raise NPKError("Axis %d should be complex" % axis, data=data)
#---------------------------------------------------------------------------
def ft_sim(data):
"""performs the fourier transform of a data-set acquired on a Bruker in
simultaneous mode
Processing is performed only along the F2 (F3) axis if in 2D (3D)
(Bruker QSIM mode)"""
todo = data.dim
data.revf().fft(axis=todo)
return data
NPKData_plugin("ft_sim", ft_sim)
#---------------------------------------------------------------------------
def ft_seq(data):
"""performs the fourier transform of a data-set acquired on a Bruker in simultaneous mode
Processing is performed only along the F2 (F3) axis if in 2D (3D)
(Bruker QSIM mode)"""
todo = data.dim
data.revf().rfft(axis=todo)
return data
NPKData_plugin("ft_seq", ft_seq)
Binv,
y,
eta,
nbiter=nbiter,
lamda=lamda,
prec=precision,
full_output=full_output)
Ok = not np.isnan(
x.sum()) # the current algo sometimes produces NaN values
if not Ok:
NaN_found += 1
uncertainty *= 1.4
npkd.set_buffer(x[:, 0])
npkd.noise = eta
if full_output:
npkd.full_output = c
if NaN_found > 0:
print("%d NaN conditions encountered during PALMA processing" %
NaN_found)
return npkd
def test(npkd):
print('Not implemented')
NPKData_plugin("palma", palma)
NPKData_plugin("do_palma", do_palma)
NPKData_plugin("prepare_palma", prepare_palma)
NPKData_plugin("calibdosy", dcalibdosy)
if npkd.dim == 1:
z = denoise1D(npkd.get_buffer(),
nsigma * npkd.get_buffer().std(),
wavelet=wavelet)
elif npkd.dim == 2:
z = denoise2D(npkd.get_buffer(),
nsigma * npkd.get_buffer().std(),
wavelet=wavelet)
else:
raise NPKError("not implemented")
npkd.set_buffer(z)
return npkd
if ok:
NPKData_plugin("wavelet", wavelet)
class WaveLetTest(unittest.TestCase):
""" - Testing Wavelet plugin- """
def setUp(self):
self.verbose = 1 # verbose > 0 switches messages on
def announce(self):
if self.verbose > 0:
print("\n========", self.shortDescription(), '===============')
def test_wave(self):
""" - testing wavelet - """
from spike.util.signal_tools import findnoiselevel
from spike.NPKData import NPKData
if xlabel == "_def_":
xlabel = npkd.axis2.currentunit
if ylabel == "_def_":
ylabel = npkd.axis1.currentunit
if xlabel is not None:
dbk['x_axis_label'] = xlabel
if ylabel is not None:
dbk['y_axis_label'] = ylabel
dbk.update(dbkdic)
xs, ys, col = get_contour_data(ax)
if redraw:
npkd.bokeh_fig['xs'] = xs
npkd.bokeh_fig['ys'] = ys
else:
p = bk.figure(**dbk)
xs, ys, col = get_contour_data(ax)
dfig = {}
dfig['xs'] = xs
dfig['ys'] = ys
dfig['color'] = col
dfig.update(dfigdic)
p.multi_line(**dfig)
npkd.bokeh_fig = dfig
npkd.bokeh_plot = p
del fig, ax
if show:
bk.show(npkd.bokeh_plot)
return npkd
NPKData_plugin("bokeh", bokeh_display)
d.peaks.report(NbMaxPeaks=10)
def test_center2d(self):
M = np.zeros((20, 20))
# add one peak at (F1,F2) 5.3, 7.9 with widthes (5.0,1.3)
for y in range(1, 10):
for x in range(6, 11):
# yo, x0 intens, widthy, widthx
M[y, x] = center2d(np.array([y, x]), 5.3, 7.9, 20.0, 5.0, 1.3)
#print (M[1:10,6:11])
self.assertAlmostEqual(M[2, 7], 5.87777515)
d = _NPKData(buffer=np.maximum(M, 0.0))
d.peaks = Peak2DList(source=d)
# self, Id, label, intens, posF1, posF2
d.peaks.append(Peak2D(0, "0", 18.0, 5, 8))
d.centroid(npoints_F1=5)
self.assertAlmostEqual(d.peaks[0].posF1, 5.3)
self.assertAlmostEqual(d.peaks[0].posF2, 7.9)
self.assertAlmostEqual(d.peaks[0].intens, 20.0)
self.assertAlmostEqual(d.peaks[0].widthF1, 5.0 * np.sqrt(2))
self.assertAlmostEqual(d.peaks[0].widthF2, 1.3 * np.sqrt(2))
NPKData_plugin("pp", peakpick)
NPKData_plugin("peakpick", peakpick)
NPKData_plugin("centroid", centroid)
NPKData_plugin("report_peaks", report_peaks)
NPKData_plugin("display_peaks", display_peaks)
NPKData_plugin("peaks2d", peaks2d)
NPKData_plugin("pk2pandas", pk2pandas)
window_size)
else:
raise NPKError("a faire")
return npkd
########################################################################
def sg2D(npkd, window_size, order, deriv=None):
"""applies a 2D Savitzky-Golay of order filter to data
window_size : int
the length of the square window. Must be an odd integer number.
