def photoID_base_stats(db, filename=''):
'''
This function takes as argument a pandas DataFrame and adds new columns.
The filename should be the full path to where the database will be saved. If a filename is not specified, the database will not be saved.
This function computed basic statistics for all clusters (e.g. laser responsiveness, sound responsiveness, preferred frequency).
'''
soundLoc = []
NaKpeakLatency = np.empty(len(db))
laserTestStatistic = np.empty(len(db))
laserPVal = np.empty(len(db))
laserTrainTestStatistic = np.empty(len(db))
laserTrainPVal = np.empty(len(db))
laserChangeFR = np.empty(len(db))
soundResponseTestStatistic = np.empty(len(db))
soundResponsePVal = np.empty(len(db))
onsetSoundResponseTestStatistic = np.empty(len(db))
onsetSoundResponsePVal = np.empty(len(db))
sustainedSoundResponseTestStatistic = np.empty(len(db))
sustainedSoundResponsePVal = np.empty(len(db))
AMRate = np.empty(len(db))
gaussFit = []
tuningTimeRange = []
Rsquared = np.empty(len(db))
prefFreq = np.empty(len(db))
octavesFromPrefFreq = np.empty(len(db))
bestBandSession = np.empty(len(db))
for indRow, (dbIndex, dbRow) in enumerate(db.iterrows()):
cellObj = ephyscore.Cell(dbRow, useModifiedClusters=True)
print "Now processing", dbRow['subject'], dbRow['date'], dbRow[
'depth'], dbRow['tetrode'], dbRow['cluster']
# --- Determine if sound presentation was ipsi or contra to recording location ---
soundSide = dbRow['info'][2]
recordingSide = dbRow['brainArea']
if (soundSide == 'sound_left' and recordingSide == 'left_AC') or (
soundSide == 'sound_right' and recordingSide == 'right_AC'):
soundLoc.append('ipsi')
else:
soundLoc.append('contra')
# --- Determine time difference between Na and K peak (spike width) ---
peakTimes = dbRow['spikePeakTimes']
latency = peakTimes[2] - peakTimes[1]
NaKpeakLatency[indRow] = latency
# --- Determine laser responsiveness of each cell (using laser pulse) ---
try:
laserEphysData, noBehav = cellObj.load('laserPulse')
except IndexError:
print "No laser pulse session for this cell"
testStatistic = np.nan
pVal = np.nan
changeFR = np.nan
else:
testStatistic, pVal, changeFR = funcs.laser_response(
laserEphysData)
laserTestStatistic[indRow] = testStatistic
laserPVal[indRow] = pVal
laserChangeFR[indRow] = changeFR
# --- Determine laser responsiveness of each cell (using laser train) ---
try:
laserTrainEphysData, noBehav = cellObj.load('laserTrain')
except IndexError:
print "No laser train session for this cell"
testStatistic = np.nan
pVal = np.nan
else:
testStatistic, pVal, changeFR = funcs.laser_response(
laserTrainEphysData)
laserTrainTestStatistic[indRow] = testStatistic
laserTrainPVal[indRow] = pVal
# --- Determine sound responsiveness during bandwidth sessions and other statistics about bandwidth session---
try:
bandEphysData, bandBehavData = cellObj.load('bandwidth')
except IndexError:
print "No bandwidth session for this cell"
testStatistic = np.nan
pVal = np.nan
onsetTestStatistic = np.nan
onsetpVal = np.nan
sustainedTestStatistic = np.nan
sustainedpVal = np.nan
AM = np.nan
else:
bandEventOnsetTimes = funcs.get_sound_onset_times(
bandEphysData, 'bandwidth')
bandSpikeTimestamps = bandEphysData['spikeTimes']
bandEachTrial = bandBehavData['currentBand']
secondSort = bandBehavData['currentAmp']
numBands = np.unique(bandEachTrial)
numSec = np.unique(secondSort)
AM = np.unique(bandBehavData['modRate'])
trialsEachComb = behavioranalysis.find_trials_each_combination(
bandEachTrial, numBands, secondSort, numSec)
trialsEachBaseCond = trialsEachComb[:, :,
-1] #using high amp trials for photoidentified, no laser for inactivation
testStatistic, pVal = funcs.sound_response_any_stimulus(
bandEventOnsetTimes, bandSpikeTimestamps, trialsEachBaseCond,
[0.0, 1.0], [-1.2, -0.2])
onsetTestStatistic, onsetpVal = funcs.sound_response_any_stimulus(
bandEventOnsetTimes, bandSpikeTimestamps, trialsEachBaseCond,
[0.0, 0.05], [-0.25, 0.2])
sustainedTestStatistic, sustainedpVal = funcs.sound_response_any_stimulus(
bandEventOnsetTimes, bandSpikeTimestamps, trialsEachBaseCond,
