import rdkit from rdkit import Chem from rdkit.Chem import AllChem from rdkit.Chem import PandasTools from rdkit.Chem import Descriptors import pandas as pd import numpy as np import molvs import sys import os import subprocess import tempfile import fnmatch import xlrd import sqlite3 from molvs.tautomer import TautomerTransform from rdkit import RDLogger from progress.bar import Bar from time import sleep from time import process_time TAUTOMER_TRANSFORMS = ( TautomerTransform('1,3 (thio)keto/enol f', '[CX4!H0]-[C]=[O,S,Se,Te;X1]'), TautomerTransform('1,3 (thio)keto/enol r', '[O,S,Se,Te;X2!H0]-[C]=[C]'), TautomerTransform('1,5 (thio)keto/enol f', '[CX4,NX3;!H0]-[C]=[C][CH0]=[O,S,Se,Te;X1]'), TautomerTransform('1,5 (thio)keto/enol r', '[O,S,Se,Te;X2!H0]-[CH0]=[C]-[C]=[C,N]'), TautomerTransform('aliphatic imine f', '[CX4!H0]-[C]=[NX2]'), TautomerTransform('aliphatic imine r', '[NX3!H0]-[C]=[CX3]'), TautomerTransform('special imine f', '[N!H0]-[C]=[CX3R0]'), TautomerTransform('special imine r', '[CX4!H0]-[c]=[n]'), TautomerTransform('1,3 aromatic heteroatom H shift f', '[#7!H0]-[#6R1]=[O,#7X2]'), TautomerTransform('1,3 aromatic heteroatom H shift r', '[O,#7;!H0]-[#6R1]=[#7X2]'), TautomerTransform('1,3 heteroatom H shift', '[#7,S,O,Se,Te;!H0]-[#7X2,#6,#15]=[#7,#16,#8,Se,Te]'), TautomerTransform('1,5 aromatic heteroatom H shift', '[#7,#16,#8;!H0]-[#6,#7]=[#6]-[#6,#7]=[#7,#16,#8;H0]'), TautomerTransform('1,5 aromatic heteroatom H shift f', '[#7,#16,#8,Se,Te;!H0]-[#6,nX2]=[#6,nX2]-[#6,#7X2]=[#7X2,S,O,Se,Te]'), TautomerTransform('1,5 aromatic heteroatom H shift r', '[#7,S,O,Se,Te;!H0]-[#6,#7X2]=[#6,nX2]-[#6,nX2]=[#7,#16,#8,Se,Te]'), TautomerTransform('1,7 aromatic heteroatom H shift f', '[#7,#8,#16,Se,Te;!H0]-[#6,#7X2]=[#6,#7X2]-[#6,#7X2]=[#6]-[#6,#7X2]=[#7X2,S,O,Se,Te,CX3]'), TautomerTransform('1,7 aromatic heteroatom H shift r', '[#7,S,O,Se,Te,CX4;!H0]-[#6,#7X2]=[#6]-[#6,#7X2]=[#6,#7X2]-[#6,#7X2]=[NX2,S,O,Se,Te]'), TautomerTransform('1,9 aromatic heteroatom H shift f', '[#7,O;!H0]-[#6,#7X2]=[#6,#7X2]-[#6,#7X2]=[#6,#7X2]-[#6,#7X2]=[#6,#7X2]-[#6,#7X2]=[#7,O]'), TautomerTransform('1,11 aromatic heteroatom H shift f', '[#7,O;!H0]-[#6,nX2]=[#6,nX2]-[#6,nX2]=[#6,nX2]-[#6,nX2]=[#6,nX2]-[#6,nX2]=[#6,nX2]-[#6,nX2]=[#7X2,O]'), TautomerTransform('furanone f', '[O,S,N;!H0]-[#6r5]=[#6X3r5;$([#6]([#6r5])=[#6r5])]'), TautomerTransform('furanone r', '[#6r5!H0;$([#6]([#6r5])[#6r5])]-[#6r5]=[O,S,N]'), TautomerTransform('keten/ynol f', '[C!H0]=[C]=[O,S,Se,Te;X1]', bonds='#-'), TautomerTransform('keten/ynol r', '[O,S,Se,Te;!H0X2]-[C]#[C]', bonds='=='), TautomerTransform('ionic nitro/aci-nitro f', '[C!H0]-[N+;$([N][O-])]=[O]'), TautomerTransform('ionic nitro/aci-nitro r', '[O!H0]-[N+;$([N][O-])]=[C]'), TautomerTransform('oxim/nitroso f', '[O!H0]-[N]=[C]'), TautomerTransform('oxim/nitroso r', '[C!H0]-[N]=[O]'), TautomerTransform('oxim/nitroso via phenol f', '[O!H0]-[N]=[C]-[C]=[C]-[C]=[OH0]'), TautomerTransform('oxim/nitroso