def qeds(df):
N = df.shape[0]
n = int(N * threshold)
df = df.sort_values('score')
df = df[:n] # top 10 % statistics
smiles = df.smile
smiles = [s for s in smiles if Chem.MolFromSmiles(s) is not None]
mols = [Chem.MolFromSmiles(s) for s in smiles]
q = np.array([QED.default(m) for m in mols])
return (np.mean(q), np.std(q))
N = 3
# Plot : 2500 first :
optimol = pd.read_csv(
os.path.join(script_dir, '..',
'cbas/slurm/results/big_new_lr/optimol_scored.csv'))
optimol = optimol[:10000]
optimol = optimol.sample(10)
# Top molecules
samples = optimol.sort_values('score')
smiles, scores = samples.smile, samples.score
smiles = smiles[:N]
mols = [Chem.MolFromSmiles(s) for s in smiles]
qeds = np.array([QED.default(m) for m in mols])
sas = [calculateScore(m) for m in mols]
scores = scores[:N]
img1 = Draw.MolsToGridImage(mols,
molsPerRow=1,
useSVG=False,
legends=[f'{sc:.2f}' for sc in scores])
img2 = Draw.MolsToGridImage(mols, molsPerRow=1, useSVG=False)
svg2pdf(str(img), write_to='optimol_samp_2.pdf')
"""
['Cc1ccccc1CCC(=O)OCC(=O)NCCc1ccc2ccccc2c1'
'COCC1CCCCN(C(=O)C(=O)NCc2cc3ccccc3c3ccccc23)C1'
'Cc1ccccc1CC1CCCN1C(=O)CC1Cc2ccccc2NC1=O']
"""
prop_names = [
'QED', 'logP', 'molWt', 'maxCharge', 'minCharge', 'valence', 'TPSA', 'HBA',
'HBD', 'jIndex'
]
for name in prop_names:
d[f'{name}'] = []
for i, s in enumerate(smiles):
if (i % 10000 == 0):
print(i)
m = Chem.MolFromSmiles(s)
if (m == None or 'i' in s or '.' in s):
DUD = DUD.drop(i)
print(s, i)
else:
d['QED'].append(QED.default(m))
d['logP'].append(Crippen.MolLogP(m))
d['molWt'].append(Descriptors.MolWt(m))
d['maxCharge'].append(Descriptors.MaxPartialCharge(m))
d['minCharge'].append(Descriptors.MinPartialCharge(m))
d['valence'].append(Descriptors.NumValenceElectrons(m))
d['TPSA'].append(rdMolDescriptors.CalcTPSA(m))
d['HBA'].append(rdMolDescriptors.CalcNumHBA(m))
d['HBD'].append(rdMolDescriptors.CalcNumHBD(m))
d['jIndex'].append(GraphDescriptors.BalabanJ(m))
df = pd.DataFrame.from_dict(d)
df_merge = pd.merge(df, DUD, on=df.index)
#df_merge.to_csv('/home/mcb/jboitr/data/DUD_full.csv')
