def eval_function(text):
generated = ''.join(text)
try:
decoded = DeepSMILESLanguageModelUtils.decode(generated,
start='<s>',
end='</s>')
DeepSMILESLanguageModelUtils.sanitize(decoded)
except Exception:
return 0
# extracted = DeepSMILESLanguageModelUtils.extract(generated)
# tokenized = DeepSMILESTokenizer(extracted)
# len_score = len(tokenized.get_tokens()) / (text_length - 1) # provide more reward for longer text sequences
decoded = DeepSMILESLanguageModelUtils.decode(generated,
start='<s>',
end='</s>')
smiles = DeepSMILESLanguageModelUtils.sanitize(decoded)
mol = Chem.MolFromSmiles(smiles)
logp = factor * MolLogP(mol)
logp_score = (logp - logp_min) / (logp_max - logp_min
) # normalize logP between 0 and 1
score = logp_score # (logp_score * 0.5) + (len_score * 0.5)
logger.info("%s, %s" % (generated, str(score)))
return score
def GetNeighborLists(probes,topN,pool,
simMetric=DataStructs.DiceSimilarity,
silent=False):
probeFps = [x[1] for x in probes]
validProbes = [x for x in range(len(probeFps)) if probeFps[x] is not None]
validFps=[probeFps[x] for x in validProbes]
from rdkit.DataStructs.TopNContainer import TopNContainer
nbrLists = [TopNContainer(topN) for x in range(len(probeFps))]
nDone=0
for nm,fp in pool:
nDone+=1
if not silent and not nDone%1000: logger.info(' searched %d rows'%nDone)
if(simMetric==DataStructs.DiceSimilarity):
scores = DataStructs.BulkDiceSimilarity(fp,validFps)
for i,score in enumerate(scores):
nbrLists[validProbes[i]].Insert(score,nm)
elif(simMetric==DataStructs.TanimotoSimilarity):
scores = DataStructs.BulkTanimotoSimilarity(fp,validFps)
for i,score in enumerate(scores):
nbrLists[validProbes[i]].Insert(score,nm)
else:
for i in range(len(probeFps)):
pfp = probeFps[i]
if pfp is not None:
score = simMetric(probeFps[i],fp)
nbrLists[i].Insert(score,nm)
return nbrLists
def dividetask(data, task, silent=True):
data = mpi.broadcast(mpi.world, data, 0)
nProcs = mpi.world.size
chunkSize = len(data) // nProcs
extraBits = len(data) % nProcs
res = []
allRes = []
# the root node handles the extra pieces:
if mpi.world.rank == 0:
for i in range(extraBits):
elem = data[i]
res.append(task(elem))
if not silent:
logger.info('task(%d) done %d' % (mpi.world.rank, i + 1))
pos = extraBits + mpi.world.rank * chunkSize
for i in range(chunkSize):
elem = data[pos]
pos += 1
res.append(task(elem))
if not silent:
logger.info('task(%d) done %d' % (mpi.world.rank, i + 1))
if mpi.world.rank == 0:
tmp = mpi.gather(mpi.world, res, 0)
for res in tmp:
allRes.extend(res)
else:
mpi.gather(mpi.world, res, 0)
return allRes
def eval_function(text):
generated = ''.join(text)
try:
decoded = DeepSMILESLanguageModelUtils.decode(generated,
start='<s>',
end='</s>')
DeepSMILESLanguageModelUtils.sanitize(decoded)
except Exception:
return 0
decoded = DeepSMILESLanguageModelUtils.decode(generated,
start='<s>',
end='</s>')
smiles = DeepSMILESLanguageModelUtils.sanitize(decoded)
mol = Chem.MolFromSmiles(smiles)
num_atoms = mol.GetNumAtoms()
num_aromatic_atoms = 0
for i in range(num_atoms):
if mol.GetAtomWithIdx(i).GetIsAromatic():
num_aromatic_atoms += 1
arom_reward = num_aromatic_atoms / 23
perplexity = lm.perplexity(text)
perplexity_reward = perplexity / (1 + perplexity)
score = (perplexity_reward * 0.5) + (arom_reward * 0.5)
logger.info("%s, %s" % (generated, str(score)))
return score
def dividetask(data,task,silent=True):
data=mpi.broadcast(mpi.world,data,0)
nProcs = mpi.world.size
chunkSize=len(data)//nProcs
extraBits =len(data)%nProcs
res=[]
allRes=[]
# the root node handles the extra pieces:
if mpi.world.rank == 0:
for i in range(extraBits):
elem=data[i]
res.append(task(elem))
if not silent:
logger.info('task(%d) done %d'%(mpi.world.rank,i+1))
pos=extraBits+mpi.world.rank*chunkSize;
for i in range(chunkSize):
elem=data[pos]
pos += 1
res.append(task(elem))
if not silent:
logger.info('task(%d) done %d'%(mpi.world.rank,i+1))
if mpi.world.rank==0:
tmp=mpi.gather(mpi.world,res,0)
for res in tmp: allRes.extend(res)
else:
mpi.gather(mpi.world,res,0)
return allRes
def eval_function(text):
generated = ''.join(text)
try:
decoded = DeepSMILESLanguageModelUtils.decode(generated,
start='<s>',
end='</s>')
smiles = DeepSMILESLanguageModelUtils.sanitize(decoded)
except Exception:
return -1.0
jscore = jscorer.score(smiles)
score = jscore / (1 + np.abs(jscore))
logger.info("%s, %s" % (generated, str(score)))
return score
def GetNeighborLists(probes,
topN,
pool,
simMetric=DataStructs.DiceSimilarity,
simThresh=-1.,
silent=False,
**kwargs):
probeFps = [x[1] for x in probes]
validProbes = [x for x in range(len(probeFps)) if probeFps[x] is not None]
validFps = [probeFps[x] for x in validProbes]
from rdkit.DataStructs.TopNContainer import TopNContainer
if simThresh <= 0:
nbrLists = [TopNContainer(topN) for x in range(len(probeFps))]
else:
nbrLists = [TopNContainer(-1) for x in range(len(probeFps))]
nDone = 0
for nm, fp in pool:
nDone += 1
if not silent and not nDone % 1000:
logger.info(' searched %d rows' % nDone)
if (simMetric == DataStructs.DiceSimilarity):
scores = DataStructs.BulkDiceSimilarity(fp, validFps)
for i, score in enumerate(scores):
if score > simThresh:
nbrLists[validProbes[i]].Insert(score, nm)
elif (simMetric == DataStructs.TanimotoSimilarity):
scores = DataStructs.BulkTanimotoSimilarity(fp, validFps)
for i, score in enumerate(scores):
if score > simThresh:
nbrLists[validProbes[i]].Insert(score, nm)
elif (simMetric == DataStructs.TverskySimilarity):
av = float(kwargs.get('tverskyA', 0.5))
bv = float(kwargs.get('tverskyB', 0.5))
scores = DataStructs.BulkTverskySimilarity(fp, validFps, av, bv)
for i, score in enumerate(scores):