via phenol r', '[O!H0]-[c]=[c]-[c]=[c]-[N]=[OH0]'), TautomerTransform('cyano/iso-cyanic acid f', '[O!H0]-[C]#[N]', bonds='=='), TautomerTransform('cyano/iso-cyanic acid r', '[N!H0]=[C]=[O]', bonds='#-'), # TautomerTransform('formamidinesulfinic acid f', '[O,N;!H0]-[C]=[S,Se,Te]=[O]', bonds='=--'), # TODO: WAT!? # TautomerTransform('formamidinesulfinic acid r', '[O!H0]-[S,Se,Te]-[C]=[O,N]', bonds='=--'), TautomerTransform('isocyanide f', '[C-0!H0]#[N+0]', bonds='#', charges='-+'), TautomerTransform('isocyanide r', '[N+!H0]#[C-]', bonds='#', charges='-+'), TautomerTransform('phosphinic acid f', '[OH]-[PD3X3H0]', bonds='='), TautomerTransform('phosphinic acid r', '[PD3X3H1]=[O]', bonds='-'), ) s = molvs.Standardizer(tautomer_transforms=TAUTOMER_TRANSFORMS, prefer_organic=True) HYD_RXNS = dict({ 'hyd_Xali_SN_A':Chem.AllChem.ReactionFromSmarts('[$([CX4]);$(C[Cl,Br,I]);!$(C([F,Cl,Br,I])([F,Cl,Br,I])[F,Cl,Br,I]);!$(C([F,Cl,Br,I])[F,Cl,Br,I]):1][Cl,Br,I:2]>>[*:1][OH1].[*:2]'), 'hyd_Xali_SN_B':Chem.AllChem.ReactionFromSmarts('[$([CX4]);$(C[Cl,Br,I]);$(CC[Cl,Br,I]);!$(C([F,Cl,Br,I])([F,Cl,Br,I])[F,Cl,Br,I]);!$(C([F,Cl,Br,I])[F,Cl,Br,I]):1][Cl,Br,I:2]>>[*:1][OH1].[*:2]'), 'hyd_Xali_SN_C':Chem.AllChem.ReactionFromSmarts('[Cl,Br,I:5]-[CX4:1]([F,Cl,Br,I:3])([*:4])([*:2])>>[*:5].[OH1][A:1]([*:3])([*:4])([*:2])'), 'hyd_Xali_E':Chem.AllChem.ReactionFromSmarts('[F,Cl,Br,I:3]-[CX4:1]-[C;X4;H1,H2:2]-[#1:4]>>[C:1]=[C:2].[*:4][*:3]'), 'hyd_epox':Chem.AllChem.ReactionFromSmarts('([OX2r3]1[#6r3:1][#6r3:2]1)>>([OX2r3][*:1][*:2][OH1])'), 'hyd_OPE_1':Chem.AllChem.ReactionFromSmarts('[PX4D4:1](=[!#6,O,S:2])([N,O,S;H0:3])([N,O,S;H0:4])[N,O,S;H0:5]>>[*:1]([OH1])(=[*:2])([*:3])[*:4].[H][*:5]'), 'hyd_ester':Chem.AllChem.ReactionFromSmarts('[CX3;$([#6]),$([H1]):1](=[OX1:4])[OX2:2][#6;!$(C=[O,N,S]):3]>>[*:1](=[*:4])[OH1].[*H1:2][*:3]'), 'hyd_lactone':Chem.AllChem.ReactionFromSmarts('([C][CX3R:1](=[OX1])[#8X2:2][C;!$(C=[O,N,S])])>>([*:1][OH1].[H][*:2])'), 'hyd_carbonate':Chem.AllChem.ReactionFromSmarts('[#6;!$(C=[O,N,S])][#8X2:1][#6X3:2](=[OX1])[#8X2:3][#6;!$(C=[O,N,S])]>>[*H1:1].[*:2].[H][*:3]'), 'hyd_anhydride':Chem.AllChem.ReactionFromSmarts('[O:1]([C](=O)[C])([C:2](=O)[C])>>[OH1:1]([C](=O)[C]).[OH1][C:2](=O)[C]'), 'hyd_cyanhydride':Chem.AllChem.ReactionFromSmarts('([$([CR]=O),$([CR]=[OX1]):1][#8X2R:2][$([CR]=O),$([CR]=[OX1]):3])>>([*:1][OH1].[OH1][*:3]).[*:2]'), 'hyd_amide':Chem.AllChem.ReactionFromSmarts('[CX3;$([R0][#6]),$([H1R0]):1](=[OX1:3])[#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])]):2]>>[OH1][*:1](=[*:3]).[H][*:2]'), 'hyd_lactam':Chem.AllChem.ReactionFromSmarts('([CR:4][CX3R:1](=[OX1])[#7X3;$([H1][C;!$(C=[O,N,S])]),$([H0]([C;!$(C=[O,N,S])])[C;!$(C=[O,N,S])]):2][CR:3])>>([*:4][*:1](=O)[OH1].[*:2][*:3])'), 'hyd_carbamate':Chem.AllChem.ReactionFromSmarts('[$([NH2]),$([NH][c,CX4]),$(N([c,CX4])[c,CX4]):1][C:2](=O)[O:3][$([#6]);!$(C=[O,S,N])]>>[H][*:1].