def confs_from_formula(formula,
threshold=0.001,
total_prob=None,
charge=1,
adduct=None):
"""Simulate and return spectrum peaks for given formula.
Parameters as in __init__ method. `formula` must be a nonempty string.
"""
parsed = re.findall('([A-Z][a-z]*)([0-9]*)', formula)
formula = Counter()
for e, n in parsed:
n = int(n) if n else 1
formula[e] += n
if adduct:
formula[adduct] += charge
assert all(v >= 0 for v in formula.values())
formula = ''.join(x + str(formula[x]) for x in formula if formula[x])
if total_prob is not None:
isospec = IsoSpecPy.IsoTotalProb(formula=formula,
prob_to_cover=total_prob,
get_minimal_pset=True,
get_confs=False)
else:
isospec = IsoSpecPy.IsoThreshold(formula=formula,
threshold=threshold,
absolute=False,
get_confs=False)
confs = [(x[0] / abs(charge), x[1])
for x in zip(isospec.masses, isospec.probs)]
return confs
def test_isospec_threshold_(formula, thr): t0 = time() res = iso.IsoThreshold( formula=formula, threshold=thr, get_confs=True ) t1 = time() return t1 - t0
def expand_isotopes(peptide, charge_states=[2, 3]):
'''
Convert peptide to DataFrame of isotopic peaks
Input
Series, should contain 'sequence', 'z+' columns, and model columns
Return
DataFrame with one row for each isotopic peak
columns are:
mz - m/z of ion
ic_XX - ion abundance acording to XX model
z - charge
sequence - peptide sequence
'''
formula = ''.join([
'{}{}'.format(x, y)
for x, y in mass.Composition(peptide['sequence']).items()
])
cluster = IsoSpecPy.IsoThreshold(formula=formula,
threshold=0.005,
absolute=True)
mz0 = cluster.np_masses()
int0 = cluster.np_probs()
mz = np.concatenate([get_ions(mz0, z) for z in charge_states])
ic = np.concatenate(
[int0 * peptide['{}+'.format(z)] for z in charge_states])
charge = np.concatenate(
[np.repeat(z, mz0.shape[0]) for z in charge_states])
result = pd.DataFrame({'mz': mz, 'ic': ic, 'z': charge})
result['sequence'] = peptide['sequence']
for model in params.ion_models:
result['ic_{}'.format(model)] = result['ic'] * peptide[model]
return result
def confs_from_layered_generator(formula, target_prob):
ret = ([], [], [])
for conf in IsoSpecPy.IsoLayered(formula=formula,
prob_to_cover=target_prob,
get_confs=True,
get_minimal_pset=True):
conf = (conf[0], log(conf[1]), conf[2])
ret[0].append(conf[0])
ret[1].append(conf[1])
ret[2].append([item for sublist in conf[2] for item in sublist])
return sort_confs(ret)
def confs_from_ordered_generator(formula, target_prob):
ret = ([], [], [])
prob = 0.0
for conf in IsoSpecPy.IsoOrderedGenerator(formula=formula, get_confs=True):
conf = (conf[0], log(conf[1]), conf[2])
if prob >= target_prob and target_prob < 1.0:
return ret
ret[0].append(conf[0])
prob += exp(conf[1])
ret[1].append(conf[1])
ret[2].append([item for sublist in conf[2] for item in sublist])
return ret
def confs_from_threshold_generator(formula, target_prob):
ret = ([], [], [])
for conf in IsoSpecPy.IsoThresholdGenerator(formula=formula,
threshold=target_prob,
absolute=True,
get_confs=True):
conf = (conf[0], log(conf[1]), conf[2])
ret[0].append(conf[0])
ret[1].append(conf[1])
ret[2].append([item for sublist in conf[2] for item in sublist])
return sort_confs(ret)
def sample_isospec(formula, count, precision):
population = IsoSpecPy.IsoLayeredGenerator(formula, t_prob_hint = precision, reorder_marginals = False)
#population = IsoSpecPy.IsoThresholdGenerator(formula = formula, threshold = -1.0)
#for x in population:
# yield x
'''Performs sampling with replacement from population argument, with
associated probabilities from second argument. The probabilities must
sum to 1. Yields a stream of tuples: (population_member, times_chosen).
Accepts generators as first and second argument. May return duplicate
tuples and tuples with times_chosen == 0.
