def build_cwt(timeseries,
max_scale=256,
mother_freq=2,
Norm=True,
fmin=1000.0,
fmax=4096,
maplen=2048):
# Make sure the timeseries peaks in the middle of the map
paddata = np.zeros(maplen)
peakidx = np.argmax(timeseries.data)
paddata[0.5 * maplen - peakidx:0.5 * maplen] = timeseries.data[:peakidx]
paddata[0.5 * maplen:0.5 * maplen + len(timeseries.data) -
peakidx] = timeseries.data[peakidx:]
timeseries = pycbc.types.TimeSeries(paddata, delta_t=timeseries.delta_t)
sample_rate = 1. / timeseries.delta_t
# Range of wavelet scales we're interested in
scales = 1 + np.arange(max_scale)
#scales = np.logspace(0,np.log10(max_scale),max_scale)
# Construct the 'mother wavelet'; we'll begin using a Morlet wavelet
# (sine-Gaussian) but we should/could investigate others
# Could also experiment with values of f0. See cwt.Morlet for info
mother_wavelet = cwt.Morlet(len_signal = \
len(timeseries), scales = scales,
sampf=sample_rate, f0=mother_freq)
# Compute the CWT
wavelet = cwt.cwt(timeseries.data, mother_wavelet)
# Take the absolute values of coefficients
tfmap = np.abs(wavelet.coefs)
# Reduce to useful frequencies
freqs = sample_rate * wavelet.motherwavelet.fc \
/ wavelet.motherwavelet.scales
tfmap[(freqs < fmin) + (freqs > fmax), :] = 0.0
# Normalise
tfmap /= max(map(max, abs(wavelet.coefs)))
# Return a dictionary
timefreq = dict()
timefreq['analysed_data'] = timeseries
timefreq['map'] = tfmap
timefreq['times'] = timeseries.sample_times.data
timefreq['frequencies'] = freqs
timefreq['scales'] = scales
timefreq['mother_wavelet'] = mother_wavelet
timefreq['image_shape'] = np.shape(tfmap)
return timefreq
data = noisysignal.data.data[startidx:startidx+400]
#data = noiselesssignal.data.data[startidx:startidx+512]
data_noisefree = noiselesssignal.data.data[startidx:startidx+400]
motherfreq = 1.5
scalerange = motherfreq/(np.array([10,0.5*sample_rate])*(1.0/sample_rate))
scales = np.arange(scalerange[1],scalerange[0])
mother_wavelet = cwt.Morlet(len_signal = len(data), scales = scales,
sampf=sample_rate, f0=motherfreq)
#mother_wavelet = cwt.SDG(len_signal = len(data), scales = scales, normalize = True,
# fc = 'center')
wavelet = cwt.cwt(data, mother_wavelet)
# --- Plotting
# convert to pseudo frequency
freqs = scale2hertz(mother_wavelet.fc, scales, sample_rate)
peaktime = np.argmax(abs(data_noisefree))/16384.
time = np.arange(0, len(data)/16384.0, 1.0/16384) - peaktime
#collevs=np.linspace(0, max(map(max,abs(wavelet.coefs)**2)), 100)
fpeakmin=2000.0
collevs=np.linspace(0, 1*max(wavelet.get_wps()[freqs>fpeakmin]), 100)
fig, ax_cont = pl.subplots(figsize=(10,8))
import matplotlib.pyplot as plt import cwt fs = 1e3 t = np.linspace(0, 1, fs+1, endpoint=True) x = np.cos(2*np.pi*32*t) * np.logical_and(t >= 0.1, t < 0.3) + np.sin(2*np.pi*64*t) * (t > 0.7) wgnNoise = 0.05 * np.random.standard_normal(t.shape) x += wgnNoise c, f = cwt.cwt(x, 'morl', sampling_frequency=fs) fig, ax = plt.subplots() ax.imshow(np.absolute(wt), aspect='auto')
#
# Construct CWT (continuous wavelet transform) using pyCWT module
#
# Range of wavelet scales we're interested in
scales = 1+np.arange(256)
# Construct the 'mother wavelet'; we'll begin using a Morlet wavelet
# (sine-Gaussian) but we should/could investigate others
# Could also experiment with values of f0. See cwt.Morlet for info
mother_wavelet = cwt.Morlet(len_signal = len(signal), scales = scales,
sampf=sample_rate, f0=1)
# Compute the CWT
wavelet = cwt.cwt(signal.data, mother_wavelet)
# ------------------------
#
# --- Plot The Results ---
#
# ------------------------
# Compute the center frequencies corresponding to each scale
freqs = 0.5 * sample_rate * wavelet.motherwavelet.fc / wavelet.motherwavelet.scales
# Choose the number of colour levels to plo
collevs=np.linspace(0, max(map(max,abs(wavelet.coefs)**2)), 100)
# Open the figure
fig, ax_cont = pl.subplots(figsize=(10,5))
def print_eic_ms(mzML_file):
basename0=os.path.basename(mzML_file)
print(basename0)
ms1_scans=['MS1']
ms2_scans=['MS2']
tree=ET.parse(open(mzML_file,'rb'))
for element in tree.iter(tag='{http://psi.hupo.org/ms/mzml}spectrum'):
if element.findtext(".//*{http://psi.hupo.org/ms/mzml}binary"):
mslevel_elem=element.find("*[@accession='MS:1000511']")
if mslevel_elem is None:
ref_id=element.find("{http://psi.hupo.org/ms/mzml}referenceableParamGroupRef").attrib['ref']
mslevel_elem=(tree.find(".//*{http://psi.hupo.org/ms/mzml}referenceableParamGroup[@id='"+ref_id+"']/*[@accession='MS:1000511']"))
centroid_elem=(tree.find(".//*{http://psi.hupo.org/ms/mzml}referenceableParamGroup[@id='"+ref_id+"']/*[@accession='MS:1000127']"))
else:
centroid_elem=element.find("*[@accession='MS:1000127']")
if centroid_elem is None:
print("error: profile mode!")
