def tospectrum(self,
signal,
fwhm=0.1,
forceFwhm=True,
autoAlign=True,
iterLimit=None,
relThreshold=0.,
pickingHeight=0.90,
baseline=None):
"""Fit modeled profiles to spectrum using tmp peaklist.
signal (numpy array) - m/z / intensity pairs
fwhm (float) - defaut fwhm
forceFwhm (bool) - use default fwhm
autoAlign (bool) - automatic m/z shift
iterLimit (int) - maximum number of iterations
pickingHeight (float) - peak picking height for centroiding
relThreshold (float) - relative intensity threshold
baseline (numpy array) - signal baseline
"""
# crop signal to relevant m/z range
i1 = mod_signal.locate(signal, self.mzrange[0])
i2 = mod_signal.locate(signal, self.mzrange[1])
signal = signal[i1:i2]
# get peaklist from signal
peaklist = mod_peakpicking.labelscan(signal=signal,
pickingHeight=pickingHeight,
relThreshold=relThreshold,
baseline=baseline)
# remove shoulder peaks
peaklist.remshoulders(fwhm=fwhm)
# correct signal baseline
if baseline != None:
self.spectrum = mod_signal.subbase(signal, baseline)
# fit to peaklist
return self.topeaklist(
peaklist=peaklist,
fwhm=fwhm,
forceFwhm=forceFwhm,
autoAlign=autoAlign,
iterLimit=iterLimit,
relThreshold=relThreshold,
)
def labelpoint(signal, mz, baseline=None):
"""Return labeled peak at given x-value.
signal (numpy array) - signal data points
mz (float) - x-value to label
baseline (numpy array) - signal baseline
"""
# check signal type
if not isinstance(signal, numpy.ndarray):
raise TypeError, "Signal must be NumPy array!"
# check baseline type
if baseline != None and not isinstance(baseline, numpy.ndarray):
raise TypeError, "Baseline must be NumPy array!"
# check signal data
if len(signal) == 0:
return None
# check m/z value
if mz <= 0:
return None
# get peak intensity
ai = mod_signal.intensity(signal, mz)
if not ai:
return None
# get peak baseline and s/n
base = 0.0
sn = None
if baseline == None:
base, noise = mod_signal.noise(signal, x=mz)
if noise:
sn = (ai - base) / noise
else:
idx = mod_signal.locate(baseline, mz)
if (idx > 0) and (idx < len(baseline)):
base = mod_signal.interpolate( (baseline[idx-1][0], baseline[idx-1][1]), (baseline[idx][0], baseline[idx][1]), x=mz)
noise = mod_signal.interpolate( (baseline[idx-1][0], baseline[idx-1][2]), (baseline[idx][0], baseline[idx][2]), x=mz)
if noise:
sn = (ai - base) / noise
# check peak intensity
if ai <= base:
return None
# get peak fwhm
height = base + (ai - base) * 0.5
fwhm = mod_signal.width(signal, mz, height)
# make peak object
peak = obj_peak.peak(mz=mz, ai=ai, base=base, sn=sn, fwhm=fwhm)
return peak
def labelpoint(signal, mz, baseline=None):
"""Return labeled peak at given x-value.
signal (numpy array) - signal data points
mz (float) - x-value to label
baseline (numpy array) - signal baseline
"""
# check signal type
if not isinstance(signal, numpy.ndarray):
raise TypeError, "Signal must be NumPy array!"
# check baseline type
if baseline != None and not isinstance(baseline, numpy.ndarray):
raise TypeError, "Baseline must be NumPy array!"