'''
pprob = 0.0
cprob = 0.0
accumulated = 0
iso_iter = population.__iter__()
while count > 0:
if accumulated > 0:
yield (pop_next, accumulated)
accumulated = 0
pop_next, prob_next = next(iso_iter)
pprob += prob_next
# Beta mode
while (pprob - cprob) * count / (1.0 - cprob) < 1.0:
cprob += _beta_1_b(count) * (1.0 - cprob)
while pprob < cprob:
if accumulated > 0:
yield (pop_next, accumulated)
accumulated = 0
pop_next, prob_next = next(iso_iter)
pprob += prob_next
accumulated += 1
count -= 1
if count == 0: break
if count == 0: break
# Binomial mode
nrtaken = _safe_binom(count, (pprob-cprob)/(1.0-cprob))
accumulated += nrtaken
count -= nrtaken
cprob = pprob
if accumulated > 0:
yield (pop_next, accumulated)
import IsoSpecPy
from math import exp
from IsoSpecPy.Formulas import *
from IsoSpecPy.approximations import approximate_subisotopologues
test_on = horse_myoglobin
test_prob = 0.9999
print("Formula:", test_on, "Probability:", test_prob)
i = IsoSpecPy.Iso(test_on)
print("From C++ code:", list(map(exp,
i.getMarginalLogSizeEstimates(test_prob))))
symbols, _ = IsoSpecPy.ParseFormula(test_on)
dct = approximate_subisotopologues(test_on, test_prob)
print("From Python: ", [dct[s] for s in symbols])
v = IsoSpecPy.IsoTotalProb(formula=test_on,
prob_to_cover=test_prob,
get_confs=True,
get_minimal_pset=True)
acc = [set() for _ in range(v.dimNumber)]
for conf in v.confs:
for i in range(v.dimNumber):
acc[i].add(conf[i])
print("Real:", list(map(len, acc)))
print(len(v), "total confs.")
for conf in IsoSpecPy.IsoThreshold(formula=formula,
threshold=target_prob,
absolute=True,
get_confs=True):
conf = (conf[0], log(conf[1]), conf[2])
ret[0].append(conf[0])
ret[1].append(conf[1])
ret[2].append([item for sublist in conf[2] for item in sublist])
return sort_confs(ret)
is_ok = False
try:
i = IsoSpecPy.IsoThreshold(0.1,
atomCounts=[100],
isotopeMasses=[[1.0, 2.0, 3.0]],
isotopeProbabilities=[[0.0, 0.6, 0.4]])
for x in i:
print(x)
except ValueError:
is_ok = True
assert is_ok
total_confs = 0
for molecule in molecules:
for parameter in parameters:
if not silentish_run:
sprint("{} {}... ".format(molecule, parameter))
old_ordered = OldIsoSpecPy.IsoSpecPy.IsoSpec.IsoFromFormula(
molecule, parameter, method="ordered").getConfs()
import IsoSpecPy
from tqdm import tqdm
t = 0.0
for x in tqdm(xrange(100000)):
i = IsoSpecPy.Iso("C100H100N100O100")
t += i.getTheoreticalAverageMass()
print t
def makeplot(sth):
x = []
y = []
z = []
with open("results205.csv", 'r') as csvfile:
plots = csv.reader(csvfile, delimiter=",")
for row in tqdm(plots):
#x.append(float(row[0]))
#y.append(float(row[1]))
conf1 = next(IsoSpecPy.IsoOrderedGenerator(row[2], get_confs = True).__iter__())[2]
conf2 = next(IsoSpecPy.IsoOrderedGenerator(row[3], get_confs = True).__iter__())[2]
#C H N O S
if sth == 'nucleons':
nukleons1 = [0,0,0]
nukleons2 = [0,0,0]
i = 0
nukleons1[0]+=conf1[i][0]*12
nukleons2[0]+=conf2[i][0]*12
nukleons1[0]+=conf1[i][1]*13
nukleons2[0]+=conf2[i][1]*13
i = 1
nukleons1[0]+=conf1[i][0]*1
nukleons2[0]+=conf2[i][0]*1
nukleons1[0]+=conf1[i][1]*2
nukleons2[0]+=conf2[i][1]*2
i = 2
nukleons1[0]+=conf1[i][0]*14
nukleons2[0]+=conf2[i][0]*14
nukleons1[0]+=conf1[i][1]*15
nukleons2[0]+=conf2[i][1]*15
i = 3
nukleons1[0]+=conf1[i][0]*16
nukleons2[0]+=conf2[i][0]*16
nukleons1[0]+=conf1[i][1]*17
nukleons2[0]+=conf2[i][1]*17
nukleons1[0]+=conf1[i][2]*18
nukleons2[0]+=conf2[i][2]*18
i = 4
nukleons1[0]+=conf1[i][0]*32
nukleons2[0]+=conf2[i][0]*32
nukleons1[0]+=conf1[i][1]*33
nukleons2[0]+=conf2[i][1]*33
nukleons1[0]+=conf1[i][2]*34
nukleons2[0]+=conf2[i][2]*34
nukleons1[0]+=conf1[i][3]*36