if score > simThresh:
nbrLists[validProbes[i]].Insert(score, nm)
else:
for i in range(len(probeFps)):
pfp = probeFps[i]
if pfp is not None:
score = simMetric(probeFps[i], fp)
if score > simThresh:
nbrLists[validProbes[i]].Insert(score, nm)
return nbrLists
def eval_function(text):
generated = ''.join(text)
try:
decoded = DeepSMILESLanguageModelUtils.decode(generated,
start='<s>',
end='</s>')
DeepSMILESLanguageModelUtils.sanitize(decoded)
except Exception:
return 0
extracted = DeepSMILESLanguageModelUtils.extract(generated,
start='<s>',
end='</s>')
tokenized = DeepSMILESTokenizer(extracted)
score = len(tokenized.get_tokens()) / (
text_length - 1) # provide more reward for longer text sequences
logger.info("%s, %s" % (generated, str(score)))
return score
def testCairoFile(self):
try:
from rdkit.Chem.Draw.cairoCanvas import Canvas
except ImportError:
logger.info("Skipping cairo test")
return
os.environ['RDKIT_CANVAS']='cairo'
foo,fn=tempfile.mkstemp(suffix='.png')
foo=None
self.failUnlessEqual(os.path.getsize(fn),0)
Draw.MolToFile(self.mol,fn)
self.failIfEqual(os.path.getsize(fn),0)
try:
os.unlink(fn)
except:
pass
def testSpingFile(self):
try:
from rdkit.Chem.Draw.spingCanvas import Canvas
except ImportError:
logger.info("Skipping sping test")
return
os.environ['RDKIT_CANVAS'] = 'sping'
foo, fn = tempfile.mkstemp(suffix='.png')
foo = None
self.assertEqual(os.path.getsize(fn), 0)
Draw.MolToFile(self.mol, fn)
self.assertNotEqual(os.path.getsize(fn), 0)
try:
os.unlink(fn)
except Exception:
pass
def GetNeighborLists(probes,topN,pool,
simMetric=DataStructs.DiceSimilarity,
simThresh=-1.,
silent=False,
**kwargs):
probeFps = [x[1] for x in probes]
validProbes = [x for x in range(len(probeFps)) if probeFps[x] is not None]
validFps=[probeFps[x] for x in validProbes]
from rdkit.DataStructs.TopNContainer import TopNContainer
if simThresh<=0:
nbrLists = [TopNContainer(topN) for x in range(len(probeFps))]
else:
nbrLists=[TopNContainer(-1) for x in range(len(probeFps))]
nDone=0
for nm,fp in pool:
nDone+=1
if not silent and not nDone%1000: logger.info(' searched %d rows'%nDone)
if(simMetric==DataStructs.DiceSimilarity):
scores = DataStructs.BulkDiceSimilarity(fp,validFps)
for i,score in enumerate(scores):
if score>simThresh:
nbrLists[validProbes[i]].Insert(score,nm)
elif(simMetric==DataStructs.TanimotoSimilarity):
scores = DataStructs.BulkTanimotoSimilarity(fp,validFps)
for i,score in enumerate(scores):
if score>simThresh:
nbrLists[validProbes[i]].Insert(score,nm)
elif(simMetric==DataStructs.TverskySimilarity):
av = float(kwargs.get('tverskyA',0.5))
bv = float(kwargs.get('tverskyB',0.5))
scores = DataStructs.BulkTverskySimilarity(fp,validFps,av,bv)
for i,score in enumerate(scores):
if score>simThresh:
nbrLists[validProbes[i]].Insert(score,nm)
else:
for i in range(len(probeFps)):
pfp = probeFps[i]
if pfp is not None:
score = simMetric(probeFps[i],fp)
if score>simThresh:
nbrLists[validProbes[i]].Insert(score,nm)
return nbrLists
def testSpingFile(self):
try:
from rdkit.Chem.Draw.spingCanvas import Canvas
except ImportError:
logger.info("Skipping sping test")
return
os.environ['RDKIT_CANVAS']='sping'
foo,fn=tempfile.mkstemp(suffix='.png')
foo=None
self.assertEqual(os.path.getsize(fn),0)
Draw.MolToFile(self.mol,fn)
self.assertNotEqual(os.path.getsize(fn),0)
try:
os.unlink(fn)
except Exception:
pass
def testAggFile(self):
try:
from rdkit.Chem.Draw.aggCanvas import Canvas
except ImportError:
logger.info("Skipping agg test")
return
os.environ['RDKIT_CANVAS'] = 'agg'
foo, fn = tempfile.mkstemp(suffix='.png')
foo = None
self.failUnlessEqual(os.path.getsize(fn), 0)
Draw.MolToFile(self.mol, fn)
self.failIfEqual(os.path.getsize(fn), 0)
try:
os.unlink(fn)
except:
pass
def testSpingFile(self):
try:
from rdkit.Chem.Draw.spingCanvas import Canvas
except ImportError:
logger.info("Skipping sping test")
return
os.environ["RDKIT_CANVAS"] = "sping"
foo, fn = tempfile.mkstemp(suffix=".png")
foo = None
self.failUnlessEqual(os.path.getsize(fn), 0)
Draw.MolToFile(self.mol, fn)
self.failIfEqual(os.path.getsize(fn), 0)
try:
os.unlink(fn)
except:
pass
def eval_function(text):
generated = ''.join(text)
try:
decoded = DeepSMILESLanguageModelUtils.decode(generated,
start='<s>',
end='</s>')
smiles = DeepSMILESLanguageModelUtils.sanitize(decoded)
except Exception:
return -1.0
global all_smiles
if smiles in all_smiles:
score = -1.0
else:
qedscore = qedscorer.score(smiles)
score = qedscore / (1 + np.abs(qedscore))
all_smiles[smiles] = qedscore
logger.info("%s, %s" % (smiles, str(score)))
return score
parser.add_option('--maxPathLength','--max',default=8,type=int,
help='maximum length path for the fingerprint')
parser.add_option('--similarityThreshold','--sim',default=[0.9],type='floatlist',
help='threshold for similarity')
parser.add_option('--numNeighbors','--num','-n','-k',default=50,type=int,
help='number of neighbors to consider')
parser.add_option('--neighborsFile','--nbrs',default='',
help='name of an output file to hold the neighbor lists')
parser.add_option('--scan',default=False,action="store_true")
if __name__=='__main__':
options,args = parser.parse_args()
outF = file(args[-1],'w+')
logger.info('reading training molecules and generating fingerprints')
suppl = Chem.SDMolSupplier(args[0])
train=[]
for i,mol in enumerate(suppl):
if not mol:
continue
smi = Chem.MolToSmiles(mol,True)
nm = mol.GetProp(nameField)
property = float(mol.GetProp(propField))
fp = GetMolFingerprint(mol,options.maxPathLength)
train.append((nm,smi,fp,property))
logger.info(' got %d molecules'%len(train))
if len(args)>2:
suppl = Chem.SDMolSupplier(args[1])
haveTest=True
