def analyseFile(filename):
print(filename)
seriesName = os.path.splitext(os.path.split(filename)[-1])[0]
PL.ExtendContext({'seriesName': seriesName})
try:
pipe = Pipeline(filename)
except RuntimeError:
print(('Error opening %s' % filename))
PL.PopContext()
return
#only look at first 7k frames
#pipe.filterKeys['t'] = (0, 7000)
pipe.Rebuild()
autoDrift.correctDrift(pipe)
imb = visHelpers.ImageBounds.estimateFromSource(pipe)
jitVals = 1.0 * pipe['error_x']
im = visHelpers.rendJitTriang(pipe['x'], pipe['y'], 20, jitVals, 1, imb, 5)
pipe.CloseFiles()
PL.PopContext()
def colfcnwrap(pipeline):
colourFilter = pipeline.colourFilter
metadata = pipeline.mdh
chans = colourFilter.getColourChans()
if USE_GUI:
figure(os.path.split(pipeline.filename)[-1] + ' - ' + fcn.__name__)
if len(chans) == 0:
fcn(colourFilter, metadata)
else:
curChan = colourFilter.currentColour
if 'rng' in args:
nPh = getPhotonNums(colourFilter, metadata)
rng = 6 * nPh.mean()
chanNames = chans[:]
if 'Sample.Labelling' in metadata.getEntryNames():
lab = metadata.getEntry('Sample.Labelling')
for i in range(len(lab)):
if lab[i][0] in chanNames:
chanNames[chanNames.index(lab[i][0])] = lab[i][1]
for ch, i in zip(chans, range(len(chans))):
colourFilter.setColour(ch)
PL.ExtendContext({'chan': chanNames[i]})
if 'rng' in args:
fcn(colourFilter, metadata, chanNames[i], i, rng)
else:
fcn(colourFilter, metadata, chanNames[i], i)
PL.PopContext()
colourFilter.setColour(curChan)
def fitOnTimes(colourFilter, metadata, channame='', i=0):
numPerClump = colourFilter['clumpSize']
n, bins = np.histogram(numPerClump, np.arange(20) + .001)
n = n / np.arange(1, 20)
cycTime = metadata.getEntry('Camera.CycleTime')
res = FitModelWeighted(eimod, [n[0], .2], n[1:], 1. / (np.sqrt(n[1:]) + 1),
bins[2:] * cycTime)
#res = FitModelPoisson(eimod, [n[0], .2], n[1:], bins[2:]*cycTime)
ch2, mse = chi2_mse(eimod, n[1:], res, bins[2:] * cycTime)
#mse = (res[2]['fvec']**2).mean()
res2 = FitModelWeighted(ei2mod, [n[0], .2], n[1:],
1. / (np.sqrt(n[1:]) + 1), bins[2:] * cycTime,
cycTime)
ch22, mse2 = chi2_mse(ei2mod, n[1:], res2, bins[2:] * cycTime, cycTime)
PL.AddRecord('/Photophysics/OnTimes/eimod', munge_res(eimod, res, mse=mse))
PL.AddRecord('/Photophysics/OnTimes/ei2mod',
munge_res(ei2mod, res2, mse=mse2))
#print res[0]
#figure()
if USE_GUI:
pylab.semilogy()
pylab.bar(bins[:-1] * cycTime,
n,
width=cycTime,
alpha=0.4,
fc=colours[i])
pylab.plot(np.linspace(1, 20, 50) * cycTime,
ei2mod(res2[0],
np.linspace(1, 20, 50) * cycTime, cycTime),
colours[i],
lw=3,
ls='--')
pylab.plot(np.linspace(1, 20, 50) * cycTime,
eimod(res[0],
np.linspace(1, 20, 50) * cycTime),
colours[i],
lw=3)
pylab.ylabel('Events')
pylab.xlabel('Event Duration [s]')
pylab.ylim((1, pylab.ylim()[1]))
pylab.title(
'Event Duration - CAUTION: unreliable if $\\tau <\\sim$ exposure time'
)
pylab.figtext(.6,
.8 - .05 * i,
channame + '\t$\\tau = %3.4fs$' % (res[0][1], ),
size=18,
color=colours[i])
return 0
def fitDecay(colourFilter, metadata, channame='', i=0):
#get frames in which events occured and convert into seconds
t = colourFilter['t'] * metadata.getEntry('Camera.CycleTime')
n, bins = histogram(t, 100)
b1 = bins[:-1]
Nm = n.max()
res = FitModel(e2mod, [Nm * 2, 15, -3, Nm * 3, n[1] / 1], n[1:], b1[1:],
Nm)
#mse = (res[2]['fvec']**2).mean()
#ch2 = chi2(res, n[1:])
#print ch2