# check signal data
if len(signal) == 0:
return None
# check m/z value
if mz <= 0:
return None
# get peak intensity
ai = mod_signal.intensity(signal, mz)
if not ai:
return None
# get peak baseline and s/n
base = 0.0
sn = None
if baseline == None:
base, noise = mod_signal.noise(signal, x=mz)
if noise:
sn = (ai - base) / noise
else:
idx = mod_signal.locate(baseline, mz)
if (idx > 0) and (idx < len(baseline)):
base = mod_signal.interpolate( (baseline[idx-1][0], baseline[idx-1][1]), (baseline[idx][0], baseline[idx][1]), x=mz)
noise = mod_signal.interpolate( (baseline[idx-1][0], baseline[idx-1][2]), (baseline[idx][0], baseline[idx][2]), x=mz)
if noise:
sn = (ai - base) / noise
# check peak intensity
if ai <= base:
return None
# get peak fwhm
height = base + (ai - base) * 0.5
fwhm = mod_signal.width(signal, mz, height)
# make peak object
peak = obj_peak.peak(mz=mz, ai=ai, base=base, sn=sn, fwhm=fwhm)
return peak
def tospectrum(self, signal, fwhm=0.1, forceFwhm=True, autoAlign=True, iterLimit=None, relThreshold=0., pickingHeight=0.90, baseline=None):
"""Fit modeled profiles to spectrum using tmp peaklist.
signal (numpy array) - m/z / intensity pairs
fwhm (float) - defaut fwhm
forceFwhm (bool) - use default fwhm
autoAlign (bool) - automatic m/z shift
iterLimit (int) - maximum number of iterations
pickingHeight (float) - peak picking height for centroiding
relThreshold (float) - relative intensity threshold
baseline (numpy array) - signal baseline
"""
# crop signal to relevant m/z range
i1 = mod_signal.locate(signal, self.mzrange[0])
i2 = mod_signal.locate(signal, self.mzrange[1])
signal = signal[i1:i2]
# get peaklist from signal
peaklist = mod_peakpicking.labelscan(
signal = signal,
pickingHeight = pickingHeight,
relThreshold = relThreshold,
baseline = baseline
)
# remove shoulder peaks
peaklist.remshoulders(fwhm=fwhm)
# correct signal baseline
if baseline != None:
self.spectrum = mod_signal.subbase(signal, baseline)
# fit to peaklist
return self.topeaklist(
peaklist = peaklist,
fwhm = fwhm,
forceFwhm = forceFwhm,
autoAlign = autoAlign,
iterLimit = iterLimit,
relThreshold = relThreshold,
)
def labelscan(signal, minX=None, maxX=None, pickingHeight=0.75, absThreshold=0., relThreshold=0., snThreshold=0., baseline=None):
"""Return centroided peaklist for given data points.
signal (numpy array) - signal data points
minX (float) - x-range start
maxX (float) - x-range end
pickingHeight (float) - centroiding height
absThreshold (float) - absolute intensity threshold
relThreshold (float) - relative intensity threshold
snThreshold (float) - signal to noise threshold
baseline (numpy array) - signal baseline
"""
# check signal type
if not isinstance(signal, numpy.ndarray):
raise TypeError, "Signal must be NumPy array!"
# check baseline type
if baseline != None and not isinstance(baseline, numpy.ndarray):
raise TypeError, "Baseline must be NumPy array!"
# crop data
if minX != None and maxX != None:
i1 = mod_signal.locate(signal, minX)
i2 = mod_signal.locate(signal, maxX)
signal = signal[i1:i2]
# check data points
if len(signal) == 0:
return obj_peaklist.peaklist([])
# get local maxima
buff = []
basepeak = mod_signal.basepeak(signal)
threshold = max(signal[basepeak][1] * relThreshold, absThreshold)
for peak in mod_signal.maxima(signal):
if peak[1] >= threshold:
buff.append( [peak[0], peak[1], 0., None, None] ) # mz, ai, base, sn, fwhm
CHECK_FORCE_QUIT()
# get peaks baseline and s/n
basepeak = 0.0
if baseline != None:
for peak in buff:
idx = mod_signal.locate(baseline, peak[0])
if (idx > 0) and (idx < len(baseline)):
p1 = baseline[idx-1]
p2 = baseline[idx]
peak[2] = mod_signal.interpolate( (p1[0], p1[1]), (p2[0], p2[1]), x=peak[0])
noise = mod_signal.interpolate( (p1[0], p1[2]), (p2[0], p2[2]), x=peak[0])
intens = peak[1] - peak[2]
if noise:
peak[3] = intens / noise
if intens > basepeak:
basepeak = intens
CHECK_FORCE_QUIT()
# remove peaks bellow threshold
threshold = max(basepeak * relThreshold, absThreshold)
candidates = []
for peak in buff:
if peak[0] > 0 and (peak[1] - peak[2]) >= threshold and (not peak[3] or peak[3] >= snThreshold):
candidates.append(peak)
# make centroides
if pickingHeight < 1.:
buff = []
previous = None
for peak in candidates:
CHECK_FORCE_QUIT()
# calc peak height
h = ((peak[1]-peak[2]) * pickingHeight) + peak[2]
# get centroid indexes
idx = mod_signal.locate(signal, peak[0])
if (idx == 0) or (idx == len(signal)):
continue
ileft = idx-1
while (ileft > 0) and (signal[ileft][1] > h):
ileft -= 1
iright = idx
while (iright < len(signal)-1) and (signal[iright][1] > h):
iright += 1
# calculate peak mz
leftMZ = mod_signal.interpolate(signal[ileft], signal[ileft+1], y=h)
rightMZ = mod_signal.interpolate(signal[iright-1], signal[iright], y=h)
peak[0] = (leftMZ + rightMZ)/2.