"benchmark and test fingerprint screenout and substructure searching")
parser.add_argument("--validate",
dest='validateResults',
default=False,
action='store_true',
help="validate that the screenout isn't missing anything")
parser.add_argument("--short",
dest='doShort',
default=False,
action='store_true',
help="run a small subset of the molecules")
args = parser.parse_args()
ts = []
logger.info('mols from smiles')
mols = []
t1 = time.time()
# find this file here: https://raw.githubusercontent.com/greglandrum/rdkit_blog/master/data/chembl21_25K.pairs.txt.gz
with gzip.open('../Data/chembl21_25K.pairs.txt.gz', 'rb') as inf:
for line in inf:
line = line.decode().strip().split()
smi1 = line[1]
smi2 = line[3]
mols.append(Chem.MolFromSmiles(smi1))
mols.append(Chem.MolFromSmiles(smi2))
if args.doShort and len(mols) >= 1000:
break
t2 = time.time()
ts.append(t2 - t1)
logger.info(f'Results{len(ts)}: {t2-t1 : .2f} seconds, {len(mols)} mols')
from rdkit.Chem import AllChem
from rdkit.Chem import Recap
from rdkit.RDLogger import logger
logger = logger()
tests=[1]*1001
if len(sys.argv)>1:
tests=[0]*1001
tests[1]=1
for x in sys.argv[1:]:
x = int(x)
tests[x] = 1
ts = []
sdData = gzip.open('../Data/mols.1000.sdf.gz','rb').read()
logger.info('mols from sdf')
suppl = Chem.SDMolSupplier()
suppl.SetData(sdData)
mols = []
nMols=0
nBad=0
t1=time.time()
for m in suppl:
if m:
nMols+=1
mols.append(m)
else:
nBad += 1
t2=time.time()
logger.info('Results1: %.2f seconds, %d passed, %d failed'%(t2-t1,nMols,nBad))
ts.append(t2-t1)
# propField is the name of the property (from the SD file) you want to use
# as the "activity"
propField='chemical_shift_1'
# similarity threshold for a pair to be considered interesting.
# (i.e. pairs with a similiarity below this value will not be
# added to the output.
similarityThreshold=0.5
if __name__=='__main__':
suppl = Chem.SDMolSupplier(sys.argv[1])
outF = file(sys.argv[2],'w+')
data=[]
logger.info('reading molecules and generating fingeprints')
for i,mol in enumerate(suppl):
if not mol:
continue
smi = Chem.MolToSmiles(mol,True)
nm = mol.GetProp(nameField)
property = float(mol.GetProp(propField))
fp = GetMolFingerprint(mol,maxPathLength)
data.append((nm,smi,property,fp))
logger.info(' got %d molecules'%len(data))
logger.info('calculating pairs')
pairs = []
for i in range(len(data)):
for j in range(i+1,len(data)):
def CreateDb(options,dataFilename='',supplier=None):
if not dataFilename and supplier is None:
raise ValueError('Please provide either a data filename or a supplier')
if options.errFilename:
errFile=open(os.path.join(options.outDir,options.errFilename),'w+')
else:
errFile=None
if options.noExtras:
options.doPairs=False
options.doDescriptors=False
options.doFingerprints=False
options.doPharm2D=False
options.doGobbi2D=False
options.doLayered=False
options.doMorganFps=False
if options.loadMols:
if supplier is None:
if not options.molFormat:
ext = os.path.splitext(dataFilename)[-1].lower()
if ext=='.sdf':
options.molFormat='sdf'
elif ext in ('.smi','.smiles','.txt','.csv'):
options.molFormat='smiles'
if not options.delimiter:
# guess the delimiter
import csv
sniffer = csv.Sniffer()
dlct=sniffer.sniff(open(dataFilename,'r').read(2000))
options.delimiter=dlct.delimiter
if not options.silent:
logger.info('Guessing that delimiter is %s. Use --delimiter argument if this is wrong.'%repr(options.delimiter))
if not options.silent:
logger.info('Guessing that mol format is %s. Use --molFormat argument if this is wrong.'%repr(options.molFormat))
if options.molFormat=='smiles':
if options.delimiter=='\\t': options.delimiter='\t'
supplier=Chem.SmilesMolSupplier(dataFilename,
titleLine=options.titleLine,
delimiter=options.delimiter,
smilesColumn=options.smilesColumn,
nameColumn=options.nameColumn
)
else:
supplier = Chem.SDMolSupplier(dataFilename)
if not options.silent: logger.info('Reading molecules and constructing molecular database.')
Loader.LoadDb(supplier,os.path.join(options.outDir,options.molDbName),
errorsTo=errFile,regName=options.regName,nameCol=options.molIdName,
skipProps=options.skipProps,defaultVal=options.missingPropertyVal,
addComputedProps=options.addProps,uniqNames=True,
skipSmiles=options.skipSmiles,maxRowsCached=int(options.maxRowsCached),
silent=options.silent,nameProp=options.nameProp,
lazySupplier=int(options.maxRowsCached)>0,
startAnew=not options.updateDb
)
if options.doPairs:
pairConn = DbConnect(os.path.join(options.outDir,options.pairDbName))
pairCurs = pairConn.GetCursor()
try:
pairCurs.execute('drop table %s'%(options.pairTableName))
except:
pass
pairCurs.execute('create table %s (guid integer not null primary key,%s varchar not null unique,atompairfp blob,torsionfp blob)'%(options.pairTableName,
options.molIdName))
if options.doFingerprints or options.doPharm2D or options.doGobbi2D or options.doLayered:
fpConn = DbConnect(os.path.join(options.outDir,options.fpDbName))
fpCurs=fpConn.GetCursor()
try:
fpCurs.execute('drop table %s'%(options.fpTableName))
except:
pass
try:
fpCurs.execute('drop table %s'%(options.pharm2DTableName))
except:
pass
try:
fpCurs.execute('drop table %s'%(options.gobbi2DTableName))
except:
pass
try:
fpCurs.execute('drop table %s'%(options.layeredTableName))
except:
pass
if options.doFingerprints:
fpCurs.execute('create table %s (guid integer not null primary key,%s varchar not null unique,rdkfp blob)'%(options.fpTableName,
options.molIdName))
if options.doLayered:
layeredQs = ','.join('?'*LayeredOptions.nWords)
colDefs=','.join(['Col_%d integer'%(x+1) for x in range(LayeredOptions.nWords)])
fpCurs.execute('create table %s (guid integer not null primary key,%s varchar not null unique,%s)'%(options.layeredTableName,