sys.exit()
mslevel=mslevel_elem.attrib['value']
if mslevel=='1':
ms1_scans.append(store_scan(element))
elif mslevel=='2':
...
else:
sys.exit()
element.clear()
del element
del tree
print(len(ms1_scans)-1,' MS1 scans')
def mz_slice(ms_scans):
rtdict={rt:n for n,rt in enumerate(sorted({sc.rt for sc in ms_scans[1:]}))}
ofile.write('scan '+ms_scans[0]+'\n')
ofile.write('\t'.join([str(x) for x in rtdict])+'\n')
data_points=[Point(scan.rt,mz,i) for scan in ms_scans[1:] for mz,i in zip(scan.mz,scan.I)]
data_points.sort(key=operator.attrgetter('mz'))
mzlist=array('d',(mz for _,mz,_ in data_points))
for ent in lib_ent:
for ii in [-.03,-.015,0]:
mbd=ent.Mmass+ii
pos0 = bisect_left(mzlist, mbd)
pos1 = bisect_left(mzlist, mbd+.03)
if ent.rt!='NA':
dp_sub=[x for x in data_points[pos0:pos1] if abs(ent.rt-x[0])<RT_shift+30]
else:
dp_sub=data_points[pos0:pos1]
if len(dp_sub)>min_group_size and max(I for _,_,I in dp_sub)>min_highest_I:
eic_dict=dict() # highest intensity in this m/z range
for rt,mz,I in dp_sub:
if rt not in eic_dict or eic_dict[rt][1]<I:
eic_dict[rt]=(mz,I)
if min_group_size<=len({r for r,(_,i) in eic_dict.items() if i>group_I_threshold}):
for rt,mz_i in sorted(eic_dict.items()):
ofile.write('{}\t{}\t{}\n'.format(rt,*mz_i))
ofile.write('-\n')
ofile.write('\n')
with open('eic_'+basename0+'.txt','w') as ofile:
mz_slice(ms1_scans)
cwt.cwt(mzML_file)
motherfreq = 2
scalerange = motherfreq / (np.array([10, 0.5 * sample_rate]) *
(1.0 / sample_rate))
scales = np.arange(scalerange[1], scalerange[0])
mother_wavelet = cwt.Morlet(len_signal=len(data),
scales=scales,
sampf=sample_rate,
f0=motherfreq)
#mother_wavelet = cwt.SDG(len_signal = len(data), scales = scales, normalize = True,
# fc = 'center')
wavelet = cwt.cwt(data, mother_wavelet)
# --- Plotting
# convert to pseudo frequency
freqs = scale2hertz(mother_wavelet.fc, scales, sample_rate)
#collevs=np.linspace(0, max(map(max,abs(wavelet.coefs)**2)), 100)
fpeakmin = 2000.0
collevs = np.linspace(0, 5 * max(wavelet.get_wps()[freqs > fpeakmin]), 100)
fig, ax_cont = pl.subplots(figsize=(10, 5))
c = ax_cont.contourf(time,
freqs,
np.abs(wavelet.coefs)**2,
def _padded_cwt(params, dt, dj, s0, J,mother, padding_len):
padded = np.concatenate([params,params,params])
#padded = np.pad(params, padding_len, mode='edge')
wavelet_matrix, scales, freqs, coi, fft, fftfreqs = cwt.cwt(padded, dt, dj, s0, J,mother)
wavelet_matrix = _unpad(wavelet_matrix, len(params))
return (wavelet_matrix, scales, freqs, coi, fft, fftfreqs)