# get peak intensity
intens = mod_signal.intensity(signal, peak[0])
if intens and intens <= peak[1]:
peak[1] = intens
else:
continue
# try to group with previous peak
if previous != None and leftMZ < previous:
if peak[1] > buff[-1][1]:
buff[-1] = peak
previous = rightMZ
else:
buff.append(peak)
previous = rightMZ
# store as candidates
candidates = buff
CHECK_FORCE_QUIT()
# get peaks baseline and s/n
basepeak = 0.0
if baseline != None:
for peak in candidates:
idx = mod_signal.locate(baseline, peak[0])
if (idx > 0) and (idx < len(baseline)):
p1 = baseline[idx-1]
p2 = baseline[idx]
peak[2] = mod_signal.interpolate( (p1[0], p1[1]), (p2[0], p2[1]), x=peak[0])
noise = mod_signal.interpolate( (p1[0], p1[2]), (p2[0], p2[2]), x=peak[0])
intens = peak[1] - peak[2]
if noise:
peak[3] = intens / noise
if intens > basepeak:
basepeak = intens
CHECK_FORCE_QUIT()
# remove peaks bellow threshold and calculate fwhm
threshold = max(basepeak * relThreshold, absThreshold)
centroides = []
for peak in candidates:
if peak[0] > 0 and (peak[1] - peak[2]) >= threshold and (not peak[3] or peak[3] >= snThreshold):
peak[4] = mod_signal.width(signal, peak[0], (peak[2] + ((peak[1] - peak[2]) * 0.5)))
centroides.append(obj_peak.peak(mz=peak[0], ai=peak[1], base=peak[2], sn=peak[3], fwhm=peak[4]))
# return peaklist object
return obj_peaklist.peaklist(centroides)
def labelpeak(signal, mz=None, minX=None, maxX=None, pickingHeight=0.75, baseline=None):
"""Return labeled peak in given m/z range.
signal (numpy array) - signal data points
mz (float) - x-value to label
minX (float) - x-range start
maxX (float) - x-range end
pickingHeight (float) - centroiding height
baseline (numpy array) - signal baseline
"""
# check signal type
if not isinstance(signal, numpy.ndarray):
raise TypeError, "Signal must be NumPy array!"
# check baseline type
if baseline != None and not isinstance(baseline, numpy.ndarray):
raise TypeError, "Baseline must be NumPy array!"
# check m/z value or range
if mz == None and minX == None and maxX == None:
raise TypeError, "m/z value or range must be specified!"