nukleons2[0]+=conf2[i][3]*36
diff = abs(nukleons1[0] - nukleons2[0])
if diff != 0:
z.append(diff)
x.append(float(row[0]))
y.append(float(row[1]))
else:
continue
elif sth == 'C':
cs1 = 0
cs2 = 0
cs1 += conf1[0][0] + conf1[0][1]
cs2 += conf2[0][0] + conf2[0][1]
diff = abs(cs1 - cs2)
if diff != 0:
z.append(diff)
x.append(float(row[0]))
y.append(float(row[1]))
else:
continue
elif sth == 'H':
hs1 = 0
hs2 = 0
hs1 += conf1[1][0] + conf1[1][1]
hs2 += conf2[1][0] + conf2[1][1]
diff = abs(hs1 - hs2)
if diff != 0:
z.append(diff)
x.append(float(row[0]))
y.append(float(row[1]))
elif sth == 'N':
ns1 = 0
ns2 = 0
ns1 += conf1[2][0] + conf1[2][1]
ns2 += conf2[2][0] + conf2[2][1]
diff = abs(ns1 - ns2)
if diff != 0:
z.append(diff)
x.append(float(row[0]))
y.append(float(row[1]))
elif sth == 'O':
os1 = 0
os2 = 0
os1 += conf1[3][0] + conf1[3][1] + conf1[3][2]
os2 += conf2[3][0] + conf2[3][1] + conf2[3][2]
diff = abs(os1 - os2)
if diff != 0:
z.append(diff)
x.append(float(row[0]))
y.append(float(row[1]))
elif sth == 'S':
ss1 = 0
ss2 = 0
ss1 += conf1[4][0] + conf1[4][1] + conf1[4][2] + conf1[4][3]
ss2 += conf2[4][0] + conf2[4][1] + conf2[4][2] + conf2[4][3]
diff = abs(ss1 - ss2)
if diff != 0:
z.append(diff)
x.append(float(row[0]))
y.append(float(row[1]))
fig, ax = plt.subplots()
plot = ax.scatter(x, y, c = z, alpha = 0.3, edgecolors = None, cmap = 'YlGnBu', s = 5.0)
legend = ax.legend(*plot.legend_elements(alpha = 1.0), loc = "lower right", title=('%s' % (sth)) + ' ' + "number difference")
ax.add_artist(legend)
plt.xlabel('Mean mass difference')
plt.ylabel('Wasserstein distance')
plt.title('Mean mass difference, Wasserstein distance and' + ' ' + ('%s' % (sth)) + ' ' + 'difference plot ')
#plt.gray()
plt.legend()
#plt.show()
plt.savefig("Plot" + ('%s' % (sth)) + ".png")
# 14C isn't normally considered in the isotopic distribution, here we add an extra isotope to the standard ones
radiolabelled_carbon_masses = PeriodicTbl.symbol_to_masses["C"] + (
14.003241989, )
# Assuming that the labelling was only 95% efficient, that is only 95%
# of the radiolabel atoms have standard C replaced with 14C. Non-replaced atoms have standard
# isotopic abundance (realtive to each other)
normal_carbon_probs = PeriodicTbl.symbol_to_probs["C"]
radiolabelled_carbon_probs = (0.05 * normal_carbon_probs[0],
0.05 * normal_carbon_probs[1], 0.95)
i = IsoSpecPy.IsoTotalProb(
formula="C4H12O6", # The formula for glucose, sans the radiolabel atoms
# Here we specify additional "elements" which occur *in addition* to those from the formula
atomCounts=(2, ),
isotopeMasses=(radiolabelled_carbon_masses, ),
isotopeProbabilities=(radiolabelled_carbon_probs, ),
# And the rest of parameters for configuration
prob_to_cover=0.99,
get_confs=True)
# Radiolabelling (or isotopic labelling) with more than one element looks like this:
# Let's say we wanted to have glucose with one 14C carbon, and two deuteriums, all with 95% probability
# Then it would be:
#i = IsoSpecPy.IsoLayeredGenerator(formula = "C5H10O6", # The formula for glucose, sans the radiolabel atoms
# atomCounts = (1, 2),
# isotopeMasses = (radiolabelled_carbon_masses, PeriodicTbl.symbol_to_masses["H"]),
# isotopeProbabilities = (radiolabelled_carbon_probs, (0.05, 0.95)),
# # And the rest of parameters for configuration
# prob_to_cover = 0.99,
# get_confs=True)
while S.next():
print(S.confs_prob, S.chasing_prob)
yield (S.current_conf, S.current_count)
from IsoSpecPy.Formulas import *
from scipy.stats import chisquare
import sys
if __name__ == '__main__':
test_mol = surcose
count = 10000000
print("Starting...")
