws_ep.tot_length(
's') / 60 > 5.0:
for hd in range(3):
if hd == 0 or hd == 2:
index = np.where(hd_info_neuron == 0)[0]
elif hd == 1:
index = np.where(hd_info_neuron == 1)[0]
if hd == 0:
index = np.intersect1d(index, mod_neurons)
elif hd == 2:
index = np.intersect1d(
index, np.setdiff1d(all_neurons, mod_neurons))
allpop = all_pop[index].copy()
prepop = nts.TsdFrame(pre_pop[index].copy())
pospop = nts.TsdFrame(pos_pop[index].copy())
# prepop25ms = nts.TsdFrame(pre_pop_25ms[index].copy())
# pospop25ms = nts.TsdFrame(pos_pop_25ms[index].copy())
if allpop.shape[1] and allpop.shape[1] > 5:
eigen, lambdaa = compute_eigen(allpop)
seuil = 1.2
if np.sum(lambdaa > seuil):
pre_score = compute_score(prepop, eigen, lambdaa,
seuil)
pos_score = compute_score(pospop, eigen, lambdaa,
seuil)
prerip_score = compute_rip_score(
rip_tsd.restrict(pre_sws_ep), pre_score, bins1)
posrip_score = compute_rip_score(
rip_tsd.restrict(pos_sws_ep), pos_score, bins1)
# pre_score_25ms = compute_score(prepop25ms, eigen)
# pos_score_25ms = compute_score(pospop25ms, eigen)
# prerip25ms_score = compute_rip_score(rip_tsd.restrict(pre_ep), pre_score_25ms, bins2)
# posrip25ms_score = compute_rip_score(rip_tsd.restrict(post_ep), pos_score_25ms, bins2)
# prerip25ms_score = prerip25ms_score - prerip25ms_score.mean(0)
# posrip25ms_score = posrip25ms_score - posrip25ms_score.mean(0)
# prerip25ms_score = prerip25ms_score / prerip25ms_score.std(0)
# posrip25ms_score = posrip25ms_score / posrip25ms_score.std(0)
# prerip25ms_score = prerip25ms_score.loc[-500:500]
# posrip25ms_score = posrip25ms_score.loc[-500:500]
# sys.exit()
# tmp = pd.concat([pd.DataFrame(prerip25ms_score.idxmax().values, columns = ['pre']),pd.DataFrame(posrip25ms_score.idxmax().values, columns = ['pos'])],axis = 1)
# tmp = pd.DataFrame(data = [[prerip25ms_score.mean(1).idxmax(), posrip25ms_score.mean(1).idxmax()]], columns = ['pre', 'pos'])
# tsmax[hd] = tsmax[hd].append(tmp, ignore_index = True)
premeanscore[hd]['rip'][session] = prerip_score.mean(1)
posmeanscore[hd]['rip'][session] = posrip_score.mean(1)
# if len(rem_ep.intersect(pre_ep)) and len(rem_ep.intersect(post_ep)):
# premeanscore[hd]['rem'].loc[session,'mean'] = pre_score.restrict(rem_ep.intersect(pre_ep)).mean()
# posmeanscore[hd]['rem'].loc[session,'mean'] = pos_score.restrict(rem_ep.intersect(post_ep)).mean()
# premeanscore[hd]['rem'].loc[session,'std'] = pre_score.restrict(rem_ep.intersect(pre_ep)).std()
# posmeanscore[hd]['rem'].loc[session,'std'] = pos_score.restrict(rem_ep.intersect(post_ep)).std()
return [premeanscore, posmeanscore, tsmax]