order : int
the order of the polynomial used in the filtering.
Must be less than `window_size` - 1.
deriv: None, 'col', or 'row'. 'both' mode does not work.
the direction of the derivative to compute (default = None means only smoothing)
can be applied to a 2D only.
"""
import spike.Algo.savitzky_golay as sgm
npkd.check2D()
npkd.set_buffer(
sgm.savitzky_golay2D(npkd.get_buffer(),
window_size,
order,
derivative=deriv))
return npkd
NPKData_plugin("sg", sg)
NPKData_plugin("sg2D", sg2D)
for the second pass a lower rank can be used.
"""
if npkd.dim == 1:
if npkd.axis1.itype == 0: # real
buff = as_cpx(_base_ifft(_base_rfft(npkd.buffer))) # real case, go to analytical signal
else: #complex
buff = npkd.get_buffer() # complex case, makes complex
sane_result = sane(buff, rank, orda=orda, trick=trick, iterations=iterations) # performs denoising
if npkd.axis1.itype == 0: # real
buff = _base_irfft(_base_fft(as_float(sane_result))) # real case, comes back to real
npkd.set_buffer(buff)
else:
npkd.buffer = as_float(sane_result) # complex case, makes real
elif npkd.dim == 2:
todo = npkd.test_axis(axis)
if todo == 2:
for i in xrange(npkd.size1):
r = npkd.row(i).sane(rank=rank, orda=orda, iterations=iterations)
npkd.set_row(i,r)
elif todo == 1:
for i in xrange(npkd.size2):
r = npkd.col(i).sane(rank=rank, orda=orda, iterations=iterations)
npkd.set_col(i,r)
elif npkd.dim == 3:
raise Exception("not implemented yet")
return npkd
NPKData_plugin("sane", sane_plugin)
# here, noise level is half of the largest amplitude.
np.random.seed(12345)
# noisy data
nfid = fid0 + NOISE * np.random.randn(
len(fid0)) + 1j * NOISE * np.random.randn(len(fid0))
# generate sampling
RATIO = 1. / 8
sampling = gene_sampling(RATIO)
# prepare
f = NPKData(buffer=nfid[sampling])
f.axis1.sampling = sampling
# do it
t0 = time.time()
g = f.copy().pg_sane(iterations=20, rank=15)
elaps = time.time() - t0
SNR = -20 * np.log10(
np.linalg.norm(g.get_buffer() - fid0) / np.linalg.norm(fid0))
print(
"test_NUS_sampling2: elaps %.2f sec SNR: %.1f dB should be larger than 30dB"
% (elaps, SNR))
self.assertTrue(SNR > 30.0)
ax1 = plt.subplot(211)
f.copy().apod_sin().zf(2).fft().display(
title='spectrum original data with sampling noise', figure=ax1)
ax2 = plt.subplot(212)
g.copy().apod_sin().zf(2).fft().display(
title='spectrum after pg_sane cleaning', figure=ax2)
NPKData_plugin("pg_sane", pg_sane)
"""
set to zeros all points below nsigma times the noise level
This allows the corresponding data-set, once stored to file, to be considerably more compressive.
