def smiles_to_all_labels(df): smilesList = df['SMILES'] feature_df = df.copy() # get all functions of modules all_lipinski = inspect.getmembers(l, inspect.isfunction) all_fragments = inspect.getmembers(f, inspect.isfunction) # bad features have the same value for all our compounds bad_features = [] for (columnName, columnData) in df.iteritems(): if (len(set(columnData.values)) == 1): bad_features.append(columnName) # add fragment features for i in range(len(all_fragments)): new_col = [] # exclude attributes which start with _ and exclude bad features if all_fragments[i][0].startswith( '_') == False and all_fragments[i][0] not in bad_features: for smiles in smilesList: molecule = chem.MolFromSmiles(smiles) mol_method = all_fragments[i][1](molecule) new_col.append(mol_method) # add new col with feature name to our df feature_df[all_fragments[i][0]] = new_col print('fragments over') # add lipinski features for i in range(len(all_lipinski)): new_col = [] if all_lipinski[i][0].startswith( '_') == False and all_fragments[i][0] not in bad_features: for smiles in smilesList: molecule = chem.MolFromSmiles(smiles) mol_method = all_lipinski[i][1](molecule) new_col.append(mol_method) feature_df[all_lipinski[i][0]] = new_col print('lipinski over') new_col = [] for smiles in smilesList: molecule = chem.MolFromSmiles(smiles) new_col.append(f.fr_Al_COO(molecule)) feature_df["fr_Al_COO"] = new_col # new_col = [] for smiles in smilesList: molecule = chem.MolFromSmiles(smiles) new_col.append(l.HeavyAtomCount(molecule)) feature_df["HeavyAtomCount"] = new_col # add getnumatoms as feature new_col = [] for smiles in smilesList: molecule = chem.MolFromSmiles(smiles) new_col.append(molecule.GetNumAtoms()) feature_df["GetNumAtoms"] = new_col # add CalcExactMolWt as feature new_col = [] for smiles in smilesList: molecule = chem.MolFromSmiles(smiles) new_col.append(molDesc.CalcExactMolWt(molecule)) feature_df["CalcExactMolWt"] = new_col # print('other over') return feature_df
def calc_rdkit(mol): descriptors = pd.Series( np.array([ Crippen.MolLogP(mol), Crippen.MolMR(mol), Descriptors.FpDensityMorgan1(mol), Descriptors.FpDensityMorgan2(mol), Descriptors.FpDensityMorgan3(mol), Descriptors.FractionCSP3(mol), Descriptors.HeavyAtomMolWt(mol), Descriptors.MaxAbsPartialCharge(mol), Descriptors.MaxPartialCharge(mol), Descriptors.MinAbsPartialCharge(mol), Descriptors.MinPartialCharge(mol), Descriptors.MolWt(mol), Descriptors.NumRadicalElectrons(mol), Descriptors.NumValenceElectrons(mol), EState.EState.MaxAbsEStateIndex(mol), EState.EState.MaxEStateIndex(mol), EState.EState.MinAbsEStateIndex(mol), EState.EState.MinEStateIndex(mol), EState.EState_VSA.EState_VSA1(mol), EState.EState_VSA.EState_VSA10(mol), EState.EState_VSA.EState_VSA11(mol), EState.EState_VSA.EState_VSA2(mol), EState.EState_VSA.EState_VSA3(mol), EState.EState_VSA.EState_VSA4(mol), EState.EState_VSA.EState_VSA5(mol), EState.EState_VSA.EState_VSA6(mol), EState.EState_VSA.EState_VSA7(mol), EState.EState_VSA.EState_VSA8(mol), EState.EState_VSA.EState_VSA9(mol), Fragments.fr_Al_COO(mol), Fragments.fr_Al_OH(mol), Fragments.fr_Al_OH_noTert