def process(self, data, display=True):
params = ['Amp1', 'Amp2', 'Time', 'Rise', 'Decay', 'Decay2']
guess = tuple([self.ctrls['guess' + name].value() for name in params])
bounds = [(self.ctrls['min' + name].value(), self.ctrls['max' + name].value()) for name in params]
rp = self.ctrls['risePower'].value()
minTime = self.ctrls['minTime'].value()
times = data.xvals('Time')
fullTimes = times
data = data.view(np.ndarray)
startInd = np.argwhere(times>=minTime)[0]
times = times[startInd:]
data = data[startInd:]
## Ignore specified guess amplitude; make a better guess.
mx = data.max()
mn = data.min()
if abs(mn) > abs(mx):
amp = mn
else:
amp = mx
params[0] = amp
params[1] = amp
fit = functions.fitDoublePsp(x=times, y=data, guess=guess, bounds=bounds, risePower=rp)
if self.ctrls['computeWaveform'].isChecked():
fitData = functions.doublePspFunc(fit, fullTimes, rp)
else:
fitData = None
return {'fitParams': fit, 'fitYData': fitData, 'fitXData': fullTimes}
def process(self, data, display=True):
rp = self.ctrls['risePower'].value()
stimTime = self.ctrls['stimulusTime'].value()
times = data.xvals('Time')
dt = times[1] - times[0]
data1 = data.view(np.ndarray)
if stimTime > times[-1]:
raise Exception("stimulusTime is larger than length of input data.")
stimInd = np.argwhere(times>=stimTime)[0][0]
# 1. make sure offset is removed
offset = np.median(data1[:stimInd])
data2 = data1 - offset
# 2. check for zero-crossing events within 4ms after stimulus
cross = functions.zeroCrossingEvents(data2[stimInd:])
maxInd = 4e-3 / dt
minLength = self.ctrls['minDirectDuration'].value() / dt
gotEvent = None
for start, length, sum, peak in cross:
if start > maxInd:
break
if length < minLength:
continue
if gotEvent is None or length > gotEvent[1]:
gotEvent = (start+stimInd, length)
#if len(cross) == 0: ## must be a single large event
#gotEvent = (0, len(data2)-stimInd)
fitParams = dict(
directFitAmp1=0., directFitAmp2=0., directFitTime=0.,
directFitRise=0., directFitDecay1=0., directFitDecay2=0.,
directFitPeakTime=0., directFitPeak=0., directFitSubtracted=False,
directFitValid=False,
)
# 3. if there is no large event near stimulus, return original data
if gotEvent is None:
if display:
self.plotItem.clear()
return {'output': data, 'fitParams': fitParams, 'plot': self.plotItem}
#print "============"
#print stimInd
#print gotEvent
#print cross[:20]
# 4. guess amplitude, tau from detected event
evStart = gotEvent[0]
evEnd = evStart + gotEvent[1]
evRgn = data2[evStart:evEnd]
#pg.plot(evRgn)
mx = evRgn.max()
mn = evRgn.min()
ampGuess = mx if abs(mx) > abs(mn) else mn
tauGuess = gotEvent[1] * dt
if ampGuess < 0:
ampBound = [None, 0]
else:
ampBound = [0, None]
# 5. fit
guess = [ampGuess*2, ampGuess*2, stimTime, 1e-3, tauGuess*0.5, tauGuess*2]
bounds = [
ampBound, ampBound,
[stimTime, stimTime+4e-3],
[1e-4, 100e-3],
[1e-3, 1],
[5e-3, 10],
]
#print guess
#print bounds
endInd = evStart + gotEvent[1] * 10
fitYData = data2[stimInd:endInd]
fitXData = times[stimInd:endInd]
fit = functions.fitDoublePsp(x=fitXData, y=fitYData, guess=guess, bounds=bounds, risePower=rp)
# 6. subtract fit from original data (offset included), return
y = functions.doublePspFunc(fit, times, rp)
if display:
self.plotItem.setData(x=fitXData, y=y[stimInd:endInd], pen=self.ctrls['plotColor'].color())
## prepare list of fit params for output
(xMax, yMax) = functions.doublePspMax(fit)
xMax -= fit[2] ## really interested in time-to-peak, not the exact peak time.
fitParams = dict(
directFitAmp1=fit[0], directFitAmp2=fit[1], directFitTime=fit[2],
directFitRise=fit[3], directFitDecay1=fit[4], directFitDecay2=fit[5],
directFitPeakTime=xMax, directFitPeak=yMax, directFitSubtracted=None, directFitValid=True,
)
if self.ctrls['subtractDirect'].isChecked():
out = metaarray.MetaArray(data-y, info=data.infoCopy())
fitParams['directFitSubtracted'] = True
else:
out = data
fitParams['directFitSubtracted'] = False
return {'fitParams': fitParams, 'output': out, 'plot': self.plotItem}