[0.2, 1.0], [-1.0, 0.2])
pVal *= len(numSec) #correction for multiple comparisons
onsetpVal *= len(numSec)
sustainedpVal *= len(numSec)
soundResponseTestStatistic[indRow] = testStatistic
soundResponsePVal[indRow] = pVal
onsetSoundResponseTestStatistic[indRow] = onsetTestStatistic
onsetSoundResponsePVal[indRow] = onsetpVal
sustainedSoundResponseTestStatistic[indRow] = sustainedTestStatistic
sustainedSoundResponsePVal[indRow] = sustainedpVal
AMRate[indRow] = AM
# --- Determine frequency tuning of cells ---
try:
tuningEphysData, tuningBehavData = cellObj.load('tuningCurve')
except IndexError:
print "No tuning session for this cell"
freqFit = np.full(4, np.nan)
thisRsquared = np.nan
bestFreq = np.nan
tuningWindow = [np.nan, np.nan]
octavesFromBest = np.nan
bandIndex = np.nan
else:
tuningEventOnsetTimes = funcs.get_sound_onset_times(
tuningEphysData, 'tuningCurve')
tuningSpikeTimestamps = tuningEphysData['spikeTimes']
freqEachTrial = tuningBehavData['currentFreq']
intensityEachTrial = tuningBehavData['currentIntensity']
numFreqs = np.unique(freqEachTrial)
numIntensities = np.unique(intensityEachTrial)
timeRange = [-0.2, 0.2]
spikeTimesFromEventOnset, trialIndexForEachSpike, indexLimitsEachTrial = spikesanalysis.eventlocked_spiketimes(
tuningSpikeTimestamps, tuningEventOnsetTimes, timeRange)
trialsEachType = behavioranalysis.find_trials_each_type(
intensityEachTrial, numIntensities)
trialsHighInt = trialsEachType[:, -1]
trialsEachComb = behavioranalysis.find_trials_each_combination(
freqEachTrial, numFreqs, intensityEachTrial, numIntensities)
trialsEachFreqHighInt = trialsEachComb[:, :, -1]
tuningWindow = funcs.best_window_freq_tuning(
spikeTimesFromEventOnset, indexLimitsEachTrial,
trialsEachFreqHighInt)
spikeCountMat = spikesanalysis.spiketimes_to_spikecounts(
spikeTimesFromEventOnset, indexLimitsEachTrial, tuningWindow)
tuningSpikeRates = (spikeCountMat[trialsHighInt].flatten()) / (
tuningWindow[1] - tuningWindow[0])
freqsThisIntensity = freqEachTrial[trialsHighInt]
freqFit, thisRsquared = funcs.gaussian_tuning_fit(
np.log2(freqsThisIntensity), tuningSpikeRates)
if freqFit is not None:
bestFreq = 2**freqFit[0]
bandIndex, octavesFromBest = funcs.best_index(
cellObj, bestFreq, 'bandwidth')
else:
freqFit = np.full(4, np.nan)
bestFreq = np.nan
bandIndex = np.nan
octavesFromBest = np.nan
gaussFit.append(freqFit)
tuningTimeRange.append(tuningWindow)
Rsquared[indRow] = thisRsquared
prefFreq[indRow] = bestFreq
octavesFromPrefFreq[indRow] = octavesFromBest
bestBandSession[indRow] = bandIndex
db['soundLocation'] = soundLoc
db['spikeWidth'] = NaKpeakLatency
db['AMRate'] = AMRate
db['laserPVal'] = laserPVal
db['laserUStat'] = laserTestStatistic
db['laserTrainPVal'] = laserTrainPVal
db['laserTrainUStat'] = laserTrainTestStatistic
db['laserChangeFR'] = laserChangeFR
db['soundResponseUStat'] = soundResponseTestStatistic
db['soundResponsePVal'] = soundResponsePVal
db['onsetSoundResponseUStat'] = onsetSoundResponseTestStatistic
db['onsetSoundResponsePVal'] = onsetSoundResponsePVal
db['sustainedSoundResponseUStat'] = sustainedSoundResponseTestStatistic
db['sustainedSoundResponsePVal'] = sustainedSoundResponsePVal
db['gaussFit'] = gaussFit
db['tuningTimeRange'] = tuningTimeRange
db['tuningFitR2'] = Rsquared
db['prefFreq'] = prefFreq
db['octavesFromPrefFreq'] = octavesFromPrefFreq
db['bestBandSession'] = bestBandSession
if len(filename) != 0:
celldatabase.save_hdf(db, filename)
print filename + " saved"
return db
def inactivation_base_stats(db, filename=''):
laserTestStatistic = np.empty(len(db))
laserPVal = np.empty(len(db))
soundResponseTestStatistic = np.empty(len(db))
soundResponsePVal = np.empty(len(db))
onsetSoundResponseTestStatistic = np.empty(len(db))
onsetSoundResponsePVal = np.empty(len(db))
sustainedSoundResponseTestStatistic = np.empty(len(db))
sustainedSoundResponsePVal = np.empty(len(db))
gaussFit = []
tuningTimeRange = []
Rsquared = np.empty(len(db))
prefFreq = np.empty(len(db))
octavesFromPrefFreq = np.empty(len(db))
bestBandSession = np.empty(len(db))
