def addLabel(info=None):
global labelInfo
create = False
if info is None:
create = True
l = ui.labelSpin.value()
if l in labelInfo:
return
info = {
'visible': True,
'name': 'label',
'color': pg.intColor(len(labelInfo), 16),
'id': l
}
else:
info = info.copy()
info['color'] = pg.mkColor(info['color'])
l = info['id']
item = Qt.QTreeWidgetItem([str(l), info['name'], ''])
item.setFlags(item.flags() | Qt.Qt.ItemIsEditable
| Qt.Qt.ItemIsUserCheckable)
if info['visible']:
item.setCheckState(0, Qt.Qt.Checked)
else:
item.setCheckState(0, Qt.Qt.Unchecked)
btn = pg.ColorButton(color=info['color'])
ui.labelTree.addTopLevelItem(item)
ui.labelTree.setItemWidget(item, 2, btn)
labelInfo[l] = {'item': item, 'btn': btn}
btn.sigColorChanged.connect(itemChanged)
btn.sigColorChanging.connect(imageChanged)
if create:
writeMeta()
def addROI(self, roiType):
pen = pg.mkPen(pg.intColor(len(self.ROIs)))
center = self.view.viewRect().center()
#print 'camerawindow.py: addROI:: ', self.view.viewPixelSize()
size = [x * 50 for x in self.view.viewPixelSize()]
if roiType == 'rect':
roi = PlotROI(center, size)
elif roiType == 'ellipse':
roi = pg.EllipseROI(center, size, removable=True)
elif roiType == 'polygon':
pts = [
center, center + pg.Point(0, size[1]),
center + pg.Point(size[0], 0)
]
roi = pg.PolyLineROI(pts, closed=True, removable=True)
elif roiType == 'ruler':
pts = [center, center + pg.Point(size[0], size[1])]
roi = RulerROI(pts, removable=True)
else:
raise ValueError("Invalid ROI type %s" % roiType)
roi.setZValue(40000)
roi.setPen(pen)
self.view.addItem(roi)
plot = self.roiPlot.plot(pen=pen)
self.ROIs.append({'roi': roi, 'plot': plot, 'vals': [], 'times': []})
roi.sigRemoveRequested.connect(self.removeROI)
def addLabel(info=None):
global labelInfo
create = False
if info is None:
create = True
l = ui.labelSpin.value()
if l in labelInfo:
return
info = {
'visible': True,
'name': 'label',
'color': pg.intColor(len(labelInfo), 16),
'id': l
}
else:
info = info.copy()
info['color'] = pg.mkColor(info['color'])
l = info['id']
item = QtGui.QTreeWidgetItem([str(l), info['name'], ''])
item.setFlags(item.flags() | QtCore.Qt.ItemIsEditable | QtCore.Qt.ItemIsUserCheckable)
if info['visible']:
item.setCheckState(0, QtCore.Qt.Checked)
else:
item.setCheckState(0, QtCore.Qt.Unchecked)
btn = pg.ColorButton(color=info['color'])
ui.labelTree.addTopLevelItem(item)
ui.labelTree.setItemWidget(item, 2, btn)
labelInfo[l] = {'item': item, 'btn': btn}
btn.sigColorChanged.connect(itemChanged)
btn.sigColorChanging.connect(imageChanged)
if create:
writeMeta()
def redisplayData(self, points): ## data must be [(scan, fh, <event time>), ...]
