def dist(spectrum1: sus.MsmsSpectrum, spectrum2: sus.MsmsSpectrum,
method: str = 'dot'):
if method == 'dot':
spectrum1 = copy.copy(spectrum1).scale_intensity(max_intensity=1)
spec1 = pyopenms.MSSpectrum()
spec1.set_peaks([spectrum1.mz, spectrum1.intensity])
spectrum2 = copy.copy(spectrum2).scale_intensity(max_intensity=1)
spec2 = pyopenms.MSSpectrum()
spec2.set_peaks([spectrum2.mz, spectrum2.intensity])
# Third parameter of xCorrelationPrescore is bin size in Da.
xcorr = pyopenms.XQuestScores().xCorrelationPrescore(spec1, spec2,
0.005)
return 1 - xcorr
else:
return 0
def handle_compressed_frame(allmz, allint, allim, mslevel, rtime, center,
width):
mz = np.concatenate(allmz)
intens = np.concatenate(allint)
ims = np.concatenate(allim)
fda = pyopenms.FloatDataArray()
fda.setName("Ion Mobility")
fda.resize(len(mz))
for k, val in enumerate(ims):
fda[k] = val
sframe = pyopenms.MSSpectrum()
sframe.setMSLevel(mslevel)
sframe.setRT(rtime)
sframe.setFloatDataArrays([fda])
p = pyopenms.Precursor()
if mslevel == 2:
p.setMZ(center)
p.setIsolationWindowUpperOffset(width / 2.0)
p.setIsolationWindowLowerOffset(width / 2.0)
sframe.setPrecursors([p])
sframe.set_peaks((mz, intens))
sframe.sortByPosition()
return sframe
def _mergeSpectra(self, tmp):
"""
Perform spectral merging
- we pick one spectrum (the first one) as our reference and add
all other data from the other spectra to it
- we check that merged spectra have equal precursor m/z and same float array
"""
merge_spec = pyopenms.MSSpectrum(tmp[0])
fda = tmp[0].getFloatDataArrays()[0]
fda.clear()
allmz = []
allint = []
for q in tmp:
m, i = q.get_peaks()
allmz.append(m)
allint.append(i)
# Sanity checks, precursors of merged spectra and float arrays need to match
assert q.getPrecursors()[0].getMZ() - merge_spec.getPrecursors()[0].getMZ() < 1e-5
assert len(q.getFloatDataArrays()) == len(merge_spec.getFloatDataArrays())
assert q.getFloatDataArrays()[0].getName() == merge_spec.getFloatDataArrays()[0].getName()
# TODO this is not very efficient, fix in pyOpenMS!
for d in q.getFloatDataArrays()[0]:
fda.push_back(d)
mz = np.concatenate(allmz)
intens = np.concatenate(allint)
# create merge spec
merge_spec.set_peaks( (mz, intens) )
merge_spec.setFloatDataArrays([fda])
merge_spec.sortByPosition()
return merge_spec
def _write_spectra_mzml(filename: str, spectra: Iterable[sus.MsmsSpectrum]) \
-> None:
"""
Write the given spectra to an mzML file.
Parameters
----------
filename : str
The mzML file name where the spectra will be written.
spectra : Iterable[sus.MsmsSpectrum]
The spectra to be written to the mzML file.
"""
experiment = pyopenms.MSExperiment()
for spectrum in tqdm.tqdm(spectra, desc='Spectra written', unit='spectra'):
mzml_spectrum = pyopenms.MSSpectrum()
mzml_spectrum.setMSLevel(2)
mzml_spectrum.setNativeID(spectrum.identifier)
precursor = pyopenms.Precursor()
precursor.setMZ(spectrum.precursor_mz)
precursor.setCharge(spectrum.precursor_charge)
mzml_spectrum.setPrecursors([precursor])
mzml_spectrum.set_peaks([spectrum.mz, spectrum.intensity])
if hasattr(spectrum, 'retention_time'):
mzml_spectrum.setRT(spectrum.retention_time)
if hasattr(spectrum, 'filename'):
mzml_spectrum.setMetaValue('filename',
str.encode(spectrum.filename))
if hasattr(spectrum, 'scan'):
mzml_spectrum.setMetaValue('scan', str.encode(str(spectrum.scan)))
if hasattr(spectrum, 'cluster'):
mzml_spectrum.setMetaValue('cluster',
str.encode(str(spectrum.cluster)))
experiment.addSpectrum(mzml_spectrum)
pyopenms.MzMLFile().store(filename, experiment)
def toMSSpectrum(self):
"""converts to pyopenms.MSSpectrum"""
spec = pyopenms.MSSpectrum()
spec.setRT(self.rt)
spec.setMSLevel(self.msLevel)
ins = spec.getInstrumentSettings()
pol = {
'0': pyopenms.IonSource.Polarity.POLNULL,
'+': pyopenms.IonSource.Polarity.POSITIVE,
'-': pyopenms.IonSource.Polarity.NEGATIVE
}[self.polarity]
ins.setPolarity(pol)
spec.setInstrumentSettings(ins)
oms_pcs = []
for mz, I in self.precursors:
p = pyopenms.Precursor()
p.setMZ(mz)
p.setIntensity(I)
oms_pcs.append(p)
spec.setPrecursors(oms_pcs)
if IS_PYOPENMS_2:
mz = self.peaks[:, 0]
I = self.peaks[:, 1]
spec.set_peaks((mz, I))
else:
spec.set_peaks(self.peaks)
spec.updateRanges()
return spec
def handle_stack(stack, mslevel):