for indRow, (dbIndex, dbRow) in enumerate(db.iterrows()):
cellObj = ephyscore.Cell(dbRow)
print "Now processing", dbRow['subject'], dbRow['date'], dbRow[
'depth'], dbRow['tetrode'], dbRow['cluster']
# --- Determine laser responsiveness of each cell (using first 100 ms of noise-in-laser trials) ---
try:
laserEphysData, noBehav = cellObj.load('lasernoisebursts')
except IndexError:
print "No laser pulse session for this cell"
testStatistic = np.nan
pVal = np.nan
changeFR = np.nan
else:
testStatistic, pVal, changeFR = funcs.laser_response(
laserEphysData,
baseRange=[-0.3, -0.2],
responseRange=[0.0, 0.1])
laserTestStatistic[indRow] = testStatistic
laserPVal[indRow] = pVal
# --- Determine sound responsiveness during bandwidth sessions and calculate baseline firing rates with and without laser---
#done in a kind of stupid way because regular and control sessions are handled the same way
if any(session in dbRow['sessionType']
for session in ['laserBandwidth', 'laserBandwidthControl']):
if 'laserBandwidth' in dbRow['sessionType']:
bandEphysData, bandBehavData = cellObj.load('laserBandwidth')
behavSession = 'laserBandwidth'
db.at[dbIndex, 'controlSession'] = 0
elif 'laserBandwidthControl' in dbRow['sessionType']:
bandEphysData, bandBehavData = cellObj.load(
'laserBandwidthControl')
behavSession = 'laserBandwidthControl'
db.at[dbIndex, 'controlSession'] = 1
bandEventOnsetTimes = funcs.get_sound_onset_times(
bandEphysData, 'bandwidth')
bandSpikeTimestamps = bandEphysData['spikeTimes']
bandEachTrial = bandBehavData['currentBand']
secondSort = bandBehavData['laserTrial']
numBands = np.unique(bandEachTrial)
numSec = np.unique(secondSort)
trialsEachComb = behavioranalysis.find_trials_each_combination(
bandEachTrial, numBands, secondSort, numSec)
trialsEachBaseCond = trialsEachComb[:, :,
0] #using no laser trials to determine sound responsiveness
testStatistic, pVal = funcs.sound_response_any_stimulus(
bandEventOnsetTimes, bandSpikeTimestamps, trialsEachBaseCond,
[0.0, 1.0], [-1.2, -0.2])
onsetTestStatistic, onsetpVal = funcs.sound_response_any_stimulus(
bandEventOnsetTimes, bandSpikeTimestamps, trialsEachBaseCond,
[0.0, 0.05], [-0.25, 0.2])
sustainedTestStatistic, sustainedpVal = funcs.sound_response_any_stimulus(
bandEventOnsetTimes, bandSpikeTimestamps, trialsEachBaseCond,
[0.2, 1.0], [-1.0, 0.2])
pVal *= len(numBands) #correction for multiple comparisons
onsetpVal *= len(numBands)
sustainedpVal *= len(numBands)
#pdb.set_trace()
#find baselines with and without laser
baselineRange = [-0.05, 0.0]
baselineRates, baselineSEMs = funcs.inactivated_cells_baselines(
bandSpikeTimestamps, bandEventOnsetTimes, secondSort,
baselineRange)
db.at[dbIndex, 'baselineFRnoLaser'] = baselineRates[0]
db.at[dbIndex, 'baselineFRLaser'] = baselineRates[1]
db.at[dbIndex, 'baselineFRnoLaserSEM'] = baselineSEMs[0]
db.at[dbIndex, 'baselineFRLaserSEM'] = baselineSEMs[1]
db.at[dbIndex,
'baselineChangeFR'] = baselineRates[1] - baselineRates[0]
else:
print "No bandwidth session for this cell"
testStatistic = np.nan
pVal = np.nan
onsetTestStatistic = np.nan
onsetpVal = np.nan
sustainedTestStatistic = np.nan
sustainedpVal = np.nan
#pdb.set_trace()
soundResponseTestStatistic[indRow] = testStatistic
soundResponsePVal[indRow] = pVal
onsetSoundResponseTestStatistic[indRow] = onsetTestStatistic
onsetSoundResponsePVal[indRow] = onsetpVal
sustainedSoundResponseTestStatistic[indRow] = sustainedTestStatistic
sustainedSoundResponsePVal[indRow] = sustainedpVal
# --- Determine frequency tuning of cells ---
try:
tuningEphysData, tuningBehavData = cellObj.load('tuningCurve')
except IndexError:
print "No tuning session for this cell"
freqFit = np.full(4, np.nan)
thisRsquared = np.nan
bestFreq = np.nan
tuningWindow = np.full(2, np.nan)
octavesFromBest = np.nan
bandIndex = np.nan
else:
tuningEventOnsetTimes = funcs.get_sound_onset_times(
tuningEphysData, 'tuningCurve')
tuningSpikeTimestamps = tuningEphysData['spikeTimes']
freqEachTrial = tuningBehavData['currentFreq']
intensityEachTrial = tuningBehavData['currentIntensity']
numFreqs = np.unique(freqEachTrial)
