File size: 9,649 Bytes
fb5f46a |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 |
# -*- coding: utf-8 -*-
"""
Created on Tue Aug 29 10:34:38 2023
@author: DELL
"""
import base64
import numpy as np
import pandas as pd
from rdkit import Chem
from rdkit.Chem import DataStructs, AllChem
from sklearn.metrics.pairwise import cosine_similarity
import matchms.filtering as msfilters
from molmass import Formula
from matchms import calculate_scores
from matchms.similarity import CosineGreedy
from spec2vec import SpectrumDocument
from spec2vec.vector_operations import calc_vector
from core.pycdk import IsotopeFromString, IsotopeSimilarity
from core.pubchem import retrieve_by_formula, retrieve_by_exact_mass
from core.pubchem import retrieve_by_formula_database, retrieve_by_exact_mass_database
def spectrum_processing(s):
"""This is how one would typically design a desired pre- and post-
processing pipeline."""
s = msfilters.default_filters(s)
if ('adduct_type' in s.metadata.keys()) and ('adduct' not in s.metadata.keys()):
s.set('adduct', s.get('adduct_type'))
s = msfilters.correct_charge(s)
s = msfilters.add_parent_mass(s)
s = msfilters.add_losses(s)
s = msfilters.normalize_intensities(s)
s = msfilters.select_by_mz(s, mz_from=0, mz_to=1000)
return s
def get_formula_mass(formula):
formula = formula.replace('+','').replace('-','')
f = Formula(formula)
return f.isotope.mass
def search_candidates(s, database, ms1_tolerence = 20):
if 'formula' in s.metadata.keys():
formula = s.metadata['formula']
candidate = retrieve_by_formula_database(formula, database)
return candidate
if len(candidate) == 0:
try:
candidate = retrieve_by_formula(formula)
return candidate
except:
return None
elif 'parent_mass' in s.metadata.keys():
mass = float(s.metadata['parent_mass'])
candidate = retrieve_by_exact_mass_database(mass, database, ppm = ms1_tolerence)
return candidate
if len(candidate) == 0:
try:
candidate = retrieve_by_exact_mass(mass)
return candidate
except:
return None
else:
return None
def calc_isotope_score(s, candidate_mol):
isotope_score = []
if s.get('isotope_mz') and s.get('isotope_intensity'):
isotope_mz = base64.b64decode(s.get('isotope_mz').split("'")[1]).decode("ascii").replace('\n', '')
isotope_intensity = base64.b64decode(s.get('isotope_intensity').split("'")[1]).decode("ascii").replace('\n', '')
isotope_mz = [float(s) for s in isotope_mz.replace('[', '').replace(']', '').split(' ') if s != '']
isotope_intensity = [float(s) for s in isotope_intensity.replace('[', '').replace(']', '').split(' ') if s != '']
isotope_mz = np.array(isotope_mz)
adduct_mz = isotope_mz[np.argmax(isotope_intensity)] - s.metadata['parent_mass']
isotope_mz = isotope_mz - adduct_mz
isotope_intensity = np.array(isotope_intensity)
isotope_intensity = isotope_intensity / max(isotope_intensity)
isotope_pattern = np.vstack((isotope_mz, isotope_intensity)).T
else:
return None
for i in range(len(candidate_mol)):
try:
formula = AllChem.CalcMolFormula(candidate_mol[i])
isotope_ref = IsotopeFromString(formula, minI=0.001)
except:
isotope_score.append(0)
isotope_score.append(IsotopeSimilarity(isotope_pattern, isotope_ref, 10))
return isotope_score
def calc_deepmass_score(s, candidate_mol, reference_mol, query_vector, reference_vector):
get_fp = lambda x: AllChem.GetMorganFingerprintAsBitVect(x, radius=2)
get_sim = lambda x, y: DataStructs.FingerprintSimilarity(x, y)
get_corr = lambda x, y: cosine_similarity([x], [y])[0][0]
k, reference_fp = [], []
for i, m in enumerate(reference_mol):
try:
reference_fp.append(get_fp(m))
k.append(i)
except:
pass
if len(k) != len(reference_mol):
k = np.array(k)
reference_mol = np.array(reference_mol)[k]
reference_vector = np.array(reference_vector)[k,:]
deepmass_score = []
for i in range(len(candidate_mol)):
try:
candidate_fp_i = get_fp(candidate_mol[i])
except:
deepmass_score.append(0)
candidate_vecsim_i = [get_corr(query_vector, reference_vector_) for reference_vector_ in reference_vector]
candidate_vecsim_i = np.array(candidate_vecsim_i)
candidate_fpsim_i = [get_sim(candidate_fp_i, reference_fp_) for reference_fp_ in reference_fp]
candidate_fpsim_i = np.array(candidate_fpsim_i)
top20 = np.argsort(-np.array(candidate_fpsim_i))[:20]