# print "Size", len(stack), mslevel
scan_nr_ = stack[0].split()
assert (len(scan_nr_) == 2)
scan_nr = int(scan_nr_[1])
rt_ = stack[1].split()
assert (len(rt_) == 3)
rt = float(rt_[2]) * 60
# Convert to mz/int pairs
pairs = [
it.split() for it in stack[3:]
if len(it.strip()) > 0 and float(it.split()[1]) > 0.0
]
#mz = [float(it.split()[0]) for it in stack[3:] if len(it.strip()) > 0]
#intensity = [float(it.split()[1]) for it in stack[3:] if len(it.strip()) > 0]
try:
mz = [float(it[0]) for it in pairs]
intensity = [float(it[1]) for it in pairs]
except ValueError:
print("Could not convert", len(stack), "with pairs", len(pairs))
return
assert len(mz) == len(intensity)
#
print("Handle scan at rt", rt)
peaks = np.ndarray(shape=(len(mz), 2), dtype=np.float32)
peaks[:, 0] = mz
peaks[:, 1] = intensity
s = pyopenms.MSSpectrum()
s.set_peaks(peaks)
s.setRT(rt)
s.setMSLevel(1)
outexp.addSpectrum(s)
def testPeakTypeEstimator():
"""
@tests:
PeakTypeEstimator.__init__
PeakTypeEstimator.estimateType
"""
pyopenms.PeakTypeEstimator().estimateType(pyopenms.MSSpectrum())
def fft_pick(spec, fft_low_cutoff, fft_high_cutoff):
# filter spec
filtered_spec = fft_filter_spec(spec, fft_low_cutoff, fft_high_cutoff)
# apply pyopenms peakpicker on filtered spec
picker = pyopenms.PeakPickerHiRes()
newspec_out = pyopenms.MSSpectrum()
picker.pick(filtered_spec, newspec_out)
return newspec_out
def convertToMSSpectrum(input_):
spectrum = pyopenms.MSSpectrum()
for p in input_:
rp = pyopenms.Peak1D()
rp.setMZ(p.getMZ())
rp.setIntensity(p.getIntensity())
spectrum.push_back(rp)
return spectrum
def getSwathExperiment(nr_scans, nr_swathes, precusorsisolation):
exp = pyopenms.MSExperiment()
for spec_cnt in range(1, nr_scans):
ms1_spec = pyopenms.MSSpectrum()
ms1_spec.setMSLevel(1)
ms1_spec.setRT(spec_cnt * 10)
middle_scan = nr_scans / 2
intensity = norm.pdf((spec_cnt - middle_scan) / 4.0)
pk_list = [[500.01, intensity * 3000], [510.05, intensity * 3000]]
peaks = numpy.array(pk_list, dtype=numpy.float32)
ms1_spec.set_peaks(peaks)
exp.addSpectrum(ms1_spec)
# Swath 1: 500.01, 500.15, 500.25
# Swath 2: 501.01, 501.15, 501.25
# Swath 3: 502.01, 502.15, 502.25
# Swath 4: 504.01, 504.15, 504.25
# Swath 5: 505.01, 505.15, 505.25
for i in range(nr_swathes):
middle_scan = nr_scans / 2 + i # shift the middle of the gaussian by one in each scan
intensity = norm.pdf((spec_cnt - middle_scan) / 4.0)
intensity *= 1.2 # 20% higher intensity in each swath
spec = pyopenms.MSSpectrum()
spec.setMSLevel(2)
spec.setRT(spec_cnt * 10 + i + 1)
prec = pyopenms.Precursor()
if precusorsisolation == "OpenSwath":
prec.setIsolationWindowLowerOffset(400 + i * 25)
prec.setIsolationWindowUpperOffset(425 + i * 25)
elif precusorsisolation == "Pwiz":
prec.setIsolationWindowLowerOffset(12.5)
prec.setIsolationWindowUpperOffset(12.5)
elif precusorsisolation == "Missing":
pass
else:
raise Exception(
"precusorsisolation needs to be {Missing,Pwiz,OpenSwath}")
prec.setMZ(400 + i * 25 + 12.5)
spec.setPrecursors([prec])
pk_list = [[500.01 + i, intensity * 3000],
[500.15 + i, intensity * 3000 / 2.0],
[500.25 + i, intensity * 3000 / 3.0]]
peaks = numpy.array(pk_list, dtype=numpy.float32)
spec.set_peaks(peaks)
exp.addSpectrum(spec)
return exp
def testMSSpectrum(self):
spec = pyopenms.MSSpectrum()
p = pyopenms.Peak1D()
p.setMZ(500.0)
p.setIntensity(1e5)
spec.push_back(p)
p_back, = list(spec)
assert isinstance(p_back, pyopenms.Peak1D)
assert p_back.getMZ() == 500.0
assert p_back.getIntensity() == 1e5
def main(options):
# generate fragmentationtype lookup
lookup = {}
methods = pyopenms.ActivationMethod()
for attr in dir(methods):
value = getattr(methods,attr)
if isinstance(value,int):
lookup[value] = attr
print "loading MS Experiment "
exp = pyopenms.MSExperiment()
fh = pyopenms.FileHandler()
fh.loadExperiment(options.infile,exp)
print "checking spectra types:"
fragmentationTypes = {}
for s in exp:
typ = getSpectrumType(s,lookup)
cont = continousSpectrumCheck(s)
fragmentationTypes[typ] = fragmentationTypes.get(typ, [] ) + [cont]
isContinousSpectrum = {}
for typ in fragmentationTypes:
check = percentile75(fragmentationTypes[typ])
isContinousSpectrum[typ] = check
if check == True:
print "\t" + typ + " has continous spectra data"