nsigma: float
the ratio used, typically 1.0 to 3.0 (higher compression)
nbseg: int
the number of segments used for noise evaluation, see util.signal_tools.findnoiselevel
axis: int
the axis on which the noise is evaluated, default is fastest varying dimension
"""
todo = npkd.test_axis(axis)
if npkd.dim == 1:
noise = findnoiselevel(npkd.get_buffer(), nbseg=nbseg)
npkd.zeroing(nsigma * noise)
elif npkd.dim == 2:
if todo == 2:
for i in xrange(npkd.size1):
npkd.set_row(i,
npkd.row(i).fastclean(nsigma=nsigma, nbseg=nbseg))
elif todo == 1:
for i in xrange(npkd.size2):
npkd.set_col(i,
npkd.col(i).fastclean(nsigma=nsigma, nbseg=nbseg))
else:
raise NPKError("a faire")
return npkd
NPKData_plugin("fastclean", fastclean)
def lpext(npkd, final_size, lprank=10, algotype="burg"):
"""
extends a 1D FID or 2D FID in F1 up to final_size, using lprank coefficients, and algotype mode
"""
if npkd.dim == 1:
return lpext1d(npkd, final_size, lprank=10, algotype="burg")
elif npkd.dim == 2:
return lpext2d(npkd, final_size, lprank=10, algotype="burg")
else:
raise Exception("Not implemented yet")
class LinpredicTests(unittest.TestCase):
def setUp(self):
self.verbose = 1 # verbose >0 switches on messages
def announce(self):
if self.verbose > 0:
print(self.shortDescription())
def test_log(self):
"""testing log"""
import math
self.announce()
x = 0.0
y = math.log(1.0)
self.assertAlmostEqual(x, y)
NPKData_plugin("lpext", lpext)
"""computes a new set of parameters whil moving pivot from pvbef to pvaft"""
# p (pbef) goes from 0 to 1 so, for a buffer of size, pivot position is p*size
# in position x, phase correction is prop to (x/size - p)
# ph = P0 + (x/size - p) P1 + (x/size - p)² P2
# ph = P0 + x/size P1 - p P1 + (x/size)² P2 + p² P2 - 2 x p/size P2
# ph = P0 - p P1 + p² P2 - 2 x p/size P2 + x/size P1 + (x/size)² P2
# and with a new pivot paft:
# ph = P0' - paft P1' + paft² P2' - 2 x paft/size P2' + x/size P1' + (x/size)² P2'
# P0' - paft P1' + paft² P2' + (x/size)(- 2 paft P2' + P1') + (x/size)² P2'
# for the phase to be equal with paft we need that cst, x and x² terms to be equal, which means
# so
# => P2' = P2
# - 2 paft P2' + P1' = - 2 p P2 + P1
# => P1' = P1 - 2 p P2 + 2 paft P2
# = P1 + 2(paft-p)P2
# P0' - paft P1' + paft² P2' = P0 - p P1 + p² P2
# => P0' = P0 - p P1 + p² P2 + paft P1' - paft² P2'
# = P0 - p P1 + (p²-paft²) P2 + paft P1'
p2p = p2
p1p = p1 + 2*(pvaft-pvbef)*p2
p0p = 360*(p0/360 - pvbef*p1 + (pvbef**2 - pvaft**2)*p2 + pvaft*p1p)
# then bring ph0 in [-180, 180]
while p0p > 180:
p0p -= 360
while p0p < -180:
p0p += 360
return p0p, p1p, p2p, pvaft
# NPKData_plugin("phase", phase) # supercede the regular phase correction code - not compatible with Test suite !
NPKData_plugin("phaseMS", phase)
#!/usr/bin/env python
# encoding: utf-8
"""Test procedure for plugins
"""
from __future__ import print_function
from spike.NPKData import NPKData_plugin
def fake(dd, title):
"fake method"
dd.test_title = title
return dd
NPKData_plugin("fake", fake)
pass
else:
xrange = range
def rem_ridge(data):
"""
This function removes a F1 ridge by evaluating a mean avalue over the last 10% data of each column of a 2D
"""
data.check2D()
deb = int(0.9 * data.size1) # debut et fin de l'évaluation
fin = data.size1
r = data.row(deb)
for i in xrange(deb + 1, fin): # je calcule la moyenne
r.add(data.row(i))
r.mult(-1.0 / (fin - deb))
for i in xrange(data.size1):
data.set_row(i, data.row(i).add(r))
return data # et garde la syntaxe standard NPKData
NPKData_plugin("rem_ridge", rem_ridge)
"""
rem_ridge() injection
now on (in this running version)
data.rem_ridge()
will realize a baseline ridge correction
"""
cc = npkd.col(i) # going column wise is probably faster ...
d5[i - z2lo, :] = cc[
z1lo:z1up +
1] # taking a slice out of a npkdata returns a np.array
zmax = np.amax(d5)
zmin = np.amin(d5) # 0 - some data-set are negative
xmin = zoom[2]
xmax = zoom[3]
ymin = zoom[0]
ymax = zoom[1]
mlab.figure(bgcolor=(1., 1., 1.), fgcolor=(0., 0., 0.))