options.molIdName,
colDefs))
if options.doPharm2D:
fpCurs.execute('create table %s (guid integer not null primary key,%s varchar not null unique,pharm2dfp blob)'%(options.pharm2DTableName,
options.molIdName))
sigFactory = BuildSigFactory(options)
if options.doGobbi2D:
fpCurs.execute('create table %s (guid integer not null primary key,%s varchar not null unique,gobbi2dfp blob)'%(options.gobbi2DTableName,
options.molIdName))
from rdkit.Chem.Pharm2D import Generate,Gobbi_Pharm2D
if options.doMorganFps :
fpConn = DbConnect(os.path.join(options.outDir,options.fpDbName))
fpCurs=fpConn.GetCursor()
try:
fpCurs.execute('drop table %s'%(options.morganFpTableName))
except:
pass
fpCurs.execute('create table %s (guid integer not null primary key,%s varchar not null unique,morganfp blob)'%(options.morganFpTableName,
options.molIdName))
if options.doDescriptors:
descrConn=DbConnect(os.path.join(options.outDir,options.descrDbName))
with open(options.descriptorCalcFilename,'r') as inTF:
buf = inTF.read().replace('\r\n', '\n').encode('utf-8')
inTF.close()
calc = cPickle.load(io.BytesIO(buf))
nms = [x for x in calc.GetDescriptorNames()]
descrCurs = descrConn.GetCursor()
descrs = ['guid integer not null primary key','%s varchar not null unique'%options.molIdName]
descrs.extend(['%s float'%x for x in nms])
try:
descrCurs.execute('drop table %s'%(options.descrTableName))
except:
pass
descrCurs.execute('create table %s (%s)'%(options.descrTableName,','.join(descrs)))
descrQuery=','.join([DbModule.placeHolder]*len(descrs))
pairRows = []
fpRows = []
layeredRows = []
descrRows = []
pharm2DRows=[]
gobbi2DRows=[]
morganRows = []
if not options.silent: logger.info('Generating fingerprints and descriptors:')
molConn = DbConnect(os.path.join(options.outDir,options.molDbName))
molCurs = molConn.GetCursor()
if not options.skipSmiles:
molCurs.execute('select guid,%s,smiles,molpkl from %s'%(options.molIdName,options.regName))
else:
molCurs.execute('select guid,%s,molpkl from %s'%(options.molIdName,options.regName))
i=0
while 1:
try:
tpl = molCurs.fetchone()
molGuid = tpl[0]
molId = tpl[1]
pkl = tpl[-1]
i+=1
except:
break
if isinstance(pkl,(bytes,str)):
mol = Chem.Mol(pkl)
else:
mol = Chem.Mol(str(pkl))
if not mol: continue
if options.doPairs:
pairs = FingerprintUtils.BuildAtomPairFP(mol)
torsions = FingerprintUtils.BuildTorsionsFP(mol)
pkl1 = DbModule.binaryHolder(pairs.ToBinary())
pkl2 = DbModule.binaryHolder(torsions.ToBinary())
row = (molGuid,molId,pkl1,pkl2)
pairRows.append(row)
if options.doFingerprints:
fp2 = FingerprintUtils.BuildRDKitFP(mol)
pkl = DbModule.binaryHolder(fp2.ToBinary())
row = (molGuid,molId,pkl)
fpRows.append(row)
if options.doLayered:
words = LayeredOptions.GetWords(mol)
row = [molGuid,molId]+words
layeredRows.append(row)
if options.doDescriptors:
descrs= calc.CalcDescriptors(mol)
row = [molGuid,molId]
row.extend(descrs)
descrRows.append(row)
if options.doPharm2D:
FingerprintUtils.sigFactory=sigFactory
fp= FingerprintUtils.BuildPharm2DFP(mol)
pkl = DbModule.binaryHolder(fp.ToBinary())
row = (molGuid,molId,pkl)
pharm2DRows.append(row)
if options.doGobbi2D:
FingerprintUtils.sigFactory=Gobbi_Pharm2D.factory
fp= FingerprintUtils.BuildPharm2DFP(mol)
pkl = DbModule.binaryHolder(fp.ToBinary())
row = (molGuid,molId,pkl)
gobbi2DRows.append(row)
if options.doMorganFps:
morgan = FingerprintUtils.BuildMorganFP(mol)
pkl = DbModule.binaryHolder(morgan.ToBinary())
row = (molGuid,molId,pkl)
morganRows.append(row)
if not i%500:
if len(pairRows):
pairCurs.executemany('insert into %s values (?,?,?,?)'%options.pairTableName,
pairRows)
pairRows = []
pairConn.Commit()
if len(fpRows):
fpCurs.executemany('insert into %s values (?,?,?)'%options.fpTableName,
fpRows)
fpRows = []
fpConn.Commit()
if len(layeredRows):
fpCurs.executemany('insert into %s values (?,?,%s)'%(options.layeredTableName,layeredQs),
layeredRows)
layeredRows = []
fpConn.Commit()
if len(descrRows):
descrCurs.executemany('insert into %s values (%s)'%(options.descrTableName,descrQuery),
descrRows)
descrRows = []
descrConn.Commit()
if len(pharm2DRows):
fpCurs.executemany('insert into %s values (?,?,?)'%options.pharm2DTableName,
pharm2DRows)
pharm2DRows = []
fpConn.Commit()
if len(gobbi2DRows):
fpCurs.executemany('insert into %s values (?,?,?)'%options.gobbi2DTableName,
gobbi2DRows)
gobbi2DRows = []
fpConn.Commit()
if len(morganRows):
fpCurs.executemany('insert into %s values (?,?,?)'%options.morganFpTableName,
morganRows)
morganRows = []
fpConn.Commit()
if not options.silent and not i%500:
logger.info(' Done: %d'%(i))
if len(pairRows):
pairCurs.executemany('insert into %s values (?,?,?,?)'%options.pairTableName,
pairRows)
pairRows = []
pairConn.Commit()
if len(fpRows):
fpCurs.executemany('insert into %s values (?,?,?)'%options.fpTableName,
fpRows)
fpRows = []
fpConn.Commit()
if len(layeredRows):
fpCurs.executemany('insert into %s values (?,?,%s)'%(options.layeredTableName,layeredQs),
layeredRows)
layeredRows = []
fpConn.Commit()
if len(descrRows):
descrCurs.executemany('insert into %s values (%s)'%(options.descrTableName,descrQuery),
descrRows)
descrRows = []
descrConn.Commit()
if len(pharm2DRows):
fpCurs.executemany('insert into %s values (?,?,?)'%options.pharm2DTableName,
pharm2DRows)
pharm2DRows = []
fpConn.Commit()
if len(gobbi2DRows):
fpCurs.executemany('insert into %s values (?,?,?)'%options.gobbi2DTableName,
gobbi2DRows)
gobbi2DRows = []
fpConn.Commit()
if len(morganRows):
fpCurs.executemany('insert into %s values (?,?,?)'%options.morganFpTableName,
morganRows)
morganRows = []
fpConn.Commit()
if not options.silent:
logger.info('Finished.')