else:
print "\t" + typ + " has centroided spectra data"
print "picking spectra"
expNew = pyopenms.MSExperiment()
picker = pyopenms.PeakPickerHiRes()
for s in exp:
typ = getSpectrumType(s,lookup)
if isContinousSpectrum[typ] == True:
newSpec = pyopenms.MSSpectrum()
picker.pick(s,newSpec)
expNew.addSpectrum(newSpec)
else:
expNew.addSpectrum(s)
print "saving file to ",options.outfile
mzFile = pyopenms.MzMLFile()
fileoptions = mzFile.getOptions()
fileoptions.setCompression(True)
mzFile.setOptions(fileoptions)
mzFile.store(options.outfile,expNew)
print "finished"
def testMSExperiment():
"""
@tests:
MSExperiment.__init__
MSExperiment.getLoadedFilePath
MSExperiment.getMaxMZ
MSExperiment.getMaxRT
MSExperiment.getMetaValue
MSExperiment.getMinMZ
MSExperiment.getMinRT
MSExperiment.push_back
MSExperiment.setLoadedFilePath
MSExperiment.setMetaValue
MSExperiment.size
MSExperiment.sortSpectra
MSExperiment.updateRanges
MSExperiment.__eq__
MSExperiment.__ge__
MSExperiment.__getitem__
MSExperiment.__gt__
MSExperiment.__iter__
MSExperiment.__le__
MSExperiment.__lt__
MSExperiment.__ne__
MSExperiment.clearMetaInfo
MSExperiment.getKeys
MSExperiment.isMetaEmpty
MSExperiment.metaValueExists
MSExperiment.removeMetaValue
"""
mse = pyopenms.MSExperiment()
_testMetaInfoInterface(mse)
mse.updateRanges()
mse.sortSpectra(True)
assert isinstance(mse.getMaxRT(), float)
assert isinstance(mse.getMinRT(), float)
assert isinstance(mse.getMaxMZ(), float)
assert isinstance(mse.getMinMZ(), float)
assert isinstance(mse.getLoadedFilePath(), str)
mse.setLoadedFilePath("")
assert mse.size() == 0
mse.push_back(pyopenms.MSSpectrum())
assert mse.size() == 1
assert mse[0] is not None
assert isinstance(list(mse), list)
assert mse == mse
assert not mse != mse
def run_list_conversions(self):
pc = pyopenms.Precursor()
allpcs = 500 * [pc]
li = []
for i in range(500):
if (i + 1) % 100 == 0:
show_mem("%4d runs" % i)
spec = pyopenms.MSSpectrum()
spec.setPrecursors(allpcs)
spec.setPrecursors(allpcs)
li.append(spec)
del spec
del li
def set_spec_peaks(self):
data = np.zeros((10000, 2), dtype=np.float32)
li = []
for i in range(1000):
if (i + 1) % 100 == 0:
show_mem("%4d specs processed" % i)
spec = pyopenms.MSSpectrum()
spec.set_peaks((data[:, 0], data[:, 1]))
spec.set_peaks((data[:, 0], data[:, 1]))
spec.set_peaks((data[:, 0], data[:, 1]))
li.append(spec)
for spec in li:
del spec
del data
def fft_filter_spec(spec, fft_low_cutoff, fft_high_cutoff):
# get raw data
peaks = spec.get_peaks()
mz_values, intensities = peaks.T
# fft filter on intensities
fft = scipy.fft(intensities)
fft[:fft_low_cutoff+1] = 0
fft[fft_high_cutoff:] = 0
filtered_intensities = scipy.ifft(fft).real
peaks[:, 1] = filtered_intensities
# build pyopenms.MSSpectrum from filtered raw data
new_spec = pyopenms.MSSpectrum()
new_spec.set_peaks(peaks)
return new_spec
def four_d_spectrum_to_experiment(spec):
"""Function that converts a 4D spectrum object which contains retention
time, ion mobility, mass to charge, and intensity data, into a new
experiment where the ion mobility becomes the retention time of
each of its new spectra and vice versa.
Args:
spec (MSSpectrum): An OpenMS MSSpectrum object.
Returns:
MSExperiment: A new MSExperiment where each MSSpectrum object is a 3D
spectrum from the original where the retention time and ion mobility
has been swapped.
"""
ion_mobility_to_peaks = zip(spec.getFloatDataArrays()[0],
*spec.get_peaks())
new_exp = ms.MSExperiment()
new_mz, new_int = [], []
curr_im = None
for im, mz, intensity in sorted(ion_mobility_to_peaks,
key=lambda x: (x[0], x[1], x[2])):
if im != curr_im:
if curr_im:
new_spec = ms.MSSpectrum()
new_spec.setRT(curr_im)
new_spec.set_peaks((new_mz, new_int))
rt_fda = ms.FloatDataArray()
for i in new_mz:
rt_fda.push_back(spec.getRT())
new_spec.setFloatDataArrays([rt_fda])
new_exp.addSpectrum(new_spec)
new_mz, new_int = [], []
curr_im = im
new_mz.append(mz)
new_int.append(intensity)
return new_exp
def store_frame(frame_id, filename, q, verbose=False):
"""
Store a single frame as an individual mzML file
Note that this is the easiest way to visualize and process the data but
involves a few hacks, namely storing the IM axis as the RT of each
spectrum.