ch2, mse = chi2_mse(e2mod, n[1:], res, b1[1:], Nm)
PL.AddRecord('/Photophysics/Decay/e2mod',
munge_res(e2mod, res, mse=mse, ch2=ch2))
res2 = FitModelPoisson(emod, [Nm * 2, 15, -3, Nm * 3, n[1] / 2], n[1:],
b1[1:], Nm) #[0]
ch2, mse = chi2_mse(hmod, n[1:], res2, b1[1:], Nm)
PL.AddRecord('/Photophysics/Decay/emod',
munge_res(emod, res2, mse=mse, ch2=ch2))
res3 = FitModelPoisson(hmod, [Nm * 2, 15, -3, Nm * 3, n[1] / 2], n[1:],
b1[1:], Nm) #[0]
ch2, mse = chi2_mse(hmod, n[1:], res3, b1[1:], Nm)
PL.AddRecord('/Photophysics/Decay/hmod',
munge_res(hmod, res3, mse=mse, ch2=ch2))
r4 = FitModelPoisson(e2mod, [Nm * 2, 15, -3, Nm * 3, n[1] / 2], n[1:],
b1[1:], Nm)
ch2, mse = chi2_mse(e2mod, n[1:], r4, b1[1:], Nm)
PL.AddRecord('/Photophysics/Decay/e2mod_p',
munge_res(e2mod, r4, mse=mse, ch2=ch2))
if USE_GUI:
bar(b1 / 60, n, width=(b1[1] - b1[0]) / 60, alpha=0.4, fc=colours[i])
plot(b1 / 60, e2mod(res[0], b1, Nm), colours[i], lw=3)
plot(b1 / 60, emod(res2[0], b1, Nm), colours[i], lw=2, ls='--')
plot(b1 / 60, hmod(res3[0], b1, Nm), colours[i], lw=2, ls=':')
plot(b1 / 60, e2mod(r4[0], b1, Nm), colours[i], lw=1)
ylabel('Events')
xlabel('Acquisition Time [mins]')
title('Event Rate')
b = 0.5 * (1 + erf(res[0][2])) * Nm
figtext(.4,
.8 - .05 * i,
channame + '\t$\\tau = %3.2fs,\\;b = %3.2f$' %
(res[0][1], b / res[0][0]),
size=18,
color=colours[i])
def analyseFile(filename):
print(filename)
seriesName = os.path.splitext(os.path.split(filename)[-1])[0]
PL.ExtendContext({'seriesName':seriesName})
try:
pipe = Pipeline(filename)
except RuntimeError:
print(('Error opening %s' % filename))
PL.PopContext()
return
#only look at first 7k frames
pipe.filterKeys['t'] = (0, 7000)
pipe.Rebuild()
trackUtils.findTracks(pipe, 'error_x', 2, 20)
pipe.Rebuild()
extraParams = {}
extraParams['cycleTime'] = pipe.mdh['Camera.CycleTime']
nPhot = kinModels.getPhotonNums(pipe.colourFilter, pipe.mdh)
extraParams['MedianPhotons'] = np.median(nPhot)
extraParams['MeanPhotons'] = np.mean(nPhot)
extraParams['NEvents'] = len(nPhot)
extraParams['MeanBackground'] = pipe['fitResults_background'].mean() - pipe.mdh['Camera.ADOffset']
extraParams['MedianBackground'] = np.median(pipe['fitResults_background']) - pipe.mdh['Camera.ADOffset']
extraParams['MeanClumpSize'] = pipe['clumpSize'].mean()
extraParams['MeanClumpPhotons'] = (pipe['clumpSize']*nPhot).mean()
PL.AddRecord('/Photophysics/ExtraParams', dictToRecarray(extraParams))
kinModels.fitDecay(pipe)
kinModels.fitFluorBrightness(pipe)
#kinModels.fitFluorBrightnessT(pipe)
#max_off_ts = [3,5,10,20,40]
#max_off_ts = [20]
#for ot in max_off_ts:
#PL.ExtendContext({'otMax':ot})
#find molecules appearing across multiple frames
kinModels.fitOnTimes(pipe)
#PL.PopContext()
pipe.CloseFiles()
PL.PopContext()
def processDir(dataPath, outfile):
#set the logging backend to the PyTables file
TB = TablesBackend(outfile)
PL.SetBackend(TB)
#walk the path and analyse all .h5r
for (path, dirs, files) in os.walk(dataPath):
for f in files:
if os.path.splitext(f)[1] in ['.h5r']:
filename = os.path.join(path, f)
analyseFile(filename)
def fitFluorBrightness(colourFilter,
metadata,
channame='',
i=0,
rng=None,
quiet=False):
#nPh = (colourFilter['A']*2*math.pi*(colourFilter['sig']/(1e3*metadata.getEntry('voxelsize.x')))**2)
#nPh = nPh*metadata.getEntry('Camera.ElectronsPerCount')/metadata.getEntry('Camera.TrueEMGain')