# check signal data
if len(signal) == 0:
return None
# check m/z value
if mz != None:
minX = mz
if minX <= 0:
return False
# get index of given m/z or range maximum
if mz != None:
imax = mod_signal.locate(signal, mz)
else:
i1 = mod_signal.locate(signal, minX)
i2 = mod_signal.locate(signal, maxX)
imax = i1
if i1 != i2:
imax += mod_signal.basepeak(signal[i1:i2])
if (imax == 0) or (imax == len(signal)):
return None
# get centroid height
h = signal[imax][1] * pickingHeight
if baseline != None:
idx = mod_signal.locate(baseline, signal[imax][0])
if (idx > 0) and (idx < len(baseline)):
base = mod_signal.interpolate( (baseline[idx-1][0], baseline[idx-1][1]), (baseline[idx][0], baseline[idx][1]), x=signal[imax][0])
h = ((signal[imax][1] - base) * pickingHeight) + base
# get centroid
ileft = imax-1
while (ileft > 0) and (signal[ileft][1] > h):
ileft -= 1
iright = imax
while (iright < len(signal)-1) and (signal[iright][1] > h):
iright += 1
leftMZ = mod_signal.interpolate(signal[ileft], signal[ileft+1], y=h)
rightMZ = mod_signal.interpolate(signal[iright-1], signal[iright], y=h)
# check range
if mz == None and (leftMZ < minX or rightMZ > maxX) and (leftMZ != rightMZ):
return None
# label peak in the newly found selection
if mz != None and leftMZ != rightMZ:
peak = labelpeak(
signal = signal,
minX = leftMZ,
maxX = rightMZ,
pickingHeight = pickingHeight,
baseline = baseline
)
# label current point
else:
peak = labelpoint(
signal = signal,
mz = ((leftMZ + rightMZ)/2.),
baseline = baseline
)
return peak
def topoints(self, points, fwhm=0.1, autoAlign=True, iterLimit=None):
"""Fit modeled profiles to given points.
points (numpy array or list) - m/z / intensity pairs
fwhm (float) - defaut fwhm
autoAlign (bool) - automatic m/z shift
iterLimit (int) - maximum number of iterations
"""
self.fwhm = fwhm
# reset previous results
self.composition = {}
self.ncomposition = {}
self.average = 0.
self.model = numpy.array([])
for x in self.models:
self.models[x][2] = 0.0 # abs comp
self.models[x][3] = 0.0 # rel comp
# check points type
if not isinstance(points, numpy.ndarray):
points = numpy.array(points)
# crop points to relevant m/z range
i1 = mod_signal.locate(points, self.mzrange[0])
i2 = mod_signal.locate(points, self.mzrange[1])
self.data = numpy.array(points[i1:i2])
# check data
if len(self.data) == 0:
return False
# auto align data and theoretical pattern
if autoAlign:
self._alignData()
# split data to raster and intensities
xAxis, yAxis = numpy.hsplit(self.data, 2)
raster = xAxis.flatten()
intensities = yAxis.flatten()
# model profiles
models, exchanged = self._makeModels(raster, reset=False)
# fit data to models
fit = self._leastSquare(intensities, models, iterLimit=iterLimit)
if not sum(fit):
return False
f = 1. / sum(fit)
for i, abundance in enumerate(fit):
self.models[exchanged[i]][2] = abundance
self.models[exchanged[i]][3] = f * abundance
# calc average exchange
for x in self.models:
self.average += x * self.models[x][3]
# get compositions
for x in sorted(self.models.keys()):
self.composition[x] = self.models[x][2]
self.ncomposition[x] = self.models[x][3]
# get calculated points
raster.shape = (-1, 1)
intensities = numpy.sum(models * [[x] for x in fit], axis=0)
intensities.shape = (-1, 1)
self.model = numpy.concatenate((raster, intensities), axis=1).copy()
return True
def topoints(self, points, fwhm=0.1, autoAlign=True, iterLimit=None):
"""Fit modeled profiles to given points.
points (numpy array or list) - m/z / intensity pairs
fwhm (float) - defaut fwhm
autoAlign (bool) - automatic m/z shift
iterLimit (int) - maximum number of iterations
"""
self.fwhm = fwhm
# reset previous results
self.composition = {}
self.ncomposition = {}
self.average = 0.