"""
# Get a projected mass spectrum:
q = conn.execute(
"SELECT NumScans FROM Frames WHERE Id={0}".format(frame_id))
num_scans = q.fetchone()[0]
# Get the mapping of the ion mobility axis
scan_number_axis = np.arange(num_scans, dtype=np.float64)
ook0_axis = td.scanNumToOneOverK0(frame_id, scan_number_axis)
# Get a new MSExperiment from OpenMS
e = pyopenms.MSExperiment()
for k, scan in enumerate(td.readScans(frame_id, 0, num_scans)):
index = np.array(scan[0], dtype=np.float64)
mz = td.indexToMz(frame_id, index)
intens = scan[1]
# Store data in OpenMS Spectrum file -> each TOF push is an individual
# spectrum and we set the "RT" to be the ion mobility dimension
s = pyopenms.MSSpectrum()
s.set_peaks((mz, intens))
s.setRT(ook0_axis[k])
e.addSpectrum(s)
if verbose: print "scan ", k
if len(mz) > 0 and verbose:
print " len: ", len(mz), len(intens)
print " at ook0", ook0_axis[k]
print " len: ", mz, intens
for p in s:
print p.getMZ(), p.getIntensity()
# Store file at designated position
pyopenms.MzMLFile().store(filename, e)
def testMSSpectrum(self):
spec = pyopenms.MSSpectrum()
p = pyopenms.Peak1D()
p.setMZ(500.0)
p.setIntensity(1e5)
spec.push_back(p)
p_back, = list(spec)
assert isinstance(p_back, pyopenms.Peak1D)
assert p_back.getMZ() == 500.0
assert p_back.getIntensity() == 1e5
spec.updateRanges()
assert isinstance(spec.getMinMZ(), float)
assert isinstance(spec.getMaxMZ(), float)
assert isinstance(spec.getMinIntensity(), float)
assert isinstance(spec.getMaxIntensity(), float)
assert spec.getMinIntensity() == 1e5
assert spec.getMaxIntensity() == 1e5
target = read_csv_faster(config.target)
source = read_csv_faster(config.source)
super_spectrum_mz = []
for i in target['feature']:
super_spectrum_mz.append(float(i))
mz = []
for i in source['feature']:
mz.append(float(i))
spectrums = []
based_intensity = np.transpose(source['mz_exp'])
for i in range(based_intensity.shape[0]):
peak_intensity = based_intensity[i]
spectrum = pyopenms.MSSpectrum()
spectrum.set_peaks([mz, peak_intensity])
spectrum.sortByPosition()
spectrums.append(spectrum)
super_spectrum_mz = np.array(super_spectrum_mz)
print(super_spectrum_mz.shape)
aligned_intensities = []
aligner = pyopenms.SpectrumAlignment()
target_spectrum = pyopenms.MSSpectrum()
target_spectrum.set_peaks(
[super_spectrum_mz, np.zeros_like(super_spectrum_mz)])
target_spectrum.sortByPosition()
for spectrum in spectrums:
def store_frame(frame_id, td, conn, exp, verbose=False, compressFrame=True):
"""
Store a single frame as an individual mzML file
Note that there are two ways to store the data:
(i) Multiple spectra per frame (for visualization), compressFrame is False. This is the
easiest way to visualize and process the data but involves a few hacks,
namely storing the IM axis as the RT of each spectrum.
(ii) One spectrum per frame, compressFrame is True. This puts all peaks
into a single spectrum (while storing the IM data in an extra array).
This is more efficient for storage and allows analysis that is ignorant
of the IM dimension.
"""
# Get a projected mass spectrum:
q = conn.execute("SELECT NumScans, Time, Polarity, MsMsType FROM Frames WHERE Id={0}".format(frame_id))
tmp = q.fetchone()
num_scans = tmp[0]
time = tmp[1]
pol = tmp[2]
msms = int(tmp[3])
center = -1
width = -1
mslevel = 1
if msms == 2:
q = conn.execute("SELECT TriggerMass, IsolationWidth, PrecursorCharge, CollisionEnergy FROM FrameMsMsInfo WHERE Frame={0}".format(frame_id))
tmp = q.fetchone()
center = float(tmp[0])
width = float(tmp[1])
mslevel = 2
if verbose:
print "mslevel", mslevel, msms
# Get the mapping of the ion mobility axis
scan_number_axis = np.arange(num_scans, dtype=np.float64)
ook0_axis = td.scanNumToOneOverK0(frame_id, scan_number_axis)
allmz = []
allint = []
allim = []
# Traverse in reversed order to get low ion mobilities first
for k, scan in reversed(list(enumerate(td.readScans(frame_id, 0, num_scans)))):
index = np.array(scan[0], dtype=np.float64)
mz = td.indexToMz(frame_id, index)
intens = scan[1]
drift_time = ook0_axis [k]
if compressFrame:
allmz.append(mz)
allint.append(intens)
allim.append([drift_time for k in mz])
continue
# Store data in OpenMS Spectrum file -> each TOF push is an individual
# spectrum and we store the ion mobility in the precursor. The frame
# can be reconstructed by grouping all spectra with the same RT.
s = pyopenms.MSSpectrum()
s.setMSLevel(mslevel)
s.set_peaks( (mz, intens) )
s.setRT(time)
p = pyopenms.Precursor()
p.setDriftTime(drift_time)
if msms == 2:
p.setMZ(center)
p.setIsolationWindowUpperOffset(width / 2.0)
p.setIsolationWindowLowerOffset(width / 2.0)
s.setPrecursors([p])
exp.consumeSpectrum(s)
if compressFrame:
mz = np.concatenate(allmz)
intens = np.concatenate(allint)
ims = np.concatenate(allim)
# print " leeen", len(mz), len(intens)
fda = pyopenms.FloatDataArray()
fda.setName("Ion Mobility")
fda.resize(len(mz))
for k,val in enumerate(ims):
fda[k] = val
sframe = pyopenms.MSSpectrum()
sframe.setMSLevel(mslevel)
sframe.setRT(time)
sframe.setFloatDataArrays([fda])
p = pyopenms.Precursor()
if msms == 2:
p.setMZ(center)
p.setIsolationWindowUpperOffset(width / 2.0)
p.setIsolationWindowLowerOffset(width / 2.0)
sframe.setPrecursors([p])
sframe.set_peaks( (mz, intens) )
sframe.sortByPosition()
exp.consumeSpectrum(sframe)
def bin_spectrum(self, spec: ms.MSSpectrum) -> None:
"""Bins a single spectrum in two passes.