[*:2].[H][*:3]'), 'hyd_thiocarbamate':Chem.AllChem.ReactionFromSmarts('[$([NH2]),$([NH][c,CX4]),$(N([c,CX4])[c,CX4]):1][C:2](=O)[SX2:3][$([#6]);!$(C=[O,S,N])]>>[H][*:1].[*:2].[H][*:3]'), 'hyd_urea':Chem.AllChem.ReactionFromSmarts('[#7X3;!$([#7][!#6]):1][CX3:2](=[OX1])[#7X3;!$([#7][!#6]):3]>>[H][*:1].[*:2].[H][*:3]'), 'hyd_sulfonylurea':Chem.AllChem.ReactionFromSmarts('[SX4;$([H1]),$([H0][#6])](=[!#6])(=[!#6])[#7X3;!$([#7][!#6]):1][#6X3:2](=[OX1])[#7X3;!$([#7][!#6]):3]>>[H][*:1].[*:2].[H][*:3]'), 'hyd_nitrile':Chem.AllChem.ReactionFromSmarts('[NX1:1]#[CX2:2]>>[H][*:1]([H])[*:2](=O)'), 'hyd_NS':Chem.AllChem.ReactionFromSmarts('[N:1][SX2:2][C,N,O:3]>>[H][*:1].[OH1][*:2][*:3]'), 'hyd_imdide':Chem.AllChem.ReactionFromSmarts('([#6X3;$([H0][#6]),$([H1]):1](=[OX1])[#7X3:2]([H,C,N])[#6X3;$([H0][#6]),$([H1])](=[OX1]))>>([*:1][OH1].[H][*:2])'), 'hyd_acidhalide':Chem.AllChem.ReactionFromSmarts('[$([S,C]=[O,S]):1][F,Br,Cl,I:2]>>[*:1][OH1].[*:2]'), 'hyd_gemdiol':Chem.AllChem.ReactionFromSmarts('[OX2H1:1][CX4;!$(C([OX2H])([OX2H])[O,S,#7,#15]):2][OX2H1:3]>>[*X1H0:1]=[*:2].[*:3]') }) def hydrolyze(mol = None,nsteps = 1,hyd_rxns = None): """ Hydrolyze a molecule. Subject an input molecule to hydrolyze. Multistep reactions (nsteps>1) hydrolyze products until the specified number of reaction steps is achieved. Parameters ---------- mol: rdkit.Chem.rdchem.Mol An rdkit mol object. nsteps: int The number of reaction steps to apply. filename: str A filename where hydrolysis reactions are stored as SMARTS. Returns ------- rdkit.Chem.rdchem.Mol A list of rdkit mol objects. """ if not isinstance(mol, Chem.rdchem.Mol): sys.exit('Input mol is not of type rdkit.Chem.rdchem.Mol') if mol is None: sys.exit('Missing input molecule.') if not isinstance(hyd_rxns, dict): sys.exit('hyd_rxns should be a dictionary') if not np.all([isinstance(hyd_rxns[key],rdkit.Chem.rdChemReactions.ChemicalReaction) for key in list(hyd_rxns.keys())]): sys.exit('Each element of hyd_rxns should be type rdkit.Chem.rdChemReaction.ChemicalReaction') n = 1 products = {} while n <= nsteps: if n == 1: mols = [mol] else: if len(products)>0: mols = list(products.values()) else: break for m in mols: prods = {k:list(np.concatenate(v)) for k,v in {k:hyd_rxns[k].RunReactants((m,)) for k in hyd_rxns}.items() if len(v)>0} for rxn,molprod in prods.items(): for mp in molprod: mp = s.super_parent(mp) inchikey = Chem.MolToInchiKey(mp) mp.SetProp('Reaction', '\t'.join(map(str,[Chem.MolToInchiKey(m),inchikey,rxn, n]))) if inchikey in [key for key in products]: products[inchikey] = MergeMolProps(products[inchikey],mp,True) else: products[inchikey] = mp n+=1 return(products) def biotransform(sd_in=None, sd_out=None, nsteps = 1,jar_file=None): """ Predict biotransformations. Use the biotransformer algorithm to predict environmental microbial biotransformation products. """ sdfFile = open(sd_out, 'a') sdfFile_w = Chem.SDWriter(sdfFile) supp = Chem.SDMolSupplier(sd_in) n = 5000 x = [range(len(supp))[i * n:(i + 1) * n] for i in range((len(range(len(supp))) + n - 1) // n )] for i in x: tmp_sdf_in = tempfile.NamedTemporaryFile("w+b",suffix=".sdf",delete=True).name tmp_sdf_out = tempfile.NamedTemporaryFile("w+b",suffix=".sdf",delete=True).name w = Chem.SDWriter(tmp_sdf_in) [w.write(supp[mol]) for mol in i] w.close() os.system('java -jar '+jar_file+' -b env -isdf '+tmp_sdf_in+' -k pred -osdf '+tmp_sdf_out+' -s '+str(nsteps)) supp2 = Chem.SDMolSupplier(tmp_sdf_out) for m in supp2: if m is not None: try: sdfFile_w.write(m) except ValueError: continue else: continue print('Total metabolites stored:', sdfFile_w.NumMols()) sdfFile_w.close() sdfFile.close() def MergeMolProps(mol1,mol2,inchikeymatch = True): ''' Merge properties of mols into one new mol. Example ------- mol1 = Chem.MolFromSmiles('c1ccccc1') mol2 = Chem.MolFromSmiles('c1ccccc1') mol1.SetProp('name', 'benzene') mol2.SetProp('name', 'cyclohexatriene') mol1.SetProp('id', 'abcd') mol2.SetProp('ID', 'xyz') mol3 = MergeMolProps(mol1, mol2) mol3.GetPropsAsDict() ''' if inchikeymatch and not Chem.MolToInchiKey(mol1)[:14] == Chem.MolToInchiKey(mol2)[:14]: sys.exit('InChIKeys do not match') else: if not mol1.GetPropsAsDict() == mol2.GetPropsAsDict(): props = mergeDict(mol1.GetPropsAsDict(),mol2.GetPropsAsDict(),True) for prop in props: mol1.SetProp(prop, str(props[prop])) return mol1 def mergeDict(dict1, dict2,join=True): ''' Merge dictionaries and keep values of common keys in list''' dict3 = {**dict1, **dict2} v_types = {k:type(v) for k,v in dict3.items()} for key, value in dict3.items(): if key in dict1 and key in dict2: if join: dict3[key] = ",".join(set([str(value) , str(dict1[key])])) else: dict3[key] = [str(value), str(dict1[key])] return dict3 def make_pfas_masslist(mollist_path, biotrans_jarfile, hyd_sdf, biotrans_in_sdf, biotrans_out_sdf, result_sdf, result_db): """ Make a PFAS molecular databse Take input molecule lists, apply hydrolysis and biotransformation, calculate properties, write results to sdf and sql database. Parameters ---------- mollist_path: path to a directory containing excel sheets exported from EPA Chemicals Dasboard biotrans_jarfile: Full path to a the biotransformer jar file hyd_sdf: A file name for the sdf file that will store results of hydrolysis prediction. biotrans_in_sdf: A filename for the input sdf file for biotransformation prediction. biotrans_out_sdf: A filename for the outpur sdf file from biotransformation. result_sdf: A file name for writing the resultant databsae to sdf file result_db: A database filename for storing the results as a SQLite database. """ # define substructure patterns carbon = Chem.MolFromSmarts('[#6]') CF2 = Chem.MolFromSmarts('[#6](F)F') fluorine = Chem.MolFromSmarts('[F]') # get a list of input files xls_files = {} for file in os.listdir(mollist_path): if fnmatch.fnmatch(file, '*.xls'): key = str.replace(file, '.xls', '') xls_files[key] = mollist_path+file mols = [] for key in xls_files: # read the mol list xls = xlrd.open_workbook(xls_files[key],encoding_override="gb2312") df = pd.read_excel(xls,na_values='-') # find the column with InChI string inchi_col = df.columns[df.columns.to_series().str.contains('INCHI_STRING|InChI')] df = df.dropna(subset=inchi_col).reset_index() for inchi in df[inchi_col[0]]: mol = Chem.MolFromInchi(inchi, sanitize=False) # make mols from inchi # split any non-covalent bonds frags = Chem.GetMolFrags(mol, asMols = True, sanitizeFrags = False) for f in frags: # get fragment, charge, isotope, stereochemistry and tautomer parent try: f_san = s.super_parent(f) except ValueError: continue # keep carbon and CF2 containing molecules if f_san.HasSubstructMatch(carbon) and f_san.HasSubstructMatch(CF2): f_san.SetProp('List', key) # set the input list f_san.SetProp('InChIKey', Chem.MolToInchiKey(f_san)) mols.append(f_san) # append to mols list #make mols a dictionary keyed by inchikey mols_dict = {} for mol in mols: inchikey = mol.GetProp('InChIKey')[:14] if inchikey in [key for key in mols_dict]: mols_dict[inchikey] = MergeMolProps(mols_dict[inchikey], mol) else: mols_dict[inchikey] = mol # see which molecules have hits for hydrolyzable moieties hyd_substruc = pd.DataFrame({}, columns = [key for key in HYD_RXNS]) for inchikey in mols_dict: df = pd.DataFrame({key:mols_dict[inchikey].HasSubstructMatch(HYD_RXNS[key].GetReactantTemplate(0)) for key in HYD_RXNS}, index = [inchikey]) hyd_substruc = hyd_substruc.append(df) hydrolyzable_inchikeys = hyd_substruc[hyd_substruc.any(axis = 'columns')].index.tolist() #predict hydrolysis products for each mol w = Chem.SDWriter(hyd_sdf) for inchikey in hydrolyzable_inchikeys: [w.write(v) for v in hydrolyze(mol=mols_dict[inchikey], nsteps=2, hyd_rxns=HYD_RXNS).values()] w.close() mols_rxn_hyd = pd.DataFrame({}, columns=['Precursor_InChIKey', 'Product_InChIKey', 'Reaction', 'RxnStep']) supp = Chem.SDMolSupplier(hyd_sdf) for mol in supp: if mol is None: continue if not mol.HasSubstructMatch(Chem.MolFromSmarts('[#6][F]')): continue else: rxns = mol.GetProp('Reaction').split(',') rxn = [r.split('\t') for r in rxns] rxn_entry = pd.DataFrame({ 'Precursor_InChIKey':[r[0] for r in rxn], 'Product_InChIKey':[r[1] for r in rxn], 'Reaction':[r[2] for r in rxn], 'RxnStep':[r[3] for r in rxn] }) mol.SetProp('List', 'HYD') inchikey = Chem.MolToInchiKey(mol) mol.SetProp('InChIKey', inchikey) mols_rxn_hyd = mols_rxn_hyd.append(rxn_entry) if inchikey[:14] in [key for key in mols_dict]: mols_dict[inchikey[:14]] = MergeMolProps(mols_dict[inchikey[:14]],mol) else: mols_dict[inchikey[:14]] = mol # predict biotrans products w = Chem.SDWriter(biotrans_in_sdf) [w.write(mol) for mol in mols_dict.values()] w.close() biotransform(sd_in=biotrans_in_sdf, sd_out=biotrans_out_sdf, nsteps = 2,jar_file=biotrans_jarfile) # add biotransformation