#raise Exception('blah')
#print points
try:
QtGui.QApplication.setOverrideCursor(QtGui.QCursor(QtCore.Qt.WaitCursor))
plot = self.getElement("Data Plot")
plot.clear()
eTable = self.getElement("Event Table")
sTable = self.getElement("Stats")
#num = len(point.data)
num = len(points)
statList = []
evList = []
for i in range(num):
color = pg.intColor(i, num)
#scan, fh = point.data[i]
try:
scan, fh = points[i][:2]
except:
print points[i]
raise
if len(points[i]) == 3:
evTime = points[i][2]
else:
evTime = None
scan.displayData(fh, plot, color, evTime)
## show stats
stats = scan.getStats(fh.parent())
statList.append(stats)
events = scan.getEvents(fh)['events']
if len(events) > 0:
evList.append(events)
sTable.setData(statList)
if len(evList) > 0:
try:
eTable.setData(np.concatenate(evList))
except:
for i in range(1,len(evList)):
if len(evList[i].dtype) != len(evList[i-1].dtype):
print "Cannot concatenate; event lists have different dtypes:"
print evList[i].dtype
print evList[i-1].dtype
else:
for j in range(len(evList[i].dtype)):
if evList[i-1].dtype[j] != evList[i].dtype[j]:
for l in evList:
print l
print "Warning: can not concatenate--field '%s' has inconsistent types %s, %s (data printed above)" % (evList[i].dtype.names[j], str(evList[i-1].dtype[j]), str(evList[i].dtype[j]))
raise
finally:
QtGui.QApplication.restoreOverrideCursor()
def addROI(self):
pen = pg.mkPen(pg.intColor(len(self.ROIs)))
center = self.view.viewRect().center()
#print 'camerawindow.py: addROI:: ', self.view.viewPixelSize()
size = [x*50 for x in self.view.viewPixelSize()]
roi = PlotROI(center, size)
roi.setZValue(40000)
roi.setPen(pen)
self.view.addItem(roi)
plot = self.roiPlot.plot(pen=pen)
self.ROIs.append({'roi': roi, 'plot': plot, 'vals': [], 'times': []})
roi.sigRemoveRequested.connect(self.removeROI)
def updateTracesPlot(self):
"""Update the Trace display plot to show the traces corresponding to
the timestamps selected by the region in the experiment plot."""
rgn = self.traceSelectRgn.getRegion()
self.tracesPlot.clear()
### plot all the traces with timestamps within the selected region (according to self.traceSelectRgn)
data = self.traces[(self.traces['timestamp'] > rgn[0] + self.expStart)
*
(self.traces['timestamp'] < rgn[1] + self.expStart)]
for i, d in enumerate(data['data']):
self.tracesPlot.plot(d['primary'], pen=pg.intColor(i, len(data)))
if len(data) > 0:
self.flowchart.setInput(dataIn=data[0]['fileHandle'])
def plotData():
rate = 1e3
nPts = 100
plot.clear()
params = {}
paramSpace = sg.listSequences()
for k in paramSpace:
params[k] = range(len(paramSpace[k]))
global seqPlots
seqPlots = []
SequenceRunner.runSequence(lambda p: seqPlots.append(sg.getSingle(rate, nPts, p)), params, params.keys())
for i, w in enumerate(seqPlots):
if w is not None:
plot.plot(w, pen=pg.intColor(i, len(seqPlots)*1.5))
data = sg.getSingle(rate, nPts)
plot.plot(data, pen=pg.mkPen('w', width=2))
def plotData():
rate = 1e3
nPts = 100
plot.clear()
params = {}
paramSpace = sg.listSequences()
for k in paramSpace:
params[k] = range(len(paramSpace[k]))
global seqPlots
seqPlots = []
SequenceRunner.runSequence(
lambda p: seqPlots.append(sg.getSingle(rate, nPts, p)), params,
list(params.keys()))
for i, w in enumerate(seqPlots):
if w is not None:
plot.plot(w, pen=pg.intColor(i, len(seqPlots) * 1.5))
data = sg.getSingle(rate, nPts)
plot.plot(data, pen=pg.mkPen('w', width=2))
def _load_data(self, seqDir):
man = getManager()
model = man.dataModel