numIntensities = np.unique(intensityEachTrial)
timeRange = [-0.2, 0.2]
spikeTimesFromEventOnset, trialIndexForEachSpike, indexLimitsEachTrial = spikesanalysis.eventlocked_spiketimes(
tuningSpikeTimestamps, tuningEventOnsetTimes, timeRange)
trialsEachType = behavioranalysis.find_trials_each_type(
intensityEachTrial, numIntensities)
trialsHighInt = trialsEachType[:, -1]
trialsEachComb = behavioranalysis.find_trials_each_combination(
freqEachTrial, numFreqs, intensityEachTrial, numIntensities)
trialsEachFreqHighInt = trialsEachComb[:, :, -1]
tuningWindow = funcs.best_window_freq_tuning(
spikeTimesFromEventOnset, indexLimitsEachTrial,
trialsEachFreqHighInt)
tuningWindow = np.array(tuningWindow)
spikeCountMat = spikesanalysis.spiketimes_to_spikecounts(
spikeTimesFromEventOnset, indexLimitsEachTrial, tuningWindow)
tuningSpikeRates = (spikeCountMat[trialsHighInt].flatten()) / (
tuningWindow[1] - tuningWindow[0])
freqsThisIntensity = freqEachTrial[trialsHighInt]
freqFit, thisRsquared = funcs.gaussian_tuning_fit(
np.log2(freqsThisIntensity), tuningSpikeRates)
if freqFit is not None:
bestFreq = 2**freqFit[0]
bandIndex, octavesFromBest = funcs.best_index(
cellObj, bestFreq, behavSession)
else:
freqFit = np.full(4, np.nan)
bestFreq = np.nan
bandIndex = np.nan
octavesFromBest = np.nan
gaussFit.append(freqFit)
tuningTimeRange.append(tuningWindow)
Rsquared[indRow] = thisRsquared
prefFreq[indRow] = bestFreq
octavesFromPrefFreq[indRow] = octavesFromBest
bestBandSession[indRow] = bandIndex
db['laserPVal'] = laserPVal
db['laserUStat'] = laserTestStatistic
db['soundResponseUStat'] = soundResponseTestStatistic
db['soundResponsePVal'] = soundResponsePVal
db['onsetSoundResponseUStat'] = onsetSoundResponseTestStatistic
db['onsetSoundResponsePVal'] = onsetSoundResponsePVal
db['sustainedSoundResponseUStat'] = sustainedSoundResponseTestStatistic
db['sustainedSoundResponsePVal'] = sustainedSoundResponsePVal
db['gaussFit'] = gaussFit
db['tuningTimeRange'] = tuningTimeRange
db['tuningFitR2'] = Rsquared
db['prefFreq'] = prefFreq
db['octavesFromPrefFreq'] = octavesFromPrefFreq
db['bestBandSession'] = bestBandSession
if len(filename) != 0:
celldatabase.save_hdf(db, filename)
print filename + " saved"
return db
def inactivation_database(db,
baseStats=False,
computeIndices=True,
filename='inactivation_cells.h5'):
if type(db) == str:
dbPath = os.path.join(settings.DATABASE_PATH, db)
db = celldatabase.load_hdf(dbPath)
if baseStats:
soundResponseTestStatistic = np.empty(len(db))
soundResponsePVal = np.empty(len(db))
onsetSoundResponseTestStatistic = np.empty(len(db))
onsetSoundResponsePVal = np.empty(len(db))
sustainedSoundResponseTestStatistic = np.empty(len(db))
sustainedSoundResponsePVal = np.empty(len(db))
gaussFit = []
tuningTimeRange = []
Rsquared = np.empty(len(db))
prefFreq = np.empty(len(db))
octavesFromPrefFreq = np.empty(len(db))
bestBandSession = np.empty(len(db))
for indRow, (dbIndex, dbRow) in enumerate(db.iterrows()):
cellObj = ephyscore.Cell(dbRow)
print "Now processing", dbRow['subject'], dbRow['date'], dbRow[
'depth'], dbRow['tetrode'], dbRow['cluster']
# --- Determine sound responsiveness during bandwidth sessions and calculate baseline firing rates with and without laser---
#done in a kind of stupid way because regular and control sessions are handled the same way
if any(session in dbRow['sessionType']
for session in ['laserBandwidth', 'laserBandwidthControl']):
if 'laserBandwidth' in dbRow['sessionType']:
bandEphysData, bandBehavData = cellObj.load(
'laserBandwidth')
behavSession = 'laserBandwidth'
db.at[dbIndex, 'controlSession'] = 0
elif 'laserBandwidthControl' in dbRow['sessionType']:
bandEphysData, bandBehavData = cellObj.load(
'laserBandwidthControl')
behavSession = 'laserBandwidthControl'
db.at[dbIndex, 'controlSession'] = 1
bandEventOnsetTimes = funcs.get_sound_onset_times(
bandEphysData, 'bandwidth')
bandSpikeTimestamps = bandEphysData['spikeTimes']
bandEachTrial = bandBehavData['currentBand']
secondSort = bandBehavData['laserTrial']
numBands = np.unique(bandEachTrial)