X = sorted(x for x in IsoSpecPy.IsoThresholdGenerator(formula=test_mol, threshold=sys.float_info.min, reorder_marginals = False) if x[1] > 0)
print("No configs: " + str(len(X)))
Y = dict([(v[0], 0) for v in X])
#print(Y)
s = 0
for x in sample_ciic(test_mol, count, 0.999999):
print(x)
Y[x[0]] = x[1]
s += x[1]
print("S:", s)
assert s == count
#print(X)
def sample_isospec2(formula, count, precision):
population = IsoSpecPy.IsoLayeredGenerator(formula, t_prob_hint = precision, reorder_marginals = False)
S = Sampler(population, count, precision, 1.0)
while S.advance():
yield S.current()
'''
import IsoSpecPy
from math import exp
try:
if IsoSpecPy.__version__[:4] != '2.1.':
raise AttributeError
except AttributeError:
print(
"This file is meant to be used with IsoSpecPy version 2.0.X. You seem to have a different version installed on your system."
)
import sys
sys.exit(-1)
i = IsoSpecPy.IsoTotalProb(formula="H2O1", prob_to_cover=0.999, get_confs=True)
print(
"Calculating isotopic distribution of water. Here's a list of configurations necessary to cover at least 0.999 of total probability:"
)
for mass, prob, conf in i:
print("")
print("Mass: " + str(mass))
print("probability: " + str(prob))
print("Number of Protium atoms: " + str(conf[0][0]))
print("Number of Deuterium atoms: " + str(conf[0][1]))
print("Number of O16 atoms: " + str(conf[1][0]))
print("Number of O17 atoms: " + str(conf[1][1]))
print("Number of O18 atoms: " + str(conf[1][2]))
def sample_ciic(formula, count, precision):
population = IsoSpecPy.IsoLayeredGenerator(formula, t_prob_hint = precision, reorder_marginals = False)
S = CIIC(population, count, precision, -1.0)
while S.next():
print(S.confs_prob, S.chasing_prob)
yield (S.current_conf, S.current_count)
def get_real_confs(formula, P): confs = [set() for el, atom_cnt in parse(formula) ] for _, _, C in iso.IsoLayered(formula=formula, prob_to_cover=P, get_confs = True): for i, x in enumerate(C): confs[i].add(x) return np.array([len(c) for c in confs])
def count_totalprob(mass, formula, prob):
s = IsoSpecPy.IsoTotalProb(prob, formula)
s.normalize()
return (mass, formula, s)
from __future__ import print_function
import IsoSpecPy
from IsoSpecPy.Formulas import *
import math
try:
math.isclose
except AttributeError:
def isclose(a, b, rel_tol=1e-09, abs_tol=0.0):
return abs(a-b) <= max(rel_tol * max(abs(a), abs(b)), abs_tol)
math.isclose = isclose
glu = IsoSpecPy.IsoThreshold(0.0, formula=glucose)
ca = IsoSpecPy.IsoThreshold(0.0, formula=caffeine)
print("Checking Wasserstein distance...", end=' ')
print(ca.wassersteinDistance(glu), end=' ')
assert(math.isclose(ca.wassersteinDistance(glu), 14.03495145836358))
print("OK!")