candidate_score_i = np.sqrt(np.sum(candidate_vecsim_i[top20] * candidate_fpsim_i[top20]))
deepmass_score.append(candidate_score_i / 20)
deepmass_score = np.array(deepmass_score)
deepmass_score /= np.max(deepmass_score)
return deepmass_score
def calc_wt_score(s, candidate_mol):
candidate_mass = [AllChem.CalcExactMolWt(m) for m in candidate_mol]
diff_mass = np.array([abs(m - float(s.get['parent_mass'])) for m in candidate_mass])
wt_score = 1 - 50000 * diff_mass / float(s.get['parent_mass'])
return wt_score
def identify_unknown(s, p, model, references, database):
candidate = search_candidates(s, database)
if candidate is None:
return s
if len(candidate) == 0:
return s
candidate_mol = [Chem.MolFromSmiles(s) for s in candidate['CanonicalSMILES']]
query_vector = calc_vector(model, SpectrumDocument(s, n_decimals=2))
xq = np.array(query_vector).astype('float32')
I, D = p.knn_query(xq, 300)
reference_spectrum = np.array(references)[I[0,:]]
reference_smile = [s.metadata['smiles'] for s in reference_spectrum]
reference_mol = [Chem.MolFromSmiles(s) for s in reference_smile]
reference_vector = np.array(p.get_items(I[0, :]))
candidate_deepmass_score = calc_deepmass_score(s, candidate_mol, reference_mol, query_vector, reference_vector)
candidate['DeepMass Score'] = np.round(candidate_deepmass_score, 4)
if s.get('formula') is None:
candidate_wt_score = calc_wt_score(s, candidate_mol)
candidate['MolWt Score'] = np.round(candidate_wt_score, 4)
if s.get('isotope_mz') and s.get('isotope_intensity'):
candidate_isotopic_score = calc_isotope_score(s, candidate_mol)
candidate['Isotope Score'] = np.round(candidate_isotopic_score, 4)
candidate['Consensus Score'] = 0.8*candidate['DeepMass Score'] + 0.1*candidate['Isotope Score'] + 0.1*candidate['MolWt Score']
else:
candidate['Consensus Score'] = 0.8*candidate['DeepMass Score'] + 0.2*candidate['MolWt Score']
candidate = candidate.sort_values('Consensus Score', ignore_index = True, ascending = False)
else:
candidate = candidate.sort_values('DeepMass Score', ignore_index = True, ascending = False)
s.set('annotation', candidate)
s.set('reference', reference_spectrum)
return s
def match_spectrum(s, precursors, references):
precursor = s.get('precursor_mz')
if precursor is None:
return s
lb, ub = precursor - 0.05, precursor + 0.05
li = np.searchsorted(precursors, lb)
ui = np.searchsorted(precursors, ub)
if ui <= li:
return s
match_scores = calculate_scores(references = references[li:ui], queries = [s], similarity_function = CosineGreedy())
# print(match_scores.scores)
match_scores = np.array([s[0].tolist()[0] for s in match_scores.scores])
w = np.argsort(-match_scores)
match_scores = match_scores[w]
reference = np.array(references)[li:ui][w]
annotation, inchikeys = [], []
for i, r in enumerate(reference):
mol = Chem.MolFromSmiles(r.get('smiles'))
score = match_scores[i]
if mol is None:
continue
inchikey = r.get('inchikey')
if inchikey == '':
continue
title = r.get('compound_name')
smiles = Chem.MolToSmiles(mol)
try:
formula = AllChem.CalcMolFormula(mol)
except:
formula = ''
if inchikey not in inchikeys:
inchikeys.append(inchikey)
annotation.append([title, formula, smiles, inchikey, score])
annotation = pd.DataFrame(annotation, columns = ['Title', 'MolecularFormula', 'CanonicalSMILES', 'InChIKey', 'Matching Score'])
if s.get('formula') is not None:
annotation = annotation[annotation['MolecularFormula'] == s.get('formula')]
annotation = annotation.reset_index(drop = True)
s.set('annotation', annotation)
s.set('reference', reference)
return s
if __name__ == '__main__':
'''
import hnswlib
import pickle
import pandas as pd
from matchms.importing import load_from_mgf
from gensim.models import Word2Vec
model = Word2Vec.load("model/Ms2Vec_allGNPSpositive.hdf5")
p = hnswlib.Index(space='l2', dim=300)
p.load_index('data/references_index_positive_spec2vec.bin')
with open('data/references_spectrums_positive.pickle', 'rb') as file:
references = pickle.load(file)
references = np.array(references)
precursors = [s.get('precursor_mz') for s in references]
precursors = np.array(precursors)
spectrums = [s for s in load_from_mgf("D:/DeepMASS2_Data_Processing/Example/CASMI/all_casmi.mgf")]
s = spectrums[200]
s = identify_unknown(s, p, model, reference, database)
'''
|