mlab.surf(d5,
extent=[0, 1000, 0, 1000, 0, 1000],
warp_scale='auto',
colormap=colormap)
ax = mlab.axes(x_axis_visibility=showaxes,
y_axis_visibility=showaxes,
z_axis_visibility=showaxes,
xlabel="F2 " + npkd.axis2.currentunit,
ylabel="F1 " + npkd.axis1.currentunit,
zlabel='Intensity',
ranges=[xmin, xmax, ymin, ymax, zmin, zmax],
nb_labels=5)
ax.label_text_property.font_family = font
ax.title_text_property.font_family = font
ax.axes.font_factor = fontsize
if ok:
NPKData_plugin("zoomwindow", zoom3D)
P1step = -P1step
break
P0min = P0minnext
P1min = P1minnext
if debug:
dd = d.copy()
P0, P1 = phase_pivot(dd, P0min, P1min, pivot)
print("*** P0 P1 :", P0, P1)
color_sequence = [
'#1f77b4', '#aec7e8', '#ff7f0e', '#ffbb78', '#2ca02c',
'#98df8a', '#d62728', '#ff9896', '#9467bd', '#c5b0d5',
'#8c564b', '#c49c94', '#e377c2', '#f7b6d2', '#7f7f7f',
'#c7c7c7', '#bcbd22', '#dbdb8d', '#17becf', '#9edae5'
]
dd.display(new_fig=False,
label="%.0f %.0f" % (P0, P1),
color=color_sequence[neval % len(color_sequence)])
P0step = P0step / 2.0
P1step = P1step / 2.0
if P0step < 5.0:
bcorr = baselinecorr # bcorr is expensive, so we make it only at the end if needed
if debug: print('bcorr = True')
(P0, P1) = phase_pivot(d, P0min, P1min, pivot)
if debug:
print("**FINAL** %.2f %.2f in %d evaluations" % (P0, P1, neval))
d.axis1.P0 = P0
d.axis1.P1 = P1
return d
NPKData_plugin("apmin", apmin)
for mol, values in calib.items():
mass, charge = values
peptide = False
try:
nmass = iso.parse_peptide(mol).monoisotop(charge)
peptide = True
except:
pass
if not peptide:
try:
nmass = iso.parse_formula(mol).monoisotop(charge)
peptide = False
except:
print('could not understand %s' % mol)
continue
if nmass < 10.0:
continue
if peptide and charge > 0: # then add H+
nmass += iso.parse_formula('H').monoisotop()
diff.append(1E6 * abs(mass - nmass) / mass)
calib[mol] = (nmass, charge)
print('recalibrate mean: %.3f ppm max: %.3f ppm' %
(sum(diff) / len(diff), max(diff)))
return calib
# and plug the whole stuf into NPKData
NPKData_plugin("set_calib", set_calib)
NPKData_plugin("calib", calib)
NPKData_plugin("display_calib", display_calib)
simple 1D correction
spline:
a cubic spline correction
both linear and spline use an additional list of pivot points 'xpoints' used to calculate the baseline
if xpoints absent, pivots are estimated automaticaly
if xpoints is integer, it determines the number of computed pivots (defaut is 8 if xpoints is None)
if xpoints is a list of integers, there will used as pivots
if nsmooth >0, buffer is smoothed by moving average over 2*nsmooth+1 positions around pivots.
if dataset is complex, the xpoints are computed on the modulus spectrum, unless modulus is False
default is spline with automatic detection of 8 baseline points
"""
if method == 'auto':
return bcorr_auto(npkd)
else:
if xpoints is None or isinstance(xpoints, int):
xpoints = autopoints(npkd, xpoints, modulus=modulus)
if method == 'linear':
return linear_interpolate(npkd, xpoints, nsmooth=nsmooth)
elif method == 'spline':
return spline_interpolate(npkd, xpoints, nsmooth=nsmooth)
else:
raise Exception("Wrong method in bcorr plugin")
NPKData_plugin("bcorr_lin", linear_interpolate)
NPKData_plugin("bcorr_spline", spline_interpolate)
NPKData_plugin("bcorr_auto", bcorr_auto)
NPKData_plugin("bcorr", bcorr)
#!/usr/bin/env python
# encoding: utf-8
"""computes diagonal of 2D-MS spectra
Created by DELSUC Marc-André on 2020-12-10.