Keyword arguments:
spec: the spectrum to bin
"""
points = util.get_spectrum_points(spec)
points.sort(key=itemgetter(3)) # Ascending IM
temp_bins = [[], [[]]] # 2D spectra
new_bins = [[], [[]]] # Final 1D spectra
for i in range(self.num_bins):
for j in range(2):
temp_bins[j].append([])
new_bins[j].append([])
for i in range(len(points)): # Assign points to bins
bin_idx = int((points[i][3] - self.im_start) / self.bin_size)
if bin_idx >= self.num_bins:
bin_idx = self.num_bins - 1
temp_bins[0][bin_idx].append(points[i])
bin_idx = int((points[i][3] - self.im_offset) / self.bin_size) + 1
if points[i][3] < self.im_offset:
bin_idx = 0
elif bin_idx > self.num_bins:
bin_idx = self.num_bins
temp_bins[1][bin_idx].append(points[i])
for i in range(self.num_bins): # First pass
if len(temp_bins[0][i]) == 0:
continue
temp_bins[0][i].sort(key=itemgetter(1)) # Ascending m/z
mz_start, curr_mz = 0, temp_bins[0][i][0][1]
running_intensity = 0
for j in range(len(temp_bins[0][i])):
if self.within_epsilon(curr_mz, temp_bins[0][i][j][1]):
running_intensity += temp_bins[0][i][j][2]
else: # Reached a new m/z slice
point = list(
temp_bins[0][i][mz_start]) # Prevents aliasing
point[2] = running_intensity
new_bins[0][i].append(point)
mz_start, curr_mz = j, temp_bins[0][i][j][1]
running_intensity = temp_bins[0][i][j][2]
point = list(
temp_bins[0][i][mz_start]) # Take care of the last slice
point[2] = running_intensity
new_bins[0][i].append(point)
transpose = list(zip(*new_bins[0][i]))
new_spec = ms.MSSpectrum() # The final binned spectrum
im_fda = ms.FloatDataArray()
for im in transpose[3]:
im_fda.push_back(im)
new_spec.setRT(spec.getRT())
new_spec.set_peaks((list(transpose[1]), list(transpose[2])))
new_spec.setFloatDataArrays([im_fda])
self.exps[0][i].addSpectrum(new_spec)
for i in range(self.num_bins + 1): # Second pass
if len(temp_bins[1][i]) == 0:
continue
temp_bins[1][i].sort(key=itemgetter(1))
mz_start, curr_mz = 0, temp_bins[1][i][0][1]
running_intensity = 0
for j in range(len(temp_bins[1][i])):
if self.within_epsilon(curr_mz, temp_bins[1][i][j][1]):
running_intensity += temp_bins[1][i][j][2]
else:
point = list(temp_bins[1][i][mz_start])
point[2] = running_intensity
new_bins[1][i].append(point)
mz_start, curr_mz = j, temp_bins[1][i][j][1]
running_intensity = temp_bins[1][i][j][2]
point = list(temp_bins[1][i][mz_start])
point[2] = running_intensity
new_bins[1][i].append(point)
transpose = list(zip(*new_bins[1][i]))
new_spec = ms.MSSpectrum()
im_fda = ms.FloatDataArray()
for im in transpose[3]:
im_fda.push_back(im)
new_spec.setRT(spec.getRT())
new_spec.set_peaks((list(transpose[1]), list(transpose[2])))
new_spec.setFloatDataArrays([im_fda])
self.exps[1][i].addSpectrum(new_spec)
def store_frame(frame_id, td, conn, exp, verbose=False, compressFrame=True):
"""
Store a single frame as an individual mzML file
Note that there are two ways to store the data:
(i) Multiple spectra per frame (for visualization), compressFrame is False. This is the
easiest way to visualize and process the data but involves a few hacks,
namely storing the IM axis as the RT of each spectrum.
(ii) One spectrum per frame, compressFrame is True. This puts all peaks
into a single spectrum (while storing the IM data in an extra array).
This is more efficient for storage and allows analysis that is ignorant
of the IM dimension.
Note that msms = 2 means that we have an MS2 scan whereas msms = 8 stands
for pasef scan.
"""
# Get a projected mass spectrum:
q = conn.execute(
"SELECT NumScans, Time, Polarity, MsMsType FROM Frames WHERE Id={0}".
format(frame_id))
tmp = q.fetchone()
num_scans = tmp[0]
time = tmp[1]
pol = tmp[2]
msms = int(tmp[3])
center = -1
width = -1
next_scan_switch = -1
mslevel = 1
scan_data = []
scan_data_it = 0
# Check whether we have a MS2 or a PASEF scan
if msms == 2:
q = conn.execute(
"SELECT TriggerMass, IsolationWidth, PrecursorCharge, CollisionEnergy FROM FrameMsMsInfo WHERE Frame={0}"
.format(frame_id))
tmp = q.fetchone()
center = float(tmp[0])
width = float(tmp[1])
mslevel = 2
elif msms == 8:
q = conn.execute(
"SELECT IsolationMz, IsolationWidth, ScanNumBegin, ScanNumEnd, CollisionEnergy FROM PasefFrameMsMsInfo WHERE Frame={0} ORDER BY IsolationMz ASC"
.format(frame_id))
scandata = q.fetchall()
tmp = scandata[scan_data_it]
center = float(tmp[0])
width = float(tmp[1])
scan_start = int(tmp[2])
scan_end = int(tmp[3])
next_scan_switch = scan_start
mslevel = 2
if verbose:
print "mslevel", mslevel, msms
# Get the mapping of the ion mobility axis
scan_number_axis = np.arange(num_scans, dtype=np.float64)
ook0_axis = td.scanNumToOneOverK0(frame_id, scan_number_axis)
allmz = []
allint = []
allim = []
# Traverse in reversed order to get low ion mobilities first
for k, scan in reversed(
list(enumerate(td.readScans(frame_id, 0, num_scans)))):
index = np.array(scan[0], dtype=np.float64)
mz = td.indexToMz(frame_id, index)
intens = scan[1]
drift_time = ook0_axis[k]
if compressFrame:
allmz.append(mz)
allint.append(intens)
allim.append([drift_time for dr_time in mz])
# We have multiple MS2 spectra in each frame, we need to separate
# them based on the information from PasefFrameMsMsInfo which
# indicates the switch scan and the isolation parameter for each
# quadrupole isolation.
if next_scan_switch != -1 and next_scan_switch == k:
if verbose:
print "Switch to new scan at", k, "with mapping", scandata
sframe = handle_compressed_frame(allmz, allint, allim, mslevel,
time, center, width)
exp.consumeSpectrum(sframe)
allmz = []
allint = []
allim = []
if k == 0: continue
scan_data_it += 1
tmp = scandata[scan_data_it]
center = float(tmp[0])
width = float(tmp[1])
scan_start = int(tmp[2])
scan_end = int(tmp[3])