# read all image data
self.img_data = model.buildSequenceArray(
seqDir, lambda dh: dh['Camera']['frames.ma'].read(),
join=False).asarray()
seqParams = list(model.listSequenceParams(seqDir).items())
if self.img_data.ndim == 1:
self.img_data = self.img_data[np.newaxis, :]
seqParams.insert(0, (None, [0]))
transpose = seqParams[0][0] == ('protocol', 'repetitions')
if transpose:
self.img_data = np.swapaxes(self.img_data, 0, 1)
seqParams = seqParams[::-1]
self.seqParams = seqParams
if seqParams[0][0] is None:
self.seqColors = [pg.mkColor('w')]
else:
nSeq = len(seqParams[0][1])
if nSeq == 2:
# Special case: add in difference between two sequence trials
seqParams[0] = (seqParams[0][0], list(seqParams[0][1]) +
[np.mean(seqParams[0][1])])
nSeq = 3
img_data = np.empty((3, ) + self.img_data.shape[1:],
dtype=self.img_data.dtype)
img_data[:2] = self.img_data
for i in range(img_data.shape[1]):
minlen = min(self.img_data[0, i].shape[0],
self.img_data[1, i].shape[0])
img_data[2, i] = self.img_data[0, i][:minlen].copy()
img_data[2, i]._data = self.img_data[
0, i][:minlen].asarray().astype(
'float') - self.img_data[1, i][:minlen].asarray()
self.img_data = img_data
self.seqColors = [pg.intColor(i, nSeq * 1.6) for i in range(nSeq)]
# cull out truncated recordings :(
self.img_data = [[
d['Time':0:200e-3] for d in row if d.xvals('Time')[-1] > 150e-3
] for row in self.img_data]
# crop / concatenate
img_len = min(
[min([d.shape[0] for d in row]) for row in self.img_data])
self.img_data = [[d[:img_len] for d in row] for row in self.img_data]
self.img_arrays = [
np.concatenate([d.asarray()[np.newaxis, ...] for d in row], axis=0)
for row in self.img_data
]
# average
self.img_mean = [img_arr.mean(axis=0) for img_arr in self.img_arrays]
for p in self.clamp_plots:
self.plt2.removeItem(p)
# read all clamp data
first_subdir = seqDir[seqDir.ls()[0]]
clamp_name = 'Clamp1'
if not first_subdir[clamp_name + '.ma'].exists():
clamp_name = 'Clamp2'
clamp_file = first_subdir[clamp_name + '.ma']
if clamp_file.exists():
self.clamp_mode = model.getClampMode(clamp_file)
chan = 'command' if self.clamp_mode == 'VC' else 'primary'
self.clamp_data = model.buildSequenceArray(
seqDir,
lambda dh: dh[clamp_name + '.ma'].read()['Channel':chan],
join=False).asarray()
if self.clamp_data.ndim == 1:
self.clamp_data = self.clamp_data[np.newaxis, :]
if transpose:
self.clamp_data = np.swapaxes(self.clamp_data, 0, 1)
self.plt2.setLabels(left=('Vm', 'V'))
for i in range(self.clamp_data.shape[0]):
for j in range(self.clamp_data.shape[1]):
trace = self.clamp_data[i, j]
pen = self.seqColors[i]
p = self.plt2.plot(trace.xvals('Time'),
trace.asarray(),
antialias=True,
pen=pen)
self.clamp_plots.append(p)
self.plt2.show()
else:
self.clamp_mode = None
self.plt2.hide()
self.img_t = self.img_data[0][0].xvals('Time')
self.img1.setImage(self.img_mean[-1].mean(axis=0))
self.roi_changed(None)
self.time_rgn_changed(None)
self.update_sequence_analysis()
def _load_data(self, seqDir):
man = getManager()
model = man.dataModel
# read all image data
self.img_data = model.buildSequenceArray(
seqDir,
lambda dh: dh['Camera']['frames.ma'].read(),
join=False).asarray()
seqParams = list(model.listSequenceParams(seqDir).items())
if self.img_data.ndim == 1:
self.img_data = self.img_data[np.newaxis, :]
seqParams.insert(0, (None, [0]))
transpose = seqParams[0][0] == ('protocol', 'repetitions')
if transpose:
self.img_data = np.swapaxes(self.img_data, 0, 1)
seqParams = seqParams[::-1]
self.seqParams = seqParams
if seqParams[0][0] is None:
self.seqColors = [pg.mkColor('w')]
else:
nSeq = len(seqParams[0][1])
if nSeq == 2:
# Special case: add in difference between two sequence trials
seqParams[0] = (seqParams[0][0], list(seqParams[0][1]) + [np.mean(seqParams[0][1])])
nSeq = 3
img_data = np.empty((3,) + self.img_data.shape[1:], dtype=self.img_data.dtype)
img_data[:2] = self.img_data
for i in range(img_data.shape[1]):
minlen = min(self.img_data[0,i].shape[0], self.img_data[1,i].shape[0])
img_data[2,i] = self.img_data[0,i][:minlen].copy()
img_data[2,i]._data = self.img_data[0,i][:minlen].asarray().astype('float') - self.img_data[1,i][:minlen].asarray()
self.img_data = img_data
self.seqColors = [pg.intColor(i, nSeq*1.6) for i in range(nSeq)]
# cull out truncated recordings :(
self.img_data = [[d['Time':0:200e-3] for d in row if d.xvals('Time')[-1] > 150e-3] for row in self.img_data]
# crop / concatenate
img_len = min([min([d.shape[0] for d in row]) for row in self.img_data])
self.img_data = [[d[:img_len] for d in row] for row in self.img_data]
self.img_arrays = [np.concatenate([d.asarray()[np.newaxis, ...] for d in row], axis=0) for row in self.img_data]
# average
self.img_mean = [img_arr.mean(axis=0) for img_arr in self.img_arrays]
for p in self.clamp_plots:
self.plt2.removeItem(p)
# read all clamp data
first_subdir = seqDir[seqDir.ls()[0]]
clamp_name = 'Clamp1'
if not first_subdir[clamp_name + '.ma'].exists():
clamp_name = 'Clamp2'
clamp_file = first_subdir[clamp_name + '.ma']
if clamp_file.exists():
self.clamp_mode = model.getClampMode(clamp_file)
chan = 'command' if self.clamp_mode == 'VC' else 'primary'
self.clamp_data = model.buildSequenceArray(
seqDir,
lambda dh: dh[clamp_name + '.ma'].read()['Channel': chan],
join=False).asarray()
if self.clamp_data.ndim == 1:
self.clamp_data = self.clamp_data[np.newaxis, :]
if transpose:
self.clamp_data = np.swapaxes(self.clamp_data, 0, 1)
self.plt2.setLabels(left=('Vm', 'V'))
for i in range(self.clamp_data.shape[0]):
for j in range(self.clamp_data.shape[1]):
trace = self.clamp_data[i,j]
pen = self.seqColors[i]
p = self.plt2.plot(trace.xvals('Time'), trace.asarray(), antialias=True, pen=pen)
self.clamp_plots.append(p)
self.plt2.show()
else:
self.clamp_mode = None
self.plt2.hide()
self.img_t = self.img_data[0][0].xvals('Time')
self.img1.setImage(self.img_mean[-1].mean(axis=0))
self.roi_changed(None)
self.time_rgn_changed(None)
self.update_sequence_analysis()
def updateProfiles(self):
#if not self.analyzeBtn.isChecked():
#return
plots = self.getElement('profiles'), self.getElement('profile fits')
for plot in plots:
plot.clear()
plot.setLabel('bottom', 'distance', units='m')
width, height = self.normData.shape
xVals = np.linspace(0, self.px[0] * width, width)
fits = []
def slopeGaussian(v, x): ## gaussian + slope
return fn.gaussian(v[:4], x) + v[4] * x
def gaussError(
v, x, y): ## center-weighted error functionfor sloped gaussian
err = abs(y - slopeGaussian(v, x))
v2 = [2.0, v[1], v[2] * 0.3, 1.0, 0.0]
return err * slopeGaussian(v2, x)
with pg.ProgressDialog("Processing..", 0, height - 1,
cancelText=None) as dlg:
for i in range(height):
row = self.normData[:, i]
guess = [
row.max() - row.min(), xVals[int(width / 2)],
self.px[0] * 3,
row.max(), 0.0
]
#fit = fn.fitGaussian(xVals=xVals, yVals=row, guess=guess)[0]
#fit = fn.fit(slopeGaussian, xVals=xVals, yVals=row, guess=guess)[0]
fit = scipy.optimize.leastsq(gaussError,
guess,
args=(xVals, row))[0]
fit[2] = abs(fit[2])
dist = fit[1] / (self.px[0] * width / 2.)