print("Checking normalization... ", end='')
ubiq = IsoSpecPy.IsoTotalProb(0.9999, ubiquitin)
print(ubiq.total_prob(), end=' ')
assert(math.isclose(ubiq.total_prob(), 0.9999, rel_tol=0.01))
ubiq = IsoSpecPy.IsoTotalProb(0.9999, ubiquitin)
ubiq.scale(0.5)
assert(math.isclose(ubiq.total_prob(), 0.9999*0.5, rel_tol=0.01))
ubiq._recalculate_everything()
assert(math.isclose(ubiq.total_prob(), 0.9999*0.5, rel_tol=0.01))
def progr(window):
with open('testowy.txt') as file:
with open('test.csv', 'w', newline='') as write_file:
writer = csv.writer(write_file)
L = []
P = []
S = []
masses = []
formulas = []
limes = float('inf')
for line in tqdm(file, total=177754527):
average_mass = float(line.split(",")[0])
formula = line.split(",")[1].strip()
if average_mass <= limes:
L.append((average_mass, formula))
limes = average_mass + window
#print(L)
else:
#rob IsoSpecPy
if len(L) > 1:
#print(L)
combined_s = []
combined_mass = []
combined_formulas = []
for i in range(len(L)):
s = IsoSpecPy.IsoTotalProb(0.99, L[i][1])
s.normalize()
mass = L[i][0]
S.append(s)
masses.append(mass)
formulas.append(L[i][1])
for i in range(len(L)):
for j in range(i + 1, len(L)):
combined_s.append((S[i], S[j]))
combined_mass.append((masses[i], masses[j]))
combined_formulas.append(
(formulas[i], formulas[j]))
S = []
masses = []
#print(len(combined_s))
for i in range(len(combined_s)):
wasserstein = combined_s[i][0].wassersteinDistance(
combined_s[i][1])
mass_difference = abs(combined_mass[i][0] -
combined_mass[i][1])
used_formulas = combined_formulas[i]
#print(mass_difference, wasserstein, used_formulas)
writer.writerow([(mass_difference, wasserstein,
used_formulas)])
#print("napisalem")
formulas = []
combined_s = []
combined_mass = []
combined_formulas = []
L.append((average_mass, formula))
for i in range(len(L)):
if L[i][0] == average_mass:
P.append(L[i])
#print(P)
L = P
P = []
limes = average_mass + window
def generate_isotopologues(formula_entry, smiles_entry, resolution_entry):
formula = ""
if formula_entry is not None and len(formula_entry):
formula = formula_entry
else:
# Getting exact mass
url = "https://gnps-structure.ucsd.edu/formula?smiles={}".format(
urllib.parse.quote(smiles_entry))
r = requests.get(url)
formula = (r.text)
i = IsoSpecPy.IsoTotalProb(
formula=
formula, # The formula for glucose, sans the radiolabel atoms # And the rest of parameters for configuration
prob_to_cover=0.99,
get_confs=True)
output_list = []
for mass, prob, conf in i:
output_dict = {}
output_dict["prob"] = prob
output_dict["mz"] = mass - 0.00054858
output_list.append(output_dict)
table_fig = dash_table.DataTable(
columns=[{
"name": i,
"id": i,
"deletable": True,
"selectable": True
} for i in ["mz", "prob"]],
data=output_list,
editable=True,
filter_action="native",
sort_action="native",
sort_mode="multi",
column_selectable="single",
selected_columns=[],
selected_rows=[],
page_action="native",
page_current=0,
page_size=10,
)
# Drawing Figure
main_mz = output_list[0]["mz"]
delta_m = main_mz / float(resolution_entry)
sigma = delta_m / 2.355
display_bins = 0.02
display_bins = sigma
import numpy as np
mz_grid = np.arange(output_list[0]["mz"] - 1, output_list[-1]["mz"] + 1,
display_bins)
intensity = np.zeros_like(mz_grid)
for peak in output_list:
# Add gaussian peak shape centered around each theoretical peak
intensity += peak["prob"] * np.exp(
-(mz_grid - peak["mz"])**2 /
(2 * sigma)) / (np.sqrt(2 * np.pi) * sigma)
# Normalize profile to 0-100
intensity = (intensity / intensity.max()) * 100
df = pd.DataFrame()
df["mz"] = mz_grid
df["intensity"] = intensity
line_fig = px.line(
df,
x="mz",
y="intensity",
title='Isotopologue Distribution - {} - Resolution - {}'.format(
formula, resolution_entry))
return [[table_fig, dcc.Graph(figure=line_fig)]]