next_scan_switch = scan_start
continue
# Store data in OpenMS Spectrum file -> each TOF push is an individual
# spectrum and we store the ion mobility in the precursor. The frame
# can be reconstructed by grouping all spectra with the same RT.
s = pyopenms.MSSpectrum()
s.setMSLevel(mslevel)
s.set_peaks((mz, intens))
s.setRT(time)
p = pyopenms.Precursor()
p.setDriftTime(drift_time)
if mslevel == 2:
p.setMZ(center)
p.setIsolationWindowUpperOffset(width / 2.0)
p.setIsolationWindowLowerOffset(width / 2.0)
s.setPrecursors([p])
exp.consumeSpectrum(s)
if compressFrame and next_scan_switch == -1:
sframe = handle_compressed_frame(allmz, allint, allim, mslevel, time,
center, width)
exp.consumeSpectrum(sframe)
def pick_spectra(self, spec: ms.MSSpectrum, peak_radius: int = 1, window_radius: float = 0.015,
pp_mode: str = 'int', min_int_mult: float = 0.10, strict: bool = True) -> ms.MSSpectrum():
"""Peak picks a single spectrum.
Keyword arguments:
spec: the spectrum to peak pick
peak_radius: the minimum peak radius of a peak set
window_radius: the maximum m/z window radius of a peak set
pp_mode: the mode to use ('ltr' or 'int')
min_int_mult: a multiplier to the maximum peak intensity in a set (for differentiating
between signal and noise)
strict: if False, allow a single increase in intensity in either direction
Returns: the peak picked spectrum.
"""
num_peaks = spec.size()
spec.sortByPosition()
peak_idx = [] # Intensity lookup table
picked = [False] * num_peaks
if pp_mode == 'int':
for i in range(num_peaks):
peak_idx.append([spec[i].getIntensity(), i])
peak_idx = sorted(peak_idx, reverse=True)
picked_spec = ms.MSSpectrum()
picked_spec.setMSLevel(1)
picked_spec.setRT(spec.getRT())
for idx in range(num_peaks): # Begin peak picking
i = idx if pp_mode == 'ltr' else peak_idx[idx][1]
if picked[i]:
continue
init_intensity = spec[i].getIntensity()
total_intensity = spec[i].getIntensity()
init_position = spec[i].getPos()
left_picked, right_picked = 0, 0
low_bound, high_bound = i, i
sFlag = False # Flag for when strict is False
threshold = peak_radius
for j in range(i - 1, -1, -1): # Walk left
if picked[j] or abs(spec[j].getPos() - init_position) > window_radius:
break
if spec[j].getIntensity() > spec[j + 1].getIntensity():
if strict or sFlag or j + 1 == i: # Don't start with an abnormal peak
break
sFlag = True
threshold += 1 # End the peak set with a lower peak ("increase peak_radius")
total_intensity += spec[j].getIntensity()
left_picked += 1
low_bound -= 1
if left_picked >= threshold and spec[j].getIntensity() <= init_intensity * min_int_mult:
break
if left_picked < threshold:
continue
sFlag = False
threshold = peak_radius
for j in range(i + 1, num_peaks): # Walk right
if picked[j] or abs(spec[j].getPos() - init_position) > window_radius:
break
if spec[j].getIntensity() > spec[j - 1].getIntensity():
if strict or sFlag or j - 1 == i:
break
sFlag = True
threshold += 1
total_intensity += spec[j].getIntensity()
right_picked += 1
high_bound += 1
if right_picked >= threshold and spec[j].getIntensity() <= init_intensity * min_int_mult:
break
if right_picked < threshold:
continue
total_position = 0
for j in range(low_bound, high_bound + 1):
picked[j] = True
if total_intensity != 0:
total_position += spec[j].getPos() * (spec[j].getIntensity() / total_intensity)
p = ms.Peak1D()
p.setIntensity(total_intensity)
p.setPos(total_position)
picked_spec.push_back(p)
return picked_spec
def testMSSpectrum():
"""
@tests:
MSSpectrum.clear
MSSpectrum.clearMetaInfo
MSSpectrum.findNearest
MSSpectrum.getAcquisitionInfo
MSSpectrum.getComment
MSSpectrum.getDataProcessing
MSSpectrum.getInstrumentSettings
MSSpectrum.getKeys
MSSpectrum.getMSLevel
MSSpectrum.getMetaValue
MSSpectrum.getName
MSSpectrum.getNativeID
MSSpectrum.getPeptideIdentifications
MSSpectrum.getPrecursors
MSSpectrum.getProducts
MSSpectrum.getRT
MSSpectrum.getSourceFile
MSSpectrum.getType
MSSpectrum.get_peaks
MSSpectrum.intensityInRange
MSSpectrum.isMetaEmpty
MSSpectrum.metaValueExists
MSSpectrum.push_back
MSSpectrum.removeMetaValue
MSSpectrum.setAcquisitionInfo
MSSpectrum.setComment
MSSpectrum.setDataProcessing
MSSpectrum.setInstrumentSettings
MSSpectrum.setMSLevel
MSSpectrum.setMetaValue
MSSpectrum.setName
MSSpectrum.setNativeID
MSSpectrum.setPeptideIdentifications
MSSpectrum.setPrecursors
MSSpectrum.setProducts
MSSpectrum.setRT
MSSpectrum.setSourceFile
MSSpectrum.setType
MSSpectrum.set_peaks
MSSpectrum.size
MSSpectrum.unify
MSSpectrum.updateRanges
MSSpectrum.__eq__
MSSpectrum.__ge__
MSSpectrum.__getitem__
MSSpectrum.__gt__
MSSpectrum.__le__
MSSpectrum.__lt__
MSSpectrum.__ne__
"""
spec = pyopenms.MSSpectrum()
_testMetaInfoInterface(spec)
testSpectrumSetting(spec)
spec.setRT(3.0)
assert spec.getRT() == 3.0
spec.setMSLevel(2)
assert spec.getMSLevel() == 2
spec.setName("spec")
assert spec.getName() == "spec"
p = pyopenms.Peak1D()
p.setMZ(1000.0)
p.setIntensity(200.0)
spec.push_back(p)
assert spec.size() == 1
assert spec[0] == p
spec.updateRanges()
assert isinstance(spec.findNearest(0.0), int)
assert spec == spec
assert not spec != spec
assert spec.get_peaks().shape == (1, 2), spec.get_peaks().shape
def _plot(self,
ax,
color,
top=True,
verificative=True,
show_iden=True,
highlight=True):
target = self if self.is_sorted() else self.sort_by_mz()
if verificative:
temp = target.clip().remove_precursor(copy=False).rank_transform(
copy=False)
mz, intensity = temp.mz, temp.intensity
else:
mz, intensity = target._binning(target.mz, target.intensity)
rects = ax.bar(mz,
intensity,
width=1,
color=color,
linewidth=0,
snap=False)
if show_iden or highlight:
identification = target.get_identification(
) if not self.override_iden else self.override_iden
if identification:
tpp_string = identification.to_tpp_string()
title = identification.to_string(tpp_string)
if self.override_iden: title = 'Overridden: ' + title
if show_iden:
y_loc = 0.9 if top else 0.1
ax.text(0.5,
y_loc,
title,
horizontalalignment='center',
verticalalignment='center',
transform=ax.transAxes,
fontsize=12)
if highlight:
tspec = om.MSSpectrum()
try:
aa_sequence = om.AASequence.fromString(tpp_string)
except Exception:
logging.info(
'String "{}" cannot be understood by pyopenms.AASequence.'