from rdkit.Chem import AllChem
from rdkit.Chem import Recap
from rdkit.RDLogger import logger
logger = logger()
tests = [1] * 1001
if len(sys.argv) > 1:
tests = [0] * 1001
tests[1] = 1
for x in sys.argv[1:]:
x = int(x)
tests[x] = 1
ts = []
sdData = gzip.open('../Data/mols.1000.sdf.gz', 'rb').read()
logger.info('mols from sdf')
suppl = Chem.SDMolSupplier()
suppl.SetData(sdData)
mols = []
nMols = 0
nBad = 0
t1 = time.time()
for m in suppl:
if m:
nMols += 1
mols.append(m)
else:
nBad += 1
t2 = time.time()
logger.info('Results1: %.2f seconds, %d passed, %d failed' %
(t2 - t1, nMols, nBad))
lm = EmptyDeepSMILESLanguageModel(vocab, n=6)
current_best_score = None
current_best_smiles = None
beats_current = lambda score: score < current_best_score
for i in range(1000):
generated = lm.generate(num_chars=25, text_seed="<s>")
try:
decoded = DeepSMILESLanguageModelUtils.decode(generated,
start='<s>',
end='</s>')
sanitized = DeepSMILESLanguageModelUtils.sanitize(decoded)
mol = Chem.MolFromSmiles(sanitized)
logp_score = MolLogP(mol)
logger.info("successful: %s , score: %s" %
(sanitized, str(logp_score)))
if current_best_score is None or beats_current(logp_score):
current_best_score = logp_score
current_best_smiles = sanitized
except Exception as e:
pass
logger.info("best: %s , score: %s" %
(current_best_smiles, str(current_best_score)))
from rdkit.Chem import AllChem
from rdkit.Chem import Recap
from rdkit.RDLogger import logger
logger = logger()
tests=[1]*1001
if len(sys.argv)>1:
tests=[0]*1001
tests[1]=1
for x in sys.argv[1:]:
x = int(x)
tests[x] = 1
ts = []
mols = []
lines = gzip.open('../Data/znp.50k.smi.gz','rb').readlines()
logger.info('mols from smiles')
nMols=0
nBad=0
t1=time.time()
for line in lines:
line = line.strip().split(' ')
m = Chem.MolFromSmiles(line[0])
if m:
nMols+=1
mols.append(m)
else:
nBad += 1
t2=time.time()
logger.info('Results1: %.2f seconds, %d passed, %d failed'%(t2-t1,nMols,nBad))
ts.append(t2-t1)
def RunSearch(options,queryFilename):
global sigFactory
if options.similarityType=='AtomPairs':
fpBuilder=FingerprintUtils.BuildAtomPairFP
simMetric=DataStructs.DiceSimilarity
dbName = os.path.join(options.dbDir,options.pairDbName)
fpTableName = options.pairTableName
fpColName = options.pairColName
elif options.similarityType=='TopologicalTorsions':
fpBuilder=FingerprintUtils.BuildTorsionsFP
simMetric=DataStructs.DiceSimilarity
dbName = os.path.join(options.dbDir,options.torsionsDbName)
fpTableName = options.torsionsTableName
fpColName = options.torsionsColName
elif options.similarityType=='RDK':
fpBuilder=FingerprintUtils.BuildRDKitFP
simMetric=DataStructs.FingerprintSimilarity
dbName = os.path.join(options.dbDir,options.fpDbName)
fpTableName = options.fpTableName
if not options.fpColName:
options.fpColName='rdkfp'
fpColName = options.fpColName
elif options.similarityType=='Pharm2D':
fpBuilder=FingerprintUtils.BuildPharm2DFP
simMetric=DataStructs.DiceSimilarity
dbName = os.path.join(options.dbDir,options.fpDbName)
fpTableName = options.pharm2DTableName
if not options.fpColName:
options.fpColName='pharm2dfp'
fpColName = options.fpColName
FingerprintUtils.sigFactory = BuildSigFactory(options)
elif options.similarityType=='Gobbi2D':
from rdkit.Chem.Pharm2D import Gobbi_Pharm2D
fpBuilder=FingerprintUtils.BuildPharm2DFP
simMetric=DataStructs.TanimotoSimilarity
dbName = os.path.join(options.dbDir,options.fpDbName)
fpTableName = options.gobbi2DTableName
if not options.fpColName:
options.fpColName='gobbi2dfp'
fpColName = options.fpColName
FingerprintUtils.sigFactory = Gobbi_Pharm2D.factory
elif options.similarityType=='Morgan':
fpBuilder=FingerprintUtils.BuildMorganFP
simMetric=DataStructs.DiceSimilarity
dbName = os.path.join(options.dbDir,options.morganFpDbName)
fpTableName = options.morganFpTableName
fpColName = options.morganFpColName
extraArgs={}
if options.similarityMetric=='tanimoto':
simMetric = DataStructs.TanimotoSimilarity
elif options.similarityMetric=='dice':
simMetric = DataStructs.DiceSimilarity
elif options.similarityMetric=='tversky':
simMetric = DataStructs.TverskySimilarity
extraArgs['tverskyA']=options.tverskyA
extraArgs['tverskyB']=options.tverskyB
if options.smilesQuery:
mol=Chem.MolFromSmiles(options.smilesQuery)
if not mol:
logger.error('could not build query molecule from smiles "%s"'%options.smilesQuery)
sys.exit(-1)
options.queryMol = mol
elif options.smartsQuery:
mol=Chem.MolFromSmarts(options.smartsQuery)
if not mol:
logger.error('could not build query molecule from smarts "%s"'%options.smartsQuery)
sys.exit(-1)
options.queryMol = mol
if options.outF=='-':
outF=sys.stdout
elif options.outF=='':
outF=None
else:
outF = file(options.outF,'w+')
molsOut=False
if options.sdfOut:
molsOut=True
if options.sdfOut=='-':
sdfOut=sys.stdout
else:
sdfOut = file(options.sdfOut,'w+')
else:
sdfOut=None
if options.smilesOut:
molsOut=True
if options.smilesOut=='-':
smilesOut=sys.stdout
else:
smilesOut = file(options.smilesOut,'w+')
else:
smilesOut=None
if queryFilename:
try:
tmpF = file(queryFilename,'r')
except IOError:
logger.error('could not open query file %s'%queryFilename)
sys.exit(1)
if options.molFormat=='smiles':
func=GetMolsFromSmilesFile
elif options.molFormat=='sdf':
func=GetMolsFromSDFile
if not options.silent:
msg='Reading query molecules'
if fpBuilder: msg+=' and generating fingerprints'
logger.info(msg)
probes=[]
i=0
nms=[]
for nm,smi,mol in func(queryFilename,None,options.nameProp):
i+=1
nms.append(nm)
if not mol:
logger.error('query molecule %d could not be built'%(i))
probes.append((None,None))
continue
if fpBuilder:
probes.append((mol,fpBuilder(mol)))
else:
probes.append((mol,None))
if not options.silent and not i%1000:
logger.info(" done %d"%i)
else:
probes=None
conn=None
idName = options.molIdName
ids=None
names=None
molDbName = os.path.join(options.dbDir,options.molDbName)
molIdName = options.molIdName
mConn = DbConnect(molDbName)
cns = [(x.lower(),y) for x,y in mConn.GetColumnNamesAndTypes('molecules')]
idCol,idTyp=cns[0]
if options.propQuery or options.queryMol:
conn = DbConnect(molDbName)
curs = conn.GetCursor()
if options.queryMol:
if not options.silent: logger.info('Doing substructure query')
if options.propQuery:
where='where %s'%options.propQuery
else:
where=''
if not options.silent:
curs.execute('select count(*) from molecules %(where)s'%locals())
nToDo = curs.fetchone()[0]
join=''
doSubstructFPs=False
fpDbName = os.path.join(options.dbDir,options.fpDbName)
if os.path.exists(fpDbName) and not options.negateQuery :
curs.execute("attach database '%s' as fpdb"%(fpDbName))
try:
curs.execute('select * from fpdb.%s limit 1'%options.layeredTableName)
except:
pass
else:
doSubstructFPs=True
join = 'join fpdb.%s using (%s)'%(options.layeredTableName,idCol)
query = LayeredOptions.GetQueryText(options.queryMol)
if query:
if not where:
where='where'
else:
where += ' and'
where += ' '+query
cmd = 'select %(idCol)s,molpkl from molecules %(join)s %(where)s'%locals()
curs.execute(cmd)
row=curs.fetchone()
nDone=0
ids=[]
while row:
id,molpkl = row
if not options.zipMols:
m = Chem.Mol(str(molpkl))
else:
m = Chem.Mol(zlib.decompress(str(molpkl)))
matched=m.HasSubstructMatch(options.queryMol)
if options.negateQuery:
matched = not matched
if matched:
ids.append(id)
nDone+=1
if not options.silent and not nDone%500:
if not doSubstructFPs:
logger.info(' searched %d (of %d) molecules; %d hits so far'%(nDone,nToDo,len(ids)))
else:
logger.info(' searched through %d molecules; %d hits so far'%(nDone,len(ids)))
row=curs.fetchone()
if not options.silent and doSubstructFPs and nToDo:
nFiltered = nToDo-nDone
logger.info(' Fingerprint screenout rate: %d of %d (%%%.2f)'%(nFiltered,nToDo,100.*nFiltered/nToDo))
elif options.propQuery:
if not options.silent: logger.info('Doing property query')
propQuery=options.propQuery.split(';')[0]
curs.execute('select %(idCol)s from molecules where %(propQuery)s'%locals())
ids = [x[0] for x in curs.fetchall()]
if not options.silent:
logger.info('Found %d molecules matching the query'%(len(ids)))
t1=time.time()
if probes:
if not options.silent: logger.info('Finding Neighbors')
conn = DbConnect(dbName)
cns = conn.GetColumnNames(fpTableName)
curs = conn.GetCursor()
if ids:
ids = [(x,) for x in ids]
curs.execute('create temporary table _tmpTbl (%(idCol)s %(idTyp)s)'%locals())
curs.executemany('insert into _tmpTbl values (?)',ids)
join='join _tmpTbl using (%(idCol)s)'%locals()
else:
join=''
if cns[0].lower() != idCol.lower():
# backwards compatibility to the days when mol tables had a guid and
# the fps tables did not:
curs.execute("attach database '%(molDbName)s' as mols"%locals())
curs.execute("""
select %(idCol)s,%(fpColName)s from %(fpTableName)s join
(select %(idCol)s,%(molIdName)s from mols.molecules %(join)s)
using (%(molIdName)s)
"""%(locals()))
else:
curs.execute('select %(idCol)s,%(fpColName)s from %(fpTableName)s %(join)s'%locals())
def poolFromCurs(curs,similarityMethod):
row = curs.fetchone()
while row:
id,pkl = row
fp = DepickleFP(str(pkl),similarityMethod)
yield (id,fp)
row = curs.fetchone()
topNLists = GetNeighborLists(probes,options.topN,poolFromCurs(curs,options.similarityType),
simMetric=simMetric,simThresh=options.simThresh,**extraArgs)
uniqIds=set()
nbrLists = {}
for i,nm in enumerate(nms):
topNLists[i].reverse()
scores=topNLists[i].GetPts()
nbrNames = topNLists[i].GetExtras()
nbrs = []
for j,nbrGuid in enumerate(nbrNames):
if nbrGuid is None:
break
else:
uniqIds.add(nbrGuid)
nbrs.append((nbrGuid,scores[j]))
nbrLists[(i,nm)] = nbrs
t2=time.time()
if not options.silent: logger.info('The search took %.1f seconds'%(t2-t1))
if not options.silent: logger.info('Creating output')
curs = mConn.GetCursor()
ids = list(uniqIds)
ids = [(x,) for x in ids]
curs.execute('create temporary table _tmpTbl (%(idCol)s %(idTyp)s)'%locals())
curs.executemany('insert into _tmpTbl values (?)',ids)
curs.execute('select %(idCol)s,%(molIdName)s from molecules join _tmpTbl using (%(idCol)s)'%locals())
nmDict={}
for guid,id in curs.fetchall():
nmDict[guid]=str(id)
ks = nbrLists.keys()
ks.sort()
if not options.transpose:
for i,nm in ks:
nbrs= nbrLists[(i,nm)]
nbrTxt=options.outputDelim.join([nm]+['%s%s%.3f'%(nmDict[id],options.outputDelim,score) for id,score in nbrs])
if outF: print >>outF,nbrTxt
else:
labels = ['%s%sSimilarity'%(x[1],options.outputDelim) for x in ks]
if outF: print >>outF,options.outputDelim.join(labels)
for i in range(options.topN):
outL = []
for idx,nm in ks:
nbr = nbrLists[(idx,nm)][i]
outL.append(nmDict[nbr[0]])
outL.append('%.3f'%nbr[1])
if outF: print >>outF,options.outputDelim.join(outL)
else:
if not options.silent: logger.info('Creating output')
curs = mConn.GetCursor()
ids = [(x,) for x in set(ids)]
curs.execute('create temporary table _tmpTbl (%(idCol)s %(idTyp)s)'%locals())
curs.executemany('insert into _tmpTbl values (?)',ids)
molIdName = options.molIdName
curs.execute('select %(idCol)s,%(molIdName)s from molecules join _tmpTbl using (%(idCol)s)'%locals())
nmDict={}
for guid,id in curs.fetchall():
nmDict[guid]=str(id)
if outF: print >>outF,'\n'.join(nmDict.values())
if molsOut and ids:
molDbName = os.path.join(options.dbDir,options.molDbName)
cns = [x.lower() for x in mConn.GetColumnNames('molecules')]
if cns[-1]!='molpkl':
cns.remove('molpkl')
cns.append('molpkl')
curs = mConn.GetCursor()
#curs.execute('create temporary table _tmpTbl (guid integer)'%locals())
#curs.executemany('insert into _tmpTbl values (?)',ids)
cnText=','.join(cns)
curs.execute('select %(cnText)s from molecules join _tmpTbl using (%(idCol)s)'%locals())
row=curs.fetchone()
molD = {}
while row:
row = list(row)
pkl = row[-1]
m = Chem.Mol(str(pkl))
guid = row[0]
nm = nmDict[guid]
if sdfOut:
m.SetProp('_Name',nm)
print >>sdfOut,Chem.MolToMolBlock(m)
for i in range(1,len(cns)-1):
pn = cns[i]
pv = str(row[i])
print >>sdfOut,'> <%s>\n%s\n'%(pn,pv)
print >>sdfOut,'$$$$'
if smilesOut:
smi=Chem.MolToSmiles(m,options.chiralSmiles)
if smilesOut:
print >>smilesOut,'%s %s'%(smi,str(row[1]))
row=curs.fetchone()
if not options.silent: logger.info('Done!')