products to list mols_rxn_bio = [] supp = Chem.SDMolSupplier(biotrans_out_sdf) for mol in supp: if not mol.HasSubstructMatch(Chem.MolFromSmarts('[#6][F]')): continue mol = MergeMolProps(s.super_parent(mol), mol,False) inchikey = Chem.MolToInchiKey(mol) inchikey_parent = mol.GetProp('Precursor_InChIKey') rxn = mol.GetProp('Reaction') for key in mol.GetPropsAsDict(): if key not in ['Reaction','Precursor_InChIKey','InChIKey']: mol.ClearProp(key) mol.SetProp('InChIKey', inchikey) mol.SetProp('List', 'BIOTRANS') mol.SetProp('BioTrans_Reaction', rxn) mol.SetProp('Reaction', '\t'.join(map(str,[inchikey_parent,inchikey,rxn]))) mols_rxn_bio.append(mol.GetPropsAsDict()) if inchikey[:14] in [key for key in mols_dict]: mols_dict[inchikey[:14]] = MergeMolProps(mols_dict[inchikey[:14]],mol,True) else: mols_dict[inchikey[:14]] = mol mols_rxn_bio = pd.DataFrame(mols_rxn_bio) # clean up mol props for inchikey in mols_dict: props = mols_dict[inchikey].GetPropsAsDict() for prop in props: mols_dict[inchikey].SetProp(prop, ",".join(set(str.split(props[prop],',')))) # keep only one inchikey per entry for inchikey in mols_dict: x = mols_dict[inchikey].GetProp('InChIKey') mols_dict[inchikey].SetProp('InChIKey_syn',x) mols_dict[inchikey].SetProp('InChIKey', x.split(',')[0]) mol_props = [] for inchikey in mols_dict: m = mols_dict[inchikey] mol_props.append({'InChIKey':inchikey,'MolForm':Chem.rdMolDescriptors.CalcMolFormula(m),'ExactMass':Chem.rdMolDescriptors.CalcExactMolWt(m),'FormalCharge':Chem.GetFormalCharge(m),'HBD':Chem.rdMolDescriptors.CalcNumHBD(m),'HBA':Chem.rdMolDescriptors.CalcNumHBA(m)}) mol_props = pd.DataFrame(mol_props) # write molecules table mols_tab = [] list_tab = [] for inchikey in mols_dict: mols_tab.append({'ID': inchikey, 'InChIKey':Chem.MolToInchiKey(mols_dict[inchikey]), 'InChI':Chem.MolToInchi(mols_dict[inchikey]), 'SMILES':Chem.MolToSmiles(mols_dict[inchikey]), 'MolStructure':Chem.MolToMolBlock(mols_dict[inchikey])}) lists = mols_dict[inchikey].GetProp('List').split(',') list_tab.append({ 'InChIKey': [Chem.MolToInchiKey(mols_dict[inchikey]) for l in lists], 'List': lists }) mols_tab_df = pd.DataFrame(mols_tab) list_tab_df = pd.DataFrame({}, columns = ['InChIKey','List']) for i in list_tab: list_tab_df = list_tab_df.append(pd.DataFrame(i)) # write sdf w = Chem.SDWriter(result_sdf) [w.write(mol) for mol in mols_dict.values()] w.close() # make a database conn = sqlite3.connect(result_db) mols_tab_df.set_index(['ID','InChIKey']).to_sql("molecules", conn, if_exists='replace') mol_props.set_index('InChIKey').to_sql("mol_props", conn, if_exists='replace') list_tab_df.set_index('InChIKey').to_sql("list_membership", conn, if_exists='replace') mols_rxn_hyd.set_index(['Precursor_InChIKey','Product_InChIKey']).to_sql("hyd_rxns", conn, if_exists='replace') mols_rxn_bio.set_index(['Precursor_InChIKey','InChIKey']).to_sql("bio_rxns", conn, if_exists='replace') conn.close()