"""
import numpy as np
from spike.NPKData import NPKData_plugin
def diagonal(self):
"""allows to extract the diagonal of a 2D FTMS spectrum"""
self.check2D()
ddiag = self.row(0) # container
diag = np.zeros(self.size2) # data
high = int(self.axis2.mztoi(self.axis1.lowmass)) # borders in index
low = int(self.axis2.mztoi(self.axis1.highmass))
i = np.arange(low, high) # convert indices to mz
z = self.axis2.itomz(i)
iz = np.int_(np.round(self.axis1.mztoi(z)))
jz = np.int_(np.round(self.axis2.mztoi(z)))
diag[jz] = self[iz, jz] # and copy
ddiag.set_buffer(diag)
return ddiag
NPKData_plugin("diagonal", diagonal)
integoff=0.3,
integscale=0.5,
color='red',
label=False,
labelxposition=1,
labelyposition=None,
regions=False,
zoom=None,
figure=None,
curvedict=None,
labeldict=None):
npkd.integrals.display(integoff=integoff,
integscale=integscale,
color=color,
label=label,
labelxposition=labelxposition,
labelyposition=labelyposition,
regions=regions,
zoom=zoom,
figure=figure,
curvedict=curvedict,
labeldict=labeldict)
return npkd
display.__doc__ = Integrals.display.__doc__
NPKData_plugin("integrate", integrate)
NPKData_plugin("integral_calibrate", calibrate)
NPKData_plugin("display_integral", display)
(ih1 - inext1), (ih2 - inext2)),
file=file)
# print(here1, here2, here1_2, here2_2, inext1, ih1, inext2, ih2, file=F)
here2_2 = next2
here2 = (here2 + bsize2)
here1_2 = next1
here1 = (here1 + bsize1)
return data
class BucketingTests(unittest.TestCase):
def setUp(self):
self.verbose = 1 # verbose >0 switches on messages
def announce(self):
if self.verbose > 0:
print(self.shortDescription())
def _test_log(self):
"""testing log"""
import math
self.announce()
x = 0.0
y = math.log(1.0)
self.assertAlmostEqual(x, y)
NPKData_plugin("bucket1d", bucket1d)
NPKData_plugin("bucket2d", bucket2d)
# create 1D spectrum
t = np.linspace(0,10,1000)
y = np.zeros_like(t)
A = (100,100,100)
W = (100, 110, 115)
TAU = (0.3, 1, 3)
for a,w,tau in zip(A,W, TAU):
y += a*np.cos(w*t)*np.exp(-t*tau)
Y = np.fft.rfft(y).real
Y -= Y[0]
# load and peak pick
d=spike.NPKData._NPKData(buffer=Y)
d.pp(threshold=1000)
# check
self.assertEqual(list(d.peaks.pos) , [159.0, 175.0, 183.0])
d.fit()
if scipy.__version__ > '0.17.0':
# first fit is not full because of constraints on widthes (third peak)
self.assertAlmostEqual(d.peaks.chi2, 121.72613405, places=2)
d.fit()
self.assertAlmostEqual(d.peaks.chi2, 15.0445981291, places=2) # second is complete
# other possibility is centroid
d.pp(threshold=1000)
d.centroid()
d.fit(zoom=(140,200))
self.assertAlmostEqual(d.peaks.chi2, 12.4304236435, places=1) # lower because of zoom.
self.assertAlmostEqual( sum(list(d.peaks.pos)), 517.74817237246634, places=2)
NPKData_plugin("simulate", simulate)
NPKData_plugin("fit", fit)
NPKData_plugin("display_fit", display_fit)
Created by DELSUC Marc-André on February 2019
Copyright (c) 2019 IGBMC. All rights reserved.
"""
import numpy as np
from spike.NPKData import as_float, NPKData_plugin
#-------------------------------------------------------------------------------
def gaussenh(npkd, width, enhancement=2.0, axis=0):
"""
apply an gaussian enhancement, width is in Hz
enhancement is the strength of the effect
multiplies by gauss(width) * exp(-enhancement*width)
"""
todo = npkd.test_axis(axis)
it = npkd.axes(todo).itype
sw = npkd.axes(todo).specwidth
size = npkd.axes(todo).size
if it == 1: # means complex
size = size // 2
baseax = width * np.arange(size) / sw
e = np.exp(enhancement * baseax)
e *= np.exp(-(baseax)**2)
if it == 1:
e = as_float((1 + 1.0j) * e)
# check NPKData.py to see how apodisations are handled.
return npkd.apod_apply(axis, e)
NPKData_plugin("gaussenh", gaussenh)