#print fit, dist
## sanity check on fit
if abs(dist - 1) > 0.5 or (0.5 < fit[3] / np.median(row) >
2.0):
#print "rejected:", fit, fit[3]/np.median(row), self.px[0]*width/2.
#fit = guess[:]
#fit[0] = 0
fit = [0, 0, 0, 0, 0]
else:
# round 2: eliminate anomalous points and re-fit
fitCurve = slopeGaussian(fit, xVals)
diff = row - fitCurve
std = diff.std()
mask = abs(diff) < std * 1.5
x2 = xVals[mask]
y2 = row[mask]
fit = fn.fit(slopeGaussian, xVals=x2, yVals=y2,
guess=fit)[0]
fits.append(fit)
dlg += 1
if dlg.wasCanceled():
raise Exception("Processing canceled by user")
for i in range(len(fits)): ## plot in reverse order
pen = pg.intColor(height - i, height * 1.4)
plots[0].plot(y=self.normData[:, -1 - i], x=xVals, pen=pen)
plots[1].plot(y=slopeGaussian(fits[-1 - i], xVals),
x=xVals,
pen=pen)
yVals = np.linspace(0, self.px[0] * height, height)
arr = np.array(fits)
info = [{
'name': 'depth',
'units': 'm',
'values': yVals
}, {
'name':
'fitParams',
'cols': [
{
'name': 'Amplitude'
},
{
'name': 'X Offset'
},
{
'name': 'Sigma',
'units': 'm'
},
{
'name': 'Y Offset'
},
{
'name': 'Slope'
},
]
}, {
'sourceImage':
self.fileHandle.name(),
'dataRegion':
self.dataRgn.saveState(),
'backgroundRegion':
self.bgRgn.saveState(),
'description':
"""
The source image was normalized for background fluorescence, then each row was fit to a sloped gaussian function:
v[0] * np.exp(-((x-v[1])**2) / (2 * v[2]**2)) + v[3] + v[4] * x
The fit parameters v[0..4] for each image row are stored in the columns of this data set.
"""
}]
#print info
self.data = MetaArray(arr, info=info)
self.showResults(self.data)
def redisplayData(self,
points): ## data must be [(scan, fh, <event time>), ...]