numSec = np.unique(secondSort)
trialsEachComb = behavioranalysis.find_trials_each_combination(
bandEachTrial, numBands, secondSort, numSec)
trialsEachBaseCond = trialsEachComb[:, :,
0] #using no laser trials to determine sound responsiveness
testStatistic, pVal = funcs.sound_response_any_stimulus(
bandEventOnsetTimes, bandSpikeTimestamps,
trialsEachBaseCond, [0.0, 1.0], [-1.2, -0.2])
onsetTestStatistic, onsetpVal = funcs.sound_response_any_stimulus(
bandEventOnsetTimes, bandSpikeTimestamps,
trialsEachBaseCond, [0.0, 0.05], [-0.25, 0.2])
sustainedTestStatistic, sustainedpVal = funcs.sound_response_any_stimulus(
bandEventOnsetTimes, bandSpikeTimestamps,
trialsEachBaseCond, [0.2, 1.0], [-1.0, 0.2])
pVal *= len(numBands) #correction for multiple comparisons
onsetpVal *= len(numBands)
sustainedpVal *= len(numBands)
#pdb.set_trace()
#find baselines with and without laser
baselineRange = [-0.05, 0.0]
baselineRates, baselineSEMs = funcs.inactivated_cells_baselines(
bandSpikeTimestamps, bandEventOnsetTimes, secondSort,
baselineRange)
db.at[dbIndex, 'baselineFRnoLaser'] = baselineRates[0]
db.at[dbIndex, 'baselineFRLaser'] = baselineRates[1]
db.at[dbIndex, 'baselineFRnoLaserSEM'] = baselineSEMs[0]
db.at[dbIndex, 'baselineFRLaserSEM'] = baselineSEMs[1]
db.at[dbIndex,
'baselineChangeFR'] = baselineRates[1] - baselineRates[0]
else:
print "No bandwidth session for this cell"
testStatistic = np.nan
pVal = np.nan
onsetTestStatistic = np.nan
onsetpVal = np.nan
sustainedTestStatistic = np.nan
sustainedpVal = np.nan
#pdb.set_trace()
soundResponseTestStatistic[indRow] = testStatistic
soundResponsePVal[indRow] = pVal
onsetSoundResponseTestStatistic[indRow] = onsetTestStatistic
onsetSoundResponsePVal[indRow] = onsetpVal
sustainedSoundResponseTestStatistic[
indRow] = sustainedTestStatistic
sustainedSoundResponsePVal[indRow] = sustainedpVal
# --- Determine frequency tuning of cells ---
try:
tuningEphysData, tuningBehavData = cellObj.load('tuningCurve')
except IndexError:
print "No tuning session for this cell"
freqFit = np.full(4, np.nan)
thisRsquared = np.nan
bestFreq = np.nan
tuningWindow = np.full(2, np.nan)
octavesFromBest = np.nan
bandIndex = np.nan
else:
tuningEventOnsetTimes = funcs.get_sound_onset_times(
tuningEphysData, 'tuningCurve')
tuningSpikeTimestamps = tuningEphysData['spikeTimes']
freqEachTrial = tuningBehavData['currentFreq']
intensityEachTrial = tuningBehavData['currentIntensity']
numFreqs = np.unique(freqEachTrial)
numIntensities = np.unique(intensityEachTrial)
timeRange = [-0.2, 0.2]
spikeTimesFromEventOnset, trialIndexForEachSpike, indexLimitsEachTrial = spikesanalysis.eventlocked_spiketimes(
tuningSpikeTimestamps, tuningEventOnsetTimes, timeRange)
trialsEachType = behavioranalysis.find_trials_each_type(
intensityEachTrial, numIntensities)
trialsHighInt = trialsEachType[:, -1]
trialsEachComb = behavioranalysis.find_trials_each_combination(
freqEachTrial, numFreqs, intensityEachTrial,
numIntensities)
trialsEachFreqHighInt = trialsEachComb[:, :, -1]
tuningWindow = funcs.best_window_freq_tuning(
spikeTimesFromEventOnset, indexLimitsEachTrial,
trialsEachFreqHighInt)
tuningWindow = np.array(tuningWindow)
spikeCountMat = spikesanalysis.spiketimes_to_spikecounts(
spikeTimesFromEventOnset, indexLimitsEachTrial,
tuningWindow)
tuningSpikeRates = (spikeCountMat[trialsHighInt].flatten()) / (
tuningWindow[1] - tuningWindow[0])
freqsThisIntensity = freqEachTrial[trialsHighInt]
freqFit, thisRsquared = funcs.gaussian_tuning_fit(
np.log2(freqsThisIntensity), tuningSpikeRates)
if freqFit is not None:
bestFreq = 2**freqFit[0]
bandIndex, octavesFromBest = funcs.best_index(
cellObj, bestFreq, behavSession)
else:
freqFit = np.full(4, np.nan)
bestFreq = np.nan
bandIndex = np.nan
octavesFromBest = np.nan
gaussFit.append(freqFit)
tuningTimeRange.append(tuningWindow)
Rsquared[indRow] = thisRsquared
prefFreq[indRow] = bestFreq
octavesFromPrefFreq[indRow] = octavesFromBest
bestBandSession[indRow] = bandIndex
db['soundResponseUStat'] = soundResponseTestStatistic
db['soundResponsePVal'] = soundResponsePVal