(mol), Fragments.fr_aldehyde(mol), Fragments.fr_alkyl_carbamate(mol), Fragments.fr_alkyl_halide(mol), Fragments.fr_allylic_oxid(mol), Fragments.fr_amide(mol), Fragments.fr_amidine(mol), Fragments.fr_aniline(mol), Fragments.fr_Ar_COO(mol), Fragments.fr_Ar_N(mol), Fragments.fr_Ar_NH(mol), Fragments.fr_Ar_OH(mol), Fragments.fr_ArN(mol), Fragments.fr_aryl_methyl(mol), Fragments.fr_azide(mol), Fragments.fr_azo(mol), Fragments.fr_barbitur(mol), Fragments.fr_benzene(mol), Fragments.fr_benzodiazepine(mol), Fragments.fr_bicyclic(mol), Fragments.fr_C_O(mol), Fragments.fr_C_O_noCOO(mol), Fragments.fr_C_S(mol), Fragments.fr_COO(mol), Fragments.fr_COO2(mol), Fragments.fr_diazo(mol), Fragments.fr_dihydropyridine(mol), Fragments.fr_epoxide(mol), Fragments.fr_ester(mol), Fragments.fr_ether(mol), Fragments.fr_furan(mol), Fragments.fr_guanido(mol), Fragments.fr_halogen(mol), Fragments.fr_hdrzine(mol), Fragments.fr_hdrzone(mol), Fragments.fr_HOCCN(mol), Fragments.fr_imidazole(mol), Fragments.fr_imide(mol), Fragments.fr_Imine(mol), Fragments.fr_isocyan(mol), Fragments.fr_isothiocyan(mol), Fragments.fr_ketone(mol), Fragments.fr_ketone_Topliss(mol), Fragments.fr_lactam(mol), Fragments.fr_lactone(mol), Fragments.fr_methoxy(mol), Fragments.fr_morpholine(mol), Fragments.fr_N_O(mol), Fragments.fr_Ndealkylation1(mol), Fragments.fr_Ndealkylation2(mol), Fragments.fr_NH0(mol), Fragments.fr_NH1(mol), Fragments.fr_NH2(mol), Fragments.fr_Nhpyrrole(mol), Fragments.fr_nitrile(mol), Fragments.fr_nitro(mol), Fragments.fr_nitro_arom(mol), Fragments.fr_nitro_arom_nonortho(mol), Fragments.fr_nitroso(mol), Fragments.fr_oxazole(mol), Fragments.fr_oxime(mol), Fragments.fr_para_hydroxylation(mol), Fragments.fr_phenol(mol), Fragments.fr_phenol_noOrthoHbond(mol), Fragments.fr_phos_acid(mol), Fragments.fr_phos_ester(mol), Fragments.fr_piperdine(mol), Fragments.fr_piperzine(mol), Fragments.fr_priamide(mol), Fragments.fr_prisulfonamd(mol), Fragments.fr_pyridine(mol), Fragments.fr_quatN(mol), Fragments.fr_SH(mol), Fragments.fr_sulfide(mol), Fragments.fr_sulfonamd(mol), Fragments.fr_sulfone(mol), Fragments.fr_term_acetylene(mol), Fragments.fr_tetrazole(mol), Fragments.fr_thiazole(mol), Fragments.fr_thiocyan(mol), Fragments.fr_thiophene(mol), Fragments.fr_unbrch_alkane(mol), Fragments.fr_urea(mol), GraphDescriptors.BalabanJ(mol), GraphDescriptors.BertzCT(mol), GraphDescriptors.Chi0(mol), GraphDescriptors.Chi0n(mol), GraphDescriptors.Chi0v(mol), GraphDescriptors.Chi1(mol), GraphDescriptors.Chi1n(mol), GraphDescriptors.Chi1v(mol), GraphDescriptors.Chi2n(mol), GraphDescriptors.Chi2v(mol), GraphDescriptors.Chi3n(mol), GraphDescriptors.Chi3v(mol), GraphDescriptors.Chi4n(mol), GraphDescriptors.Chi4v(mol), GraphDescriptors.HallKierAlpha(mol), GraphDescriptors.Ipc(mol), GraphDescriptors.Kappa1(mol), GraphDescriptors.Kappa2(mol), GraphDescriptors.Kappa3(mol), Lipinski.HeavyAtomCount(mol), Lipinski.NHOHCount(mol), Lipinski.NOCount(mol), Lipinski.NumAliphaticCarbocycles(mol), Lipinski.NumAliphaticHeterocycles(mol), Lipinski.NumAliphaticRings(mol), Lipinski.NumAromaticCarbocycles(mol), Lipinski.NumAromaticHeterocycles(mol), Lipinski.NumAromaticRings(mol), Lipinski.NumHAcceptors(mol), Lipinski.NumHDonors(mol), Lipinski.NumHeteroatoms(mol), Lipinski.NumRotatableBonds(mol), Lipinski.NumSaturatedCarbocycles(mol), Lipinski.NumSaturatedHeterocycles(mol), Lipinski.NumSaturatedRings(mol), Lipinski.RingCount(mol), MolSurf.LabuteASA(mol), MolSurf.PEOE_VSA1(mol), MolSurf.PEOE_VSA10(mol), MolSurf.PEOE_VSA11(mol), MolSurf.PEOE_VSA12(mol), MolSurf.PEOE_VSA13(mol), MolSurf.PEOE_VSA14(mol), MolSurf.PEOE_VSA2(mol), MolSurf.PEOE_VSA3(mol), MolSurf.PEOE_VSA4(mol), MolSurf.PEOE_VSA5(mol), MolSurf.PEOE_VSA6(mol), MolSurf.PEOE_VSA7(mol), MolSurf.PEOE_VSA8(mol), MolSurf.PEOE_VSA9(mol), MolSurf.SlogP_VSA1(mol), MolSurf.SlogP_VSA10(mol), MolSurf.SlogP_VSA11(mol), MolSurf.SlogP_VSA12(mol), MolSurf.SlogP_VSA2(mol), MolSurf.SlogP_VSA3(mol), MolSurf.SlogP_VSA4(mol), MolSurf.SlogP_VSA5(mol), MolSurf.SlogP_VSA6(mol), MolSurf.SlogP_VSA7(mol), MolSurf.SlogP_VSA8(mol), MolSurf.SlogP_VSA9(mol), MolSurf.SMR_VSA1(mol), MolSurf.SMR_VSA10(mol), MolSurf.SMR_VSA2(mol), MolSurf.SMR_VSA3(mol), MolSurf.SMR_VSA4(mol), MolSurf.SMR_VSA5(mol), MolSurf.SMR_VSA6(mol), MolSurf.SMR_VSA7(mol), MolSurf.SMR_VSA8(mol), MolSurf.SMR_VSA9(mol), MolSurf.TPSA(mol) ])) return descriptors
def create_features(data, types="train"): if types == "train": y = np.array(data['ACTIVE'].astype(int)) elif types == "test": y = None data = data[["SMILES"]] data["SMILES_str"] = data["SMILES"] data["SMILES"] = data["SMILES"].apply(lambda x: Chem.MolFromSmiles(x)) data["NumAtoms"] = data["SMILES"].apply( lambda x: x.GetNumAtoms()) #l.HeavyAtomCount(m) data["ExactMolWt"] = data["SMILES"].apply(lambda x: d.CalcExactMolWt(x)) data["fr_Al_COO"] = data["SMILES"].apply(lambda x: f.fr_Al_COO(x)) data["HsNumAtoms"] = data["SMILES"].apply( lambda x: Chem.AddHs(x).GetNumAtoms()) #to have the hydrogens explicitly present BondType = [[str(x.GetBondType()) for x in m.GetBonds()] for m in data["SMILES"]] BondType = [" ".join(x) for x in BondType] vec = CountVectorizer().fit(BondType) train_tfidf = vec.transform(BondType).todense() # 转化为更直观的一般矩阵 vocabulary = vec.vocabulary_ train_tfidf = pd.DataFrame(train_tfidf) train_tfidf.columns = vocabulary data = pd.concat([data, train_tfidf], axis=1) #data.columns #['SMILES', 'ACTIVE', 'SMILES_str', 'NumAtoms', 'ExactMolWt', 'fr_Al_COO','HsNumAtoms', 'double', 'single', 'aromatic', 'triple'] traindata = data[[ 'NumAtoms', 'ExactMolWt', 'fr_Al_COO', 'HsNumAtoms', 'double', 'single', 'aromatic', 'triple' ]] finger = [ np.array(AllChem.GetMorganFingerprintAsBitVect(x, 2, nBits=512)) for x in data["SMILES"] ] finger = pd.DataFrame(finger) finger.columns = ["morgan_" + str(x) for x in finger.columns] model = word2vec.Word2Vec.load('models/model_300dim.pkl') data['sentence'] = data.apply( lambda x: MolSentence(mol2alt_sentence(x['SMILES'], 1)), axis=1) m2v = [ DfVec(x) for x in sentences2vec(data['sentence'], model, unseen='UNK') ] m2v = np.array([x.vec for x in m2v]) m2v = pd.DataFrame(m2v) m2v.columns = ["m2v_" + str(x) for x in m2v.columns] datadict = { "Morgan": finger, "Despcritor": traindata, "molvec": m2v, 'y': y } return datadict