#raise Exception('blah')
#print points
try:
QtGui.QApplication.setOverrideCursor(
QtGui.QCursor(QtCore.Qt.WaitCursor))
plot = self.getElement("Data Plot")
plot.clear()
eTable = self.getElement("Event Table")
sTable = self.getElement("Stats")
#num = len(point.data)
num = len(points)
statList = []
evList = []
for i in range(num):
color = pg.intColor(i, num)
#scan, fh = point.data[i]
try:
scan, fh = points[i][:2]
except:
print points[i]
raise
if len(points[i]) == 3:
evTime = points[i][2]
else:
evTime = None
scan.displayData(fh, plot, color, evTime)
## show stats
stats = scan.getStats(fh.parent())
statList.append(stats)
events = scan.getEvents(fh)['events']
if len(events) > 0:
evList.append(events)
sTable.setData(statList)
if len(evList) > 0:
try:
eTable.setData(np.concatenate(evList))
except:
for i in range(1, len(evList)):
if len(evList[i].dtype) != len(evList[i - 1].dtype):
print "Cannot concatenate; event lists have different dtypes:"
print evList[i].dtype
print evList[i - 1].dtype
else:
for j in range(len(evList[i].dtype)):
if evList[i -
1].dtype[j] != evList[i].dtype[j]:
for l in evList:
print l
print "Warning: can not concatenate--field '%s' has inconsistent types %s, %s (data printed above)" % (
evList[i].dtype.names[j],
str(evList[i - 1].dtype[j]),
str(evList[i].dtype[j]))
raise
finally:
QtGui.QApplication.restoreOverrideCursor()
def updateProfiles(self):
#if not self.analyzeBtn.isChecked():
#return
plots = self.getElement('profiles'), self.getElement('profile fits')
for plot in plots:
plot.clear()
plot.setLabel('bottom', 'distance', units='m')
width, height = self.normData.shape
xVals = np.linspace(0, self.px[0]*width, width)
fits = []
def slopeGaussian(v, x): ## gaussian + slope
return fn.gaussian(v[:4], x) + v[4] * x
def gaussError(v, x, y): ## center-weighted error functionfor sloped gaussian
err = abs(y-slopeGaussian(v, x))
v2 = [2.0, v[1], v[2]*0.3, 1.0, 0.0]
return err * slopeGaussian(v2, x)
with pg.ProgressDialog("Processing..", 0, height-1, cancelText=None) as dlg:
for i in range(height):
row = self.normData[:, i]
guess = [row.max()-row.min(), xVals[int(width/2)], self.px[0]*3, row.max(), 0.0]
#fit = fn.fitGaussian(xVals=xVals, yVals=row, guess=guess)[0]
#fit = fn.fit(slopeGaussian, xVals=xVals, yVals=row, guess=guess)[0]
fit = scipy.optimize.leastsq(gaussError, guess, args=(xVals, row))[0]
fit[2] = abs(fit[2])
dist = fit[1] / (self.px[0] * width / 2.)
#print fit, dist
## sanity check on fit
if abs(dist-1) > 0.5 or (0.5 < fit[3]/np.median(row) > 2.0):
#print "rejected:", fit, fit[3]/np.median(row), self.px[0]*width/2.
#fit = guess[:]
#fit[0] = 0
fit = [0,0,0,0,0]
else:
# round 2: eliminate anomalous points and re-fit
fitCurve = slopeGaussian(fit, xVals)
diff = row - fitCurve
std = diff.std()
mask = abs(diff) < std * 1.5
x2 = xVals[mask]
y2 = row[mask]
fit = fn.fit(slopeGaussian, xVals=x2, yVals=y2, guess=fit)[0]
fits.append(fit)
dlg += 1
if dlg.wasCanceled():
raise Exception("Processing canceled by user")
for i in range(len(fits)): ## plot in reverse order
pen = pg.intColor(height-i, height*1.4)
plots[0].plot(y=self.normData[:, -1-i], x=xVals, pen=pen)
plots[1].plot(y=slopeGaussian(fits[-1-i], xVals), x=xVals, pen=pen)
yVals = np.linspace(0, self.px[0]*height, height)
arr = np.array(fits)
info = [
{'name': 'depth', 'units': 'm', 'values': yVals},
{'name': 'fitParams', 'cols': [
{'name': 'Amplitude'},
{'name': 'X Offset'},
{'name': 'Sigma', 'units': 'm'},
{'name': 'Y Offset'},
{'name': 'Slope'},
]},
{
'sourceImage': self.fileHandle.name(),
'dataRegion': self.dataRgn.saveState(),
'backgroundRegion': self.bgRgn.saveState(),
'description': """
The source image was normalized for background fluorescence, then each row was fit to a sloped gaussian function:
v[0] * np.exp(-((x-v[1])**2) / (2 * v[2]**2)) + v[3] + v[4] * x
The fit parameters v[0..4] for each image row are stored in the columns of this data set.
"""
}
]
#print info
self.data = MetaArray(arr, info=info)
self.showResults(self.data)
