def predict_ms(self):
base = 'RIKEN_PlaSMA_'
mode = self.ModInput.currentText()[0:3]
eneg = self.EgyInput.currentText()[0:2]
smi = self.SmiInput.toPlainText()
model = load_model(base + mode + '_' + eneg)
ms = model_predict(smi, model)
# plot_ms(ms)
self.F = MyFigure(width=3, height=2, dpi=100)
self.F.axes.cla()
self.F.axes.vlines(ms['mz'], np.zeros(ms.shape[0]),
np.array(ms['intensity']), 'red')
self.F.axes.axhline(0, color='black')
self.gridlayout = QtWidgets.QGridLayout(self.groupBox)
self.gridlayout.addWidget(self.F, 0, 1)
self.gridlayout.deleteLater()
def identification(ms, candidates, model, method='correlation'):
smiles = []
scores = []
inchis = []
masses = []
pred_ms = []
if method == 'residual':
score = ms_residual
elif method == 'correlation':
score = ms_correlation
else:
score = ms_jaccard
if 'InChI=' in candidates[0]:
read_candidate = Chem.MolFromInchi
else:
read_candidate = Chem.MolFromSmiles
for i in candidates:
try:
mol = read_candidate(i)
smi = Chem.MolToSmiles(mol)
inchi = Chem.MolToInchi(mol)
mass = CalcExactMolWt(mol)
except:
continue
pms = model_predict(smi, model)
scr = score(ms, pms)
smiles.append(smi)
inchis.append(inchi)
scores.append(scr)
masses.append(mass)
pred_ms.append(pms)
output = pd.DataFrame({
'SMILES': smiles,
'InChI': inchis,
'mass': masses,
'scores': scores,
'pred_ms': pred_ms
})
output = output.sort_values('scores', ascending=False)
return output
plt.xlabel('epoch')
plt.legend(['train forward', 'train reverse', 'test forward', 'test reverse'],
loc='upper left')
plt.show()
model_json = model.to_json()
save_path = 'Model/RIKEN_PlaSMA_Pos_10'
with open(save_path + '.json', "w") as json_file:
json_file.write(model_json)
model.save_weights(save_path + '.h5')
result = pd.DataFrame(
columns=['idx', 'smiles', 'DeepFrag', 'CFM_10', 'CFM_20', 'CFM_40'])
for i in tqdm(test_index):
try:
smi = smiles[i]
ms_cfm = cfm_predict(smi)
ms_real = ms[i]
ms_pred = model_predict(smi, model)
trans = ms_correlation(ms_real, ms_pred)
cfm_10 = ms_correlation(ms_real, ms_cfm['low_energy'])
cfm_20 = ms_correlation(ms_real, ms_cfm['medium_energy'])
cfm_40 = ms_correlation(ms_real, ms_cfm['high_energy'])
except:
continue
'''
plot_compare_ms(ms_real, ms_cfm['medium_energy'])
plot_compare_ms(ms_real, ms_pred)
'''
result.loc[len(result)] = [i, smi, trans, cfm_10, cfm_20, cfm_40]
result.to_csv('Result/RIKEN_PlaSMA_Pos_10.csv')
continue
data = pd.DataFrame(np.array(data))
data.columns = ['mz', 'intensity']
modes.append(ion_mode)
ms.append(data)
smiles.append(smi)
energies.append(energy)
summary = pd.DataFrame({'smiles': smiles, 'ion_mode': modes, 'energy': energies})
# example 1
idx = 1297
smi = smiles[idx]
mol = Chem.MolFromSmiles(smi)
t1 = time.time()
ms_pred = model_predict(smi, model)
t2 = time.time()
ms_cfm = cfm_predict(smi)
t3 = time.time()
ms_pretrain = model_predict(smi, pretrain)
ms_real = ms[idx]
plot_compare_ms(ms_real, ms_pretrain)
plot_compare_ms(ms_pretrain, ms_cfm['low_energy'])
plot_compare_ms(ms_real, ms_pred)
plot_compare_ms(ms_real, ms_cfm['low_energy'])
print ('computing time of CFM is: ' + str(t3-t2) + ' s')
print ('computing time of DeepFrag is: ' + str(t2-t1) + ' s')
## precursor
Chem.MolFromSmiles('Oc1cc(O)c2c(c1)OC(c1ccc(O)c(O)c1)C([OH2+])C2')