self.model = numpy.array([])
for x in self.models:
self.models[x][2] = 0.0 # abs comp
self.models[x][3] = 0.0 # rel comp
# check points type
if not isinstance(points, numpy.ndarray):
points = numpy.array(points)
# crop points to relevant m/z range
i1 = mod_signal.locate(points, self.mzrange[0])
i2 = mod_signal.locate(points, self.mzrange[1])
self.data = numpy.array(points[i1:i2])
# check data
if len(self.data) == 0:
return False
# auto align data and theoretical pattern
if autoAlign:
self._alignData()
# split data to raster and intensities
xAxis, yAxis = numpy.hsplit(self.data, 2)
raster = xAxis.flatten()
intensities = yAxis.flatten()
# model profiles
models, exchanged = self._makeModels(raster, reset=False)
# fit data to models
fit = self._leastSquare(intensities, models, iterLimit=iterLimit)
if not sum(fit):
return False
f = 1./sum(fit)
for i, abundance in enumerate(fit):
self.models[exchanged[i]][2] = abundance
self.models[exchanged[i]][3] = f*abundance
# calc average exchange
for x in self.models:
self.average += x * self.models[x][3]
# get compositions
for x in sorted(self.models.keys()):
self.composition[x] = self.models[x][2]
self.ncomposition[x] = self.models[x][3]
# get calculated points
raster.shape = (-1,1)
intensities = numpy.sum(models * [[x] for x in fit], axis=0)
intensities.shape = (-1,1)
self.model = numpy.concatenate((raster, intensities), axis=1).copy()
return True
def labelscan(signal, minX=None, maxX=None, pickingHeight=0.75, absThreshold=0., relThreshold=0., snThreshold=0., baseline=None):
"""Return centroided peaklist for given data points.
signal (numpy array) - signal data points
minX (float) - x-range start
maxX (float) - x-range end
pickingHeight (float) - centroiding height
absThreshold (float) - absolute intensity threshold
relThreshold (float) - relative intensity threshold
snThreshold (float) - signal to noise threshold
baseline (numpy array) - signal baseline
"""
# check signal type
if not isinstance(signal, numpy.ndarray):
raise TypeError, "Signal must be NumPy array!"
# check baseline type
if baseline != None and not isinstance(baseline, numpy.ndarray):
raise TypeError, "Baseline must be NumPy array!"
# crop data
if minX != None and maxX != None:
i1 = mod_signal.locate(signal, minX)
i2 = mod_signal.locate(signal, maxX)
signal = signal[i1:i2]
# check data points
if len(signal) == 0:
return obj_peaklist.peaklist([])
# get local maxima
buff = []
basepeak = mod_signal.basepeak(signal)
threshold = max(signal[basepeak][1] * relThreshold, absThreshold)
for peak in mod_signal.maxima(signal):
if peak[1] >= threshold:
buff.append( [peak[0], peak[1], 0., None, None] ) # mz, ai, base, sn, fwhm
CHECK_FORCE_QUIT()
# get peaks baseline and s/n
basepeak = 0.0
if baseline != None:
for peak in buff:
idx = mod_signal.locate(baseline, peak[0])
if (idx > 0) and (idx < len(baseline)):
p1 = baseline[idx-1]
p2 = baseline[idx]
peak[2] = mod_signal.interpolate( (p1[0], p1[1]), (p2[0], p2[1]), x=peak[0])
noise = mod_signal.interpolate( (p1[0], p1[2]), (p2[0], p2[2]), x=peak[0])
intens = peak[1] - peak[2]
if noise:
peak[3] = intens / noise
if intens > basepeak:
basepeak = intens
CHECK_FORCE_QUIT()
# remove peaks bellow threshold
threshold = max(basepeak * relThreshold, absThreshold)
candidates = []
for peak in buff:
if peak[0] > 0 and (peak[1] - peak[2]) >= threshold and (not peak[3] or peak[3] >= snThreshold):
candidates.append(peak)
# make centroides
if pickingHeight < 1.:
buff = []
previous = None
for peak in candidates:
CHECK_FORCE_QUIT()
# calc peak height
h = ((peak[1]-peak[2]) * pickingHeight) + peak[2]
# get centroid indexes
idx = mod_signal.locate(signal, peak[0])
if (idx == 0) or (idx == len(signal)):
continue
ileft = idx-1
while (ileft > 0) and (signal[ileft][1] > h):
ileft -= 1
iright = idx
while (iright < len(signal)-1) and (signal[iright][1] > h):
iright += 1
# calculate peak mz
leftMZ = mod_signal.interpolate(signal[ileft], signal[ileft+1], y=h)
rightMZ = mod_signal.interpolate(signal[iright-1], signal[iright], y=h)
peak[0] = (leftMZ + rightMZ)/2.