.format(tpp_string))
aa_sequence = None
if aa_sequence:
Spectrum.tsg.getSpectrum(tspec, aa_sequence, 1,
identification.charge)
ions_mz = tspec.get_peaks()[0].astype(np.uint32)
annotations = tspec.getStringDataArrays()[0]
temp_dict = {}
colors = {
'b': 'blue',
'y': 'red',
'a': 'green',
'c': 'teal',
'x': 'purple',
'z': 'orange'
}
for i in range(len(ions_mz)):
temp_dict[ions_mz[i]] = annotations[i]
for i in range(len(mz)):
if mz[i] in temp_dict.keys():
ion_name = temp_dict[mz[i]].decode()
color = colors[ion_name[0]]
rect = rects[i]
height = rect.get_height()
pos_x = rect.get_x() + rect.get_width() / 2.
pos_y = height + 0.01 * (1
if height > 0 else -1)
ax.text(pos_x,
pos_y,
ion_name,
fontsize=6,
ha='center',
va='center',
color=color)
rect.set_color(color)
return
def load(self, ifname: str, peakMap: pyopenms.MSExperiment):
inF = open(ifname, 'r')
lines = inF.read().splitlines()
curLine = 0
nLines = len(lines)
#generate spectrum list
spectraList = list()
while curLine < nLines:
if lines[curLine] == 'BEGIN IONS':
spectrum = pyopenms.MSSpectrum()
spectrum.setMSLevel(2)
precursor = pyopenms.Precursor()
curLine += 1
while curLine < nLines:
if lines[curLine][0].isalpha():
match = re.search('^([A-Z]+)=(.+)$', lines[curLine])
if match.group(1) == 'TITLE':
titleData = match.group(2).split(',')
for s in titleData:
if re.search('^scan[_=]', s):
match = re.search('^scan[_=]([0-9]+)', s)
assert (len(match.groups()) == 1)
spectrum.setNativeID('scan={}'.format(
match.group(1)))
elif match.group(1) == 'PEPMASS':
preMZ = [
float(x) for x in match.group(2).split(' ')
]
assert (len(preMZ) <= 2)
precursor.setMZ(preMZ[0])
if len(preMZ) > 1:
precursor.setIntensity(preMZ[1])
elif match.group(1) == 'CHARGE':
match = re.search('^([0-9])[+-]{0,1}$',
match.group(2))
assert (len(match.groups()) == 1)
precursor.setCharge(int(match.group(1)))
elif match.group(1) == 'RTINSECONDS':
spectrum.setRT(float(match.group(2)))
elif lines[curLine][0].isnumeric():
while curLine < nLines and lines[curLine] != 'END IONS':
ion = [float(x) for x in lines[curLine].split(' ')]
assert (len(ion) == 2)
ion_temp = pyopenms.Peak1D()
ion_temp.setMZ(ion[0])
ion_temp.setIntensity(ion[1])
spectrum.push_back(ion_temp)
curLine += 1
break
curLine += 1
spectrum.setPrecursors([precursor])
spectraList.append(spectrum)
curLine += 1
peakMap.setSpectra(spectraList)
def store_frame(frame_id,
td,
conn,
exp,
verbose=False,
compressFrame=True,
keep_frames=False):
"""
Store a single frame as an individual mzML file
Note that there are two ways to store the data:
(i) Multiple spectra per frame (for visualization), compressFrame is False. This is the
easiest way to visualize and process the data but involves a few hacks,
namely storing the IM axis as the RT of each spectrum.
(ii) One spectrum per frame, compressFrame is True. This puts all peaks
into a single spectrum (while storing the IM data in an extra array).
This is more efficient for storage and allows analysis that is ignorant
of the IM dimension.