db['onsetSoundResponseUStat'] = onsetSoundResponseTestStatistic
db['onsetSoundResponsePVal'] = onsetSoundResponsePVal
db['sustainedSoundResponseUStat'] = sustainedSoundResponseTestStatistic
db['sustainedSoundResponsePVal'] = sustainedSoundResponsePVal
db['gaussFit'] = gaussFit
db['tuningTimeRange'] = tuningTimeRange
db['tuningFitR2'] = Rsquared
db['prefFreq'] = prefFreq
db['octavesFromPrefFreq'] = octavesFromPrefFreq
db['bestBandSession'] = bestBandSession
if computeIndices:
bestCells = db.query("isiViolations<0.02") # or modifiedISI<0.02")
bestCells = bestCells.loc[bestCells['spikeShapeQuality'] > 2]
bestCells = bestCells.query(
'soundResponsePVal<0.05 or onsetSoundResponsePVal<0.05 or sustainedSoundResponsePVal<0.05'
)
bestCells = bestCells.loc[bestCells['tuningFitR2'] > R2CUTOFF]
bestCells = bestCells.loc[
bestCells['octavesFromPrefFreq'] < OCTAVESCUTOFF]
for dbIndex, dbRow in bestCells.iterrows():
cell = ephyscore.Cell(dbRow)
bandEphysData, bandBehavData = cell.load_by_index(
int(dbRow['bestBandSession']))
bandEventOnsetTimes = funcs.get_sound_onset_times(
bandEphysData, 'bandwidth')
bandSpikeTimestamps = bandEphysData['spikeTimes']
bandEachTrial = bandBehavData['currentBand']
secondSort = bandBehavData['laserTrial']
propOnset, propSustained = funcs.onset_sustained_spike_proportion(
bandSpikeTimestamps, bandEventOnsetTimes)
db.at[dbIndex, 'proportionSpikesOnset'] = propOnset
db.at[dbIndex, 'proportionSpikesSustained'] = propSustained
#by default: not subtracting baseline, but are replacing pure tone response with baseline for 0 bw condition
onsetSupInds, onsetSupIndpVals, onsetFacInds, onsetFacIndpVals, onsetPeakInds, onsetSpikeArray = funcs.bandwidth_suppression_from_peak(
bandSpikeTimestamps,
bandEventOnsetTimes,
bandEachTrial,
secondSort,
timeRange=[0.0, 0.05],
baseRange=[-0.05, 0.0])
db.at[dbIndex, 'onsetSuppressionIndexLaser'] = onsetSupInds[-1]
db.at[dbIndex, 'onsetSuppressionpValLaser'] = onsetSupIndpVals[-1]
db.at[dbIndex, 'onsetFacilitationIndexLaser'] = onsetFacInds[-1]
db.at[dbIndex, 'onsetFacilitationpValLaser'] = onsetFacIndpVals[-1]
db.at[dbIndex,
'onsetPrefBandwidthLaser'] = bandEachTrial[onsetPeakInds[-1]]
db.at[dbIndex, 'onsetSuppressionIndexNoLaser'] = onsetSupInds[0]
db.at[dbIndex, 'onsetSuppressionpValNoLaser'] = onsetSupIndpVals[0]
db.at[dbIndex, 'onsetFacilitationIndexNoLaser'] = onsetFacInds[0]
db.at[dbIndex,
'onsetFacilitationpValNoLaser'] = onsetFacIndpVals[0]
db.at[dbIndex, 'onsetPrefBandwidthNoLaser'] = bandEachTrial[
onsetPeakInds[0]]
#base range is right before sound onset so we get estimate for laser baseline
sustainedSupInds, sustainedSupIndpVals, sustainedFacInds, sustainedFacIndpVals, sustainedPeakInds, sustainedSpikeArray = funcs.bandwidth_suppression_from_peak(
bandSpikeTimestamps,
bandEventOnsetTimes,
bandEachTrial,
secondSort,
timeRange=[0.2, 1.0],
baseRange=[-0.05, 0.0])
db.at[dbIndex,
'sustainedSuppressionIndexLaser'] = sustainedSupInds[-1]
db.at[dbIndex,
'sustainedSuppressionpValLaser'] = sustainedSupIndpVals[-1]
db.at[dbIndex,
'sustainedFacilitationIndexLaser'] = sustainedFacInds[-1]
db.at[dbIndex,
'sustainedFacilitationpValLaser'] = sustainedFacIndpVals[-1]
db.at[dbIndex, 'sustainedPrefBandwidthLaser'] = bandEachTrial[
sustainedPeakInds[-1]]
db.at[dbIndex,
'sustainedSuppressionIndexNoLaser'] = sustainedSupInds[0]
db.at[dbIndex,
'sustainedSuppressionpValNoLaser'] = sustainedSupIndpVals[0]
db.at[dbIndex,
'sustainedFacilitationIndexNoLaser'] = sustainedFacInds[0]
db.at[dbIndex,
'sustainedFacilitationpValNoLaser'] = sustainedFacIndpVals[0]
db.at[dbIndex, 'sustainedPrefBandwidthNoLaser'] = bandEachTrial[
sustainedPeakInds[0]]
#no laser fit
sustainedResponseNoLaser = sustainedSpikeArray[:, 0]
bandsForFit = np.unique(bandEachTrial)
bandsForFit[-1] = 6
mFixed = 1
fitParams, R2 = fitfuncs.diff_of_gauss_fit(
bandsForFit, sustainedResponseNoLaser, mFixed=mFixed)
#fit params
db.at[dbIndex, 'R0noLaser'] = fitParams[0]
db.at[dbIndex, 'RDnoLaser'] = fitParams[3]
db.at[dbIndex, 'RSnoLaser'] = fitParams[4]
db.at[dbIndex, 'mnoLaser'] = mFixed