# get peak intensity
intens = mod_signal.intensity(signal, peak[0])
if intens and intens <= peak[1]:
peak[1] = intens
else:
continue
# try to group with previous peak
if previous != None and leftMZ < previous:
if peak[1] > buff[-1][1]:
buff[-1] = peak
previous = rightMZ
else:
buff.append(peak)
previous = rightMZ
# store as candidates
candidates = buff
CHECK_FORCE_QUIT()
# get peaks baseline and s/n
basepeak = 0.0
if baseline != None:
for peak in candidates:
idx = mod_signal.locate(baseline, peak[0])
if (idx > 0) and (idx < len(baseline)):
p1 = baseline[idx-1]
p2 = baseline[idx]
peak[2] = mod_signal.interpolate( (p1[0], p1[1]), (p2[0], p2[1]), x=peak[0])
noise = mod_signal.interpolate( (p1[0], p1[2]), (p2[0], p2[2]), x=peak[0])
intens = peak[1] - peak[2]
if noise:
peak[3] = intens / noise
if intens > basepeak:
basepeak = intens
CHECK_FORCE_QUIT()
# remove peaks bellow threshold and calculate fwhm
threshold = max(basepeak * relThreshold, absThreshold)
centroides = []
for peak in candidates:
if peak[0] > 0 and (peak[1] - peak[2]) >= threshold and (not peak[3] or peak[3] >= snThreshold):
peak[4] = mod_signal.width(signal, peak[0], (peak[2] + ((peak[1] - peak[2]) * 0.5)))
centroides.append(obj_peak.peak(mz=peak[0], ai=peak[1], base=peak[2], sn=peak[3], fwhm=peak[4]))
# return peaklist object
return obj_peaklist.peaklist(centroides)
def labelpeak(signal, mz=None, minX=None, maxX=None, pickingHeight=0.75, baseline=None):
"""Return labeled peak in given m/z range.
signal (numpy array) - signal data points
mz (float) - x-value to label
minX (float) - x-range start
maxX (float) - x-range end
pickingHeight (float) - centroiding height
baseline (numpy array) - signal baseline
"""
# check signal type
if not isinstance(signal, numpy.ndarray):
raise TypeError, "Signal must be NumPy array!"
# check baseline type
if baseline != None and not isinstance(baseline, numpy.ndarray):
raise TypeError, "Baseline must be NumPy array!"
# check m/z value or range
if mz == None and minX == None and maxX == None:
raise TypeError, "m/z value or range must be specified!"
# check signal data
if len(signal) == 0:
return None
# check m/z value
if mz != None:
minX = mz
if minX <= 0:
return False
# get index of given m/z or range maximum
if mz != None:
imax = mod_signal.locate(signal, mz)
else:
i1 = mod_signal.locate(signal, minX)
i2 = mod_signal.locate(signal, maxX)
imax = i1
if i1 != i2:
imax += mod_signal.basepeak(signal[i1:i2])
if (imax == 0) or (imax == len(signal)):
return None
# get centroid height
h = signal[imax][1] * pickingHeight
if baseline != None:
idx = mod_signal.locate(baseline, signal[imax][0])
if (idx > 0) and (idx < len(baseline)):
base = mod_signal.interpolate( (baseline[idx-1][0], baseline[idx-1][1]), (baseline[idx][0], baseline[idx][1]), x=signal[imax][0])
h = ((signal[imax][1] - base) * pickingHeight) + base
# get centroid
ileft = imax-1
while (ileft > 0) and (signal[ileft][1] > h):
ileft -= 1
iright = imax
while (iright < len(signal)-1) and (signal[iright][1] > h):
iright += 1
leftMZ = mod_signal.interpolate(signal[ileft], signal[ileft+1], y=h)
rightMZ = mod_signal.interpolate(signal[iright-1], signal[iright], y=h)
# check range
if mz == None and (leftMZ < minX or rightMZ > maxX) and (leftMZ != rightMZ):
return None
# label peak in the newly found selection
if mz != None and leftMZ != rightMZ:
peak = labelpeak(
signal = signal,
minX = leftMZ,
maxX = rightMZ,
pickingHeight = pickingHeight,
baseline = baseline
)
# label current point
else:
peak = labelpoint(
signal = signal,
mz = ((leftMZ + rightMZ)/2.),
baseline = baseline
)
return peak