TYPE_CHECKER = copy.copy(Option.TYPE_CHECKER)
TYPE_CHECKER["floatlist"] = check_floatlist
parser = OptionParser("distance predict", version="%prog", option_class=MyOption)
parser.add_option("--maxPathLength", "--max", default=8, type=int, help="maximum length path for the fingerprint")
parser.add_option("--similarityThreshold", "--sim", default=[0.9], type="floatlist", help="threshold for similarity")
parser.add_option("--numNeighbors", "--num", "-n", "-k", default=50, type=int, help="number of neighbors to consider")
parser.add_option("--neighborsFile", "--nbrs", default="", help="name of an output file to hold the neighbor lists")
parser.add_option("--scan", default=False, action="store_true")
if __name__ == "__main__":
options, args = parser.parse_args()
outF = file(args[-1], "w+")
logger.info("reading training molecules and generating fingerprints")
suppl = Chem.SDMolSupplier(args[0])
train = []
for i, mol in enumerate(suppl):
if not mol:
continue
smi = Chem.MolToSmiles(mol, True)
nm = mol.GetProp(nameField)
property = float(mol.GetProp(propField))
fp = GetMolFingerprint(mol, options.maxPathLength)
train.append((nm, smi, fp, property))
logger.info(" got %d molecules" % len(train))
if len(args) > 2:
suppl = Chem.SDMolSupplier(args[1])
haveTest = True
from rdkit.RDLogger import logger
from chemgrams import get_arpa_vocab, KenLMDeepSMILESLanguageModel, DeepSMILESLanguageModelUtils, DeepSMILESTokenizer, \
LanguageModelMCTSWithUCB1
from rdkit import rdBase
rdBase.DisableLog('rdApp.error')
rdBase.DisableLog('rdApp.warning')
logger = logger()
if __name__ == '__main__':
logger.info("loading language model...")
vocab = get_arpa_vocab(
'../resources/chemts_250k_deepsmiles_klm_10gram_200429.arpa')
lm = KenLMDeepSMILESLanguageModel(
'../resources/chemts_250k_deepsmiles_klm_10gram_200429.klm', vocab)
num_simulations = 1000
width = 3
text_length = 25
start_state = ["<s>"]
def eval_function(text):
generated = ''.join(text)
try:
decoded = DeepSMILESLanguageModelUtils.decode(generated,
start='<s>',
end='</s>')
DeepSMILESLanguageModelUtils.sanitize(decoded)
def CreateDb(options, dataFilename="", supplier=None):
if not dataFilename and supplier is None:
raise ValueError, "Please provide either a data filename or a supplier"
if options.errFilename:
errFile = file(os.path.join(options.outDir, options.errFilename), "w+")
else:
errFile = None
if options.noExtras:
options.doPairs = False
options.doDescriptors = False
options.doFingerprints = False
options.doPharm2D = False
options.doGobbi2D = False
options.doLayered = False
options.doMorganFps = False
if options.loadMols:
if supplier is None:
if not options.molFormat:
ext = os.path.splitext(dataFilename)[-1].lower()
if ext == ".sdf":
options.molFormat = "sdf"
elif ext in (".smi", ".smiles", ".txt", ".csv"):
options.molFormat = "smiles"
if not options.delimiter:
# guess the delimiter
import csv
sniffer = csv.Sniffer()
dlct = sniffer.sniff(file(dataFilename, "r").read(2000))
options.delimiter = dlct.delimiter
if not options.silent:
logger.info(
"Guessing that delimiter is %s. Use --delimiter argument if this is wrong."
% repr(options.delimiter)
)
if not options.silent:
logger.info(
"Guessing that mol format is %s. Use --molFormat argument if this is wrong."
% repr(options.molFormat)
)
if options.molFormat == "smiles":
if options.delimiter == "\\t":
options.delimiter = "\t"
supplier = Chem.SmilesMolSupplier(
dataFilename,
titleLine=options.titleLine,
delimiter=options.delimiter,
smilesColumn=options.smilesColumn,
nameColumn=options.nameColumn,
)
else:
supplier = Chem.SDMolSupplier(dataFilename)
if not options.silent:
logger.info("Reading molecules and constructing molecular database.")