db.at[dbIndex, 'sigmaDnoLaser'] = fitParams[1]
db.at[dbIndex, 'sigmaSnoLaser'] = fitParams[2]
db.at[dbIndex, 'bandwidthTuningR2noLaser'] = R2
testBands = np.linspace(bandsForFit[0], bandsForFit[-1], 500)
allFitParams = [mFixed]
allFitParams.extend(fitParams)
suppInd, prefBW = fitfuncs.extract_stats_from_fit(
allFitParams, testBands)
db.at[dbIndex, 'fitSustainedSuppressionIndexNoLaser'] = suppInd
db.at[dbIndex, 'fitSustainedPrefBandwidthNoLaser'] = prefBW
#laser fit
sustainedResponseLaser = sustainedSpikeArray[:, 1]
fitParamsLaser, R2Laser = fitfuncs.diff_of_gauss_fit(
bandsForFit, sustainedResponseLaser, mFixed=mFixed)
#fit params
db.at[dbIndex, 'R0laser'] = fitParamsLaser[0]
db.at[dbIndex, 'RDlaser'] = fitParamsLaser[3]
db.at[dbIndex, 'RSlaser'] = fitParamsLaser[4]
db.at[dbIndex, 'mlaser'] = mFixed
db.at[dbIndex, 'sigmaDlaser'] = fitParamsLaser[1]
db.at[dbIndex, 'sigmaSlaser'] = fitParamsLaser[2]
db.at[dbIndex, 'bandwidthTuningR2laser'] = R2Laser
allFitParamsLaser = [mFixed]
allFitParamsLaser.extend(fitParamsLaser)
suppIndLaser, prefBWLaser = fitfuncs.extract_stats_from_fit(
allFitParamsLaser, testBands)
db.at[dbIndex, 'fitSustainedSuppressionIndexLaser'] = suppIndLaser
db.at[dbIndex, 'fitSustainedPrefBandwidthLaser'] = prefBWLaser
meanLaserDiff = np.mean(sustainedResponseLaser -
sustainedResponseNoLaser)
db.at[dbIndex, 'laserChangeResponse'] = meanLaserDiff
laserDiff = sustainedResponseLaser - sustainedResponseNoLaser
peakInd = np.argmax(sustainedResponseNoLaser)
db.at[dbIndex, 'peakChangeFR'] = laserDiff[peakInd]
db.at[dbIndex, 'WNChangeFR'] = laserDiff[-1]
testRespsNoLaser = fitfuncs.diff_gauss_form(
testBands, *allFitParams)
testRespsLaser = fitfuncs.diff_gauss_form(testBands,
*allFitParamsLaser)
laserDiffModel = testRespsLaser - testRespsNoLaser
peakIndModel = np.argmax(testRespsNoLaser)
db.at[dbIndex, 'fitPeakChangeFR'] = laserDiffModel[peakIndModel]
db.at[dbIndex, 'fitWNChangeFR'] = laserDiffModel[-1]
#also calculating fits and suppression with pure tone being 0 bw condition
toneSustainedSupInds, toneSustainedSupIndpVals, toneSustainedFacInds, toneSustainedFacIndpVals, toneSustainedPeakInds, toneSustainedSpikeArray = funcs.bandwidth_suppression_from_peak(
bandSpikeTimestamps,
bandEventOnsetTimes,
bandEachTrial,
secondSort,
timeRange=[0.2, 1.0],
baseRange=[-0.05, 0.0],
zeroBWBaseline=False)
db.at[
dbIndex,
'sustainedSuppressionIndexNoLaserPureTone'] = toneSustainedSupInds[
0]
db.at[
dbIndex,
'sustainedSuppressionpValNoLaserPureTone'] = toneSustainedSupIndpVals[
0]
db.at[
dbIndex,
'sustainedFacilitationIndexNoLaserPureTone'] = toneSustainedFacInds[
0]
db.at[
dbIndex,
'sustainedFacilitationpValNoLaserPureTone'] = toneSustainedFacIndpVals[
0]
db.at[dbIndex,
'sustainedPrefBandwidthNoLaserPureTone'] = bandEachTrial[
toneSustainedPeakInds[0]]
db.at[
dbIndex,
'sustainedSuppressionIndexLaserPureTone'] = toneSustainedSupInds[
-1]
db.at[
dbIndex,
'sustainedSuppressionpValLaserPureTone'] = toneSustainedSupIndpVals[
-1]
db.at[
dbIndex,
'sustainedFacilitationIndexLaserPureTone'] = toneSustainedFacInds[
-1]
db.at[
dbIndex,
'sustainedFacilitationpValLaserPureTone'] = toneSustainedFacIndpVals[
-1]
db.at[dbIndex,
'sustainedPrefBandwidthLaserPureTone'] = bandEachTrial[
toneSustainedPeakInds[-1]]
toneSustainedResponseNoLaser = toneSustainedSpikeArray[:, 0]
toneFitParamsNoLaser, toneR2 = fitfuncs.diff_of_gauss_fit(
bandsForFit, toneSustainedResponseNoLaser, mFixed=mFixed)
#fit params
db.at[dbIndex, 'R0PureToneNoLaser'] = toneFitParamsNoLaser[0]
db.at[dbIndex, 'RDPureToneNoLaser'] = toneFitParamsNoLaser[3]
db.at[dbIndex, 'RSPureToneNoLaser'] = toneFitParamsNoLaser[4]
db.at[dbIndex, 'mPureToneNoLaser'] = mFixed
db.at[dbIndex, 'sigmaDPureToneNoLaser'] = toneFitParamsNoLaser[1]
db.at[dbIndex, 'sigmaSPureToneNoLaser'] = toneFitParamsNoLaser[2]
db.at[dbIndex, 'bandwidthTuningR2PureToneNoLaser'] = toneR2
allFitParamsToneNoLaser = [mFixed]
allFitParamsToneNoLaser.extend(toneFitParamsNoLaser)
suppIndTone, prefBWTone = fitfuncs.extract_stats_from_fit(
allFitParamsToneNoLaser, testBands)