nPh = getPhotonNums(colourFilter, metadata)
if rng is None:
rng = nPh.mean() * 6
n, bins = np.histogram(nPh, np.linspace(0, rng, 100))
NEvents = len(colourFilter['t'])
bins = bins[:-1]
res = FitModel(fImod, [n.max(), bins[n.argmax()] / 2, 100,
nPh.mean()], n, bins)
mse = (res[2]['fvec']**2).mean()
if not quiet:
PL.AddRecord('/Photophysics/FluorBrightness/fImod',
munge_res(fImod, res, mse=mse))
#figure()
#semilogy()
if USE_GUI:
plt.bar(bins, n, width=bins[1] - bins[0], alpha=0.4, fc=colours[i])
plt.plot(bins, fImod(res[0], bins), colours[i], lw=3)
plt.ylabel('Events')
plt.xlabel('Intensity [photons]')
#ylim((1, ylim()[1]))
plt.title(
'Event Intensity - CAUTION - unreliable if evt. duration $>\\sim$ exposure time'
)
#print res[0][2]
plt.figtext(
.4,
.8 - .15 * i,
channame +
'\t$N_{det} = %3.0f\\;\\lambda = %3.0f$\n\t$Ph.mean = %3.0f$' %
(res[0][1], res[0][3], nPh.mean()),
size=18,
color=colours[i],
verticalalignment='top',
horizontalalignment='left')
return [channame, res[0][3], NEvents]
def fitFluorBrightnessT(colourFilter, metadata, channame='', i=0, rng=None):
#nPh = (colourFilter['A']*2*math.pi*(colourFilter['sig']/(1e3*metadata.getEntry('voxelsize.x')))**2)
#nPh = nPh*metadata.getEntry('Camera.ElectronsPerCount')/metadata.getEntry('Camera.TrueEMGain')
#from mpl_toolkits.mplot3d import Axes3D
nPh = getPhotonNums(colourFilter, metadata)
t = (colourFilter['t'].astype('f') - metadata['Protocol.DataStartsAt']
) * metadata.getEntry('Camera.CycleTime')
NEvents = len(t)
if rng is None:
rng = nPh.mean() * 3
Nco = nPh.min()
n, xbins, ybins = np.histogram2d(
nPh, t, [np.linspace(0, rng, 50),
np.linspace(0, t.max(), 20)])
bins = xbins[:-1]
xb = xbins[:-1][:, None] * np.ones([1, ybins.size - 1])
yb = ybins[:-1][None, :] * np.ones([xbins.size - 1, 1])
res0 = FitModel(
fITmod2,
[n.max() * 3, 1, np.median(nPh), 20, 1e2, 1e2, 100], n, xb, yb, Nco)
print((res0[0]))
PL.AddRecord('/Photophysics/FluorBrightness/fITmod2',
munge_res(fITmod2, res0))
A, Ndet, lamb, tauI, a, Acrit, bg = res0[0]
#Ndet = Ndet**2
#NDetM = NDetM**2
#Acrit = Acrit**2
#a = (1+erf(a))/2
Ndet = np.sqrt(Ndet**2 + 1) - 1 #+ve
bg = np.sqrt(bg**2 + 1) - 1
Acrit = np.sqrt(Acrit**2 + 1) - 1
a = (1 + erf(a)) / 2 # [0,1]
#bg = sqrt(bg**2 + 1) - 1
#k = sqrt(k**2 + 1) - 1
NDetM = bg
rr = fITmod2(res0[0], xb, yb, Nco)
if USE_GUI:
pylab.figure()
pylab.subplot(131)
pylab.imshow(n, interpolation='nearest')
pylab.colorbar()
pylab.subplot(132)
pylab.imshow(rr, interpolation='nearest')
pylab.colorbar()
pylab.subplot(133)
pylab.imshow(n - rr, interpolation='nearest')
pylab.colorbar()
pylab.title(channame)
pylab.figure()
t_ = np.linspace(t[0], t[-1], 100)
#sc = (lamb/(ybins[1] - ybins[0]))
#sc = len(ybins)
sc = 1. / (1 - np.exp(-(ybins[1] - ybins[0]) / lamb))
print(('sc = ', sc))
y1 = sc * A / ((t_ / tauI)**a + 1)
pylab.plot(t_, y1)
pylab.plot(t_, sc * (Ndet / ((t_ / tauI)**a + 1) + NDetM))
pylab.bar(ybins[:-1], n.sum(0), width=ybins[1] - ybins[0], alpha=0.5)
pylab.plot(ybins[:-1], rr.sum(0), lw=2)
pylab.title(channame)
pylab.xlabel('Time [s]')
pylab.figtext(
.2,
.7,
'$A = %3.0f\\;N_{det} = %3.2f\\;\\lambda = %3.0f\\;\\tau = %3.0f$\n$\\alpha = %3.3f\\;A_{crit} = %3.2f\\;N_{det_0} = %3.2f$'
% (A, Ndet, lamb, tauI, a, Acrit, NDetM),
size=18)
return [channame, lamb, NEvents]