Note that msms = 2 means that we have an MS2 scan whereas msms = 8 stands
for pasef scan. (New tdf 5.1 has msms = 9 for pasef scan)
"""
# Get a projected mass spectrum:
q = conn.execute(
"SELECT NumScans, Time, Polarity, MsMsType FROM Frames WHERE Id={0}".
format(frame_id))
tmp = q.fetchone()
num_scans = tmp[0]
time = tmp[1]
pol = tmp[2]
msms = int(tmp[3])
center = -1
width = -1
next_scan_switch = -1
mslevel = 1
scan_data = []
scan_data_it = 0
in_scan = False
scandata = None
# Check whether we have a MS2 or a PASEF scan
if msms == 2:
q = conn.execute(
"SELECT TriggerMass, IsolationWidth, PrecursorCharge, CollisionEnergy FROM FrameMsMsInfo WHERE Frame={0}"
.format(frame_id))
tmp = q.fetchone()
center = float(tmp[0])
width = float(tmp[1])
mslevel = 2
# new tdf 5.1 has pasef scan msms = 9
elif msms == 9:
q = conn.execute(
"SELECT IsolationMz, IsolationWidth, ScanNumBegin, ScanNumEnd, CollisionEnergy, Frame FROM DiaFrameMsMsWindows INNER JOIN DiaFrameMsMsInfo ON DiaFrameMsMsWindows.WindowGroup = DiaFrameMsMsInfo.WindowGroup WHERE Frame={0} ORDER BY ScanNumBegin DESC"
.format(frame_id))
scandata = q.fetchall()
tmp = scandata[scan_data_it]
center = float(tmp[0])
width = float(tmp[1])
scan_start = int(tmp[2])
scan_end = int(tmp[3])
next_scan_switch = scan_end
# Check if we already are in the new scan (if there is no
# gap between scans, happens for diaPASEF):
if next_scan_switch == num_scans:
next_scan_switch = scan_start
in_scan = True
mslevel = 2
elif msms == 8:
q = conn.execute(
"SELECT IsolationMz, IsolationWidth, ScanNumBegin, ScanNumEnd, CollisionEnergy FROM PasefFrameMsMsInfo WHERE Frame={0} ORDER BY ScanNumBegin DESC"
.format(frame_id))
scandata = q.fetchall()
tmp = scandata[scan_data_it]
center = float(tmp[0])
width = float(tmp[1])
scan_start = int(tmp[2])
scan_end = int(tmp[3])
next_scan_switch = scan_end
# Check if we already are in the new scan (if there is no
# gap between scans, happens for diaPASEF):
if next_scan_switch == num_scans:
next_scan_switch = scan_start
in_scan = True
mslevel = 2
else:
# MS1
pass
if verbose:
print("Frame", frame_id, "mslevel", mslevel, msms,
"contains nr scans:", num_scans, "and nr pasef scans",
len(scandata) if scandata else -1)
print("Scandata for PASEF:", scandata)
if keep_frames:
next_scan_switch = -1
# Get the mapping of the ion mobility axis
scan_number_axis = np.arange(num_scans, dtype=np.float64)
ook0_axis = td.scanNumToOneOverK0(frame_id, scan_number_axis)
nr_scans_created = 0
allmz = []
allint = []
allim = []
# Traverse in reversed order to get low ion mobilities first (and high scan times first)
for k, scan in reversed(
list(enumerate(td.readScans(frame_id, 0, num_scans)))):
index = np.array(scan[0], dtype=np.float64)
mz = td.indexToMz(frame_id, index)
intens = scan[1]
drift_time = ook0_axis[k]
if compressFrame:
allmz.append(mz)
allint.append(intens)
allim.append([drift_time for dr_time in mz])
# We have multiple MS2 spectra in each frame, we need to separate
# them based on the information from PasefFrameMsMsInfo which
# indicates the switch scan and the isolation parameter for each
# quadrupole isolation.
if next_scan_switch >= 0 and next_scan_switch >= k:
if verbose:
print("Switch to new scan at", k, "/", next_scan_switch,
"store scan of size", len(allmz))
if in_scan:
# Only store spectrum when actually inside a scan, skip the "between scan" pushes
sframe = handle_compressed_frame(allmz, allint, allim,
mslevel, time, center,
width)
sframe.setNativeID("frame=%s_scan=%s" %
(frame_id, next_scan_switch))
exp.consumeSpectrum(sframe)
nr_scans_created += 1
allmz = []
allint = []
allim = []
if k == 0: continue
if in_scan:
scan_data_it += 1
if scan_data_it >= len(scandata):
if verbose:
print(
"LEFT the last scan, nothing else to do here")
next_scan_switch = -2
continue
# Already prepare for next scan
tmp = scandata[scan_data_it]
center = float(tmp[0])
width = float(tmp[1])
scan_start = int(tmp[2])
scan_end = int(tmp[3])
in_scan = False
next_scan_switch = scan_end
if verbose:
print("LEAVING scan now, next scan starts at:",
next_scan_switch)
# Check if we already are in the new scan (if there is no
# gap between scans, happens for diaPASEF):
if k == next_scan_switch:
if verbose:
print("STARTING new scan immediately at", k, ":",
center - width / 2.0, center + width / 2.0,
"scan will end at:", next_scan_switch)
next_scan_switch = scan_start
in_scan = True
else:
in_scan = True
next_scan_switch = scan_start
if verbose:
print("STARTING new scan at", k, ":",
center - width / 2.0, center + width / 2.0,
"scan will end at:", next_scan_switch)
continue
# Store data in OpenMS Spectrum file -> each TOF push is an individual
# spectrum and we store the ion mobility in the precursor. The frame
# can be reconstructed by grouping all spectra with the same RT.
s = pyopenms.MSSpectrum()
s.setMSLevel(mslevel)
s.set_peaks((mz, intens))
s.setRT(time)
s.setNativeID("frame=%s spec %s" % (frame_id, k))
p = pyopenms.Precursor()
p.setDriftTime(drift_time)
if mslevel == 2:
p.setMZ(center)
p.setIsolationWindowUpperOffset(width / 2.0)
p.setIsolationWindowLowerOffset(width / 2.0)
s.setPrecursors([p])
exp.consumeSpectrum(s)
# Store data compressed for cases where the whole frame represents a single spectrum (e.g. MS1)
if compressFrame and next_scan_switch == -1:
sframe = handle_compressed_frame(allmz, allint, allim, mslevel, time,
center, width)
sframe.setNativeID("frame=%s" % frame_id)
exp.consumeSpectrum(sframe)
nr_scans_created += 1
if scandata is not None and (nr_scans_created != len(scandata)):
raise Exception("Something went quite wrong here, we expected",
len(scandata), "scans, but only created",
nr_scans_created)