db.at[dbIndex,
'fitSustainedSuppressionIndexPureToneNoLaser'] = suppIndTone
db.at[dbIndex,
'fitSustainedPrefBandwidthPureToneNoLaser'] = prefBWTone
toneSustainedResponseLaser = toneSustainedSpikeArray[:, 1]
toneFitParamsLaser, toneR2Laser = fitfuncs.diff_of_gauss_fit(
bandsForFit, toneSustainedResponseLaser, mFixed=mFixed)
#fit params
db.at[dbIndex, 'R0PureToneLaser'] = toneFitParamsLaser[0]
db.at[dbIndex, 'RDPureToneLaser'] = toneFitParamsLaser[3]
db.at[dbIndex, 'RSPureToneLaser'] = toneFitParamsLaser[4]
db.at[dbIndex, 'mPureToneLaser'] = mFixed
db.at[dbIndex, 'sigmaDPureToneLaser'] = toneFitParamsLaser[1]
db.at[dbIndex, 'sigmaSPureToneLaser'] = toneFitParamsLaser[2]
db.at[dbIndex, 'bandwidthTuningR2PureToneLaser'] = toneR2Laser
allFitParamsToneLaser = [mFixed]
allFitParamsToneLaser.extend(toneFitParamsLaser)
suppIndToneLaser, prefBWToneLaser = fitfuncs.extract_stats_from_fit(
allFitParamsToneLaser, testBands)
db.at[
dbIndex,
'fitSustainedSuppressionIndexPureToneLaser'] = suppIndToneLaser
db.at[dbIndex,
'fitSustainedPrefBandwidthPureToneLaser'] = prefBWToneLaser
testRespsNoLaser = fitfuncs.diff_gauss_form(
testBands, *allFitParamsToneNoLaser)
testRespsLaser = fitfuncs.diff_gauss_form(testBands,
*allFitParamsToneLaser)
laserDiffModel = testRespsLaser - testRespsNoLaser
peakIndModel = np.argmax(testRespsNoLaser)
db.at[dbIndex,
'fitPeakChangeFRPureTone'] = laserDiffModel[peakIndModel]
db.at[dbIndex, 'fitWNChangeFRPureTone'] = laserDiffModel[-1]
#also calculating fits and suppression with nothing being fit for bw 0
noZeroSustainedResponseNoLaser = sustainedSpikeArray[1:, 0]
bandsForFitNoZero = bandsForFit[1:]
noZeroFitParamsNoLaser, noZeroR2 = fitfuncs.diff_of_gauss_fit(
bandsForFitNoZero,
noZeroSustainedResponseNoLaser,
mFixed=mFixed)
#fit params
db.at[dbIndex, 'R0noZeroNoLaser'] = noZeroFitParamsNoLaser[0]
db.at[dbIndex, 'RDnoZeroNoLaser'] = noZeroFitParamsNoLaser[3]
db.at[dbIndex, 'RSnoZeroNoLaser'] = noZeroFitParamsNoLaser[4]
db.at[dbIndex, 'mnoZeroNoLaser'] = mFixed
db.at[dbIndex, 'sigmaDnoZeroNoLaser'] = noZeroFitParamsNoLaser[1]
db.at[dbIndex, 'sigmaSnoZeroNoLaser'] = noZeroFitParamsNoLaser[2]
db.at[dbIndex, 'bandwidthTuningR2noZeroNoLaser'] = noZeroR2
allFitParamsNoZero = [mFixed]
allFitParamsNoZero.extend(noZeroFitParamsNoLaser)
testBandsNoZero = np.linspace(bandsForFitNoZero[0],
bandsForFitNoZero[-1], 500)
suppIndNoZero, prefBWNoZero = fitfuncs.extract_stats_from_fit(
allFitParamsNoZero, testBandsNoZero)
db.at[dbIndex,
'fitSustainedSuppressionIndexNoZeroNoLaser'] = suppIndNoZero
db.at[dbIndex,
'fitSustainedPrefBandwidthNoZeroNoLaser'] = prefBWNoZero
noZeroSustainedResponseLaser = sustainedSpikeArray[1:, 1]
bandsForFitNoZero = bandsForFit[1:]
noZeroFitParamsLaser, noZeroR2Laser = fitfuncs.diff_of_gauss_fit(
bandsForFitNoZero, noZeroSustainedResponseLaser, mFixed=mFixed)
#fit params
db.at[dbIndex, 'R0noZeroLaser'] = noZeroFitParamsLaser[0]
db.at[dbIndex, 'RDnoZeroLaser'] = noZeroFitParamsLaser[3]
db.at[dbIndex, 'RSnoZeroLaser'] = noZeroFitParamsLaser[4]
db.at[dbIndex, 'mnoZeroLaser'] = mFixed
db.at[dbIndex, 'sigmaDnoZeroLaser'] = noZeroFitParamsLaser[1]
db.at[dbIndex, 'sigmaSnoZeroLaser'] = noZeroFitParamsLaser[2]
db.at[dbIndex, 'bandwidthTuningR2noZeroLaser'] = noZeroR2Laser
allFitParamsNoZeroLaser = [mFixed]
allFitParamsNoZeroLaser.extend(noZeroFitParamsLaser)
suppIndNoZeroLaser, prefBWNoZeroLaser = fitfuncs.extract_stats_from_fit(
allFitParamsNoZeroLaser, testBandsNoZero)
db.at[
dbIndex,
'fitSustainedSuppressionIndexNoZeroLaser'] = suppIndNoZeroLaser
db.at[dbIndex,
'fitSustainedPrefBandwidthNoZeroLaser'] = prefBWNoZeroLaser
testRespsNoLaser = fitfuncs.diff_gauss_form(
testBandsNoZero, *allFitParamsNoZero)
testRespsLaser = fitfuncs.diff_gauss_form(testBandsNoZero,
*allFitParamsNoZeroLaser)
laserDiffModel = testRespsLaser - testRespsNoLaser
peakIndModel = np.argmax(testRespsNoLaser)
db.at[dbIndex,
'fitPeakChangeFRNoZero'] = laserDiffModel[peakIndModel]
db.at[dbIndex, 'fitWNChangeFRNoZero'] = laserDiffModel[-1]
if len(filename) != 0:
celldatabase.save_hdf(db, dbFilename)