Loader.LoadDb(
supplier,
os.path.join(options.outDir, options.molDbName),
errorsTo=errFile,
regName=options.regName,
nameCol=options.molIdName,
skipProps=options.skipProps,
defaultVal=options.missingPropertyVal,
addComputedProps=options.addProps,
uniqNames=True,
skipSmiles=options.skipSmiles,
maxRowsCached=int(options.maxRowsCached),
silent=options.silent,
nameProp=options.nameProp,
lazySupplier=int(options.maxRowsCached) > 0,
)
if options.doPairs:
pairConn = DbConnect(os.path.join(options.outDir, options.pairDbName))
pairCurs = pairConn.GetCursor()
try:
pairCurs.execute("drop table %s" % (options.pairTableName))
except:
pass
pairCurs.execute(
"create table %s (guid integer not null primary key,%s varchar not null unique,atompairfp blob,torsionfp blob)"
% (options.pairTableName, options.molIdName)
)
if options.doFingerprints or options.doPharm2D or options.doGobbi2D or options.doLayered:
fpConn = DbConnect(os.path.join(options.outDir, options.fpDbName))
fpCurs = fpConn.GetCursor()
try:
fpCurs.execute("drop table %s" % (options.fpTableName))
except:
pass
try:
fpCurs.execute("drop table %s" % (options.pharm2DTableName))
except:
pass
try:
fpCurs.execute("drop table %s" % (options.gobbi2DTableName))
except:
pass
try:
fpCurs.execute("drop table %s" % (options.layeredTableName))
except:
pass
if options.doFingerprints:
fpCurs.execute(
"create table %s (guid integer not null primary key,%s varchar not null unique,rdkfp blob)"
% (options.fpTableName, options.molIdName)
)
if options.doLayered:
layeredQs = ",".join("?" * LayeredOptions.nWords)
colDefs = ",".join(["Col_%d integer" % (x + 1) for x in range(LayeredOptions.nWords)])
fpCurs.execute(
"create table %s (guid integer not null primary key,%s varchar not null unique,%s)"
% (options.layeredTableName, options.molIdName, colDefs)
)
if options.doPharm2D:
fpCurs.execute(
"create table %s (guid integer not null primary key,%s varchar not null unique,pharm2dfp blob)"
% (options.pharm2DTableName, options.molIdName)
)
sigFactory = BuildSigFactory(options)
if options.doGobbi2D:
fpCurs.execute(
"create table %s (guid integer not null primary key,%s varchar not null unique,gobbi2dfp blob)"
% (options.gobbi2DTableName, options.molIdName)
)
from rdkit.Chem.Pharm2D import Generate, Gobbi_Pharm2D
if options.doMorganFps:
fpConn = DbConnect(os.path.join(options.outDir, options.fpDbName))
fpCurs = fpConn.GetCursor()
try:
fpCurs.execute("drop table %s" % (options.morganFpTableName))
except:
pass
fpCurs.execute(
"create table %s (guid integer not null primary key,%s varchar not null unique,morganfp blob)"
% (options.morganFpTableName, options.molIdName)
)
if options.doDescriptors:
descrConn = DbConnect(os.path.join(options.outDir, options.descrDbName))
calc = cPickle.load(file(options.descriptorCalcFilename, "rb"))
nms = [x for x in calc.GetDescriptorNames()]
descrCurs = descrConn.GetCursor()
descrs = ["guid integer not null primary key", "%s varchar not null unique" % options.molIdName]
descrs.extend(["%s float" % x for x in nms])
try:
descrCurs.execute("drop table %s" % (options.descrTableName))
except:
pass
descrCurs.execute("create table %s (%s)" % (options.descrTableName, ",".join(descrs)))
descrQuery = ",".join([DbModule.placeHolder] * len(descrs))
pairRows = []
fpRows = []
layeredRows = []
descrRows = []
pharm2DRows = []
gobbi2DRows = []
morganRows = []
if not options.silent:
logger.info("Generating fingerprints and descriptors:")
molConn = DbConnect(os.path.join(options.outDir, options.molDbName))
molCurs = molConn.GetCursor()
if not options.skipSmiles:
molCurs.execute("select guid,%s,smiles,molpkl from %s" % (options.molIdName, options.regName))
else:
molCurs.execute("select guid,%s,molpkl from %s" % (options.molIdName, options.regName))
i = 0
while 1:
try:
tpl = molCurs.fetchone()
molGuid = tpl[0]
molId = tpl[1]
pkl = tpl[-1]
i += 1
except:
break
mol = Chem.Mol(str(pkl))
if not mol:
continue
if options.doPairs:
pairs = FingerprintUtils.BuildAtomPairFP(mol)
torsions = FingerprintUtils.BuildTorsionsFP(mol)
pkl1 = DbModule.binaryHolder(pairs.ToBinary())
pkl2 = DbModule.binaryHolder(torsions.ToBinary())
row = (molGuid, molId, pkl1, pkl2)
pairRows.append(row)
if options.doFingerprints:
fp2 = FingerprintUtils.BuildRDKitFP(mol)
pkl = DbModule.binaryHolder(fp2.ToBinary())
row = (molGuid, molId, pkl)
fpRows.append(row)
if options.doLayered:
words = LayeredOptions.GetWords(mol)
row = [molGuid, molId] + words
layeredRows.append(row)
if options.doDescriptors:
descrs = calc.CalcDescriptors(mol)
row = [molGuid, molId]
row.extend(descrs)
descrRows.append(row)
if options.doPharm2D:
FingerprintUtils.sigFactory = sigFactory
fp = FingerprintUtils.BuildPharm2DFP(mol)
pkl = DbModule.binaryHolder(fp.ToBinary())
row = (molGuid, molId, pkl)
pharm2DRows.append(row)
if options.doGobbi2D:
FingerprintUtils.sigFactory = Gobbi_Pharm2D.factory
fp = FingerprintUtils.BuildPharm2DFP(mol)
pkl = DbModule.binaryHolder(fp.ToBinary())
row = (molGuid, molId, pkl)
gobbi2DRows.append(row)
if options.doMorganFps:
morgan = FingerprintUtils.BuildMorganFP(mol)
pkl = DbModule.binaryHolder(morgan.ToBinary())
row = (molGuid, molId, pkl)
morganRows.append(row)
if not i % 500:
if len(pairRows):
pairCurs.executemany("insert into %s values (?,?,?,?)" % options.pairTableName, pairRows)
pairRows = []
pairConn.Commit()
if len(fpRows):
fpCurs.executemany("insert into %s values (?,?,?)" % options.fpTableName, fpRows)
fpRows = []
fpConn.Commit()
if len(layeredRows):
fpCurs.executemany(
"insert into %s values (?,?,%s)" % (options.layeredTableName, layeredQs), layeredRows
)
layeredRows = []
fpConn.Commit()
if len(descrRows):
descrCurs.executemany("insert into %s values (%s)" % (options.descrTableName, descrQuery), descrRows)
descrRows = []
descrConn.Commit()
if len(pharm2DRows):
fpCurs.executemany("insert into %s values (?,?,?)" % options.pharm2DTableName, pharm2DRows)
pharm2DRows = []
fpConn.Commit()
if len(gobbi2DRows):
fpCurs.executemany("insert into %s values (?,?,?)" % options.gobbi2DTableName, gobbi2DRows)
gobbi2DRows = []
fpConn.Commit()
if len(morganRows):
fpCurs.executemany("insert into %s values (?,?,?)" % options.morganFpTableName, morganRows)
morganRows = []
fpConn.Commit()
if not options.silent and not i % 500:
logger.info(" Done: %d" % (i))
if len(pairRows):
pairCurs.executemany("insert into %s values (?,?,?,?)" % options.pairTableName, pairRows)
pairRows = []
pairConn.Commit()
if len(fpRows):
fpCurs.executemany("insert into %s values (?,?,?)" % options.fpTableName, fpRows)
fpRows = []
fpConn.Commit()
if len(layeredRows):
fpCurs.executemany("insert into %s values (?,?,%s)" % (options.layeredTableName, layeredQs), layeredRows)
layeredRows = []
fpConn.Commit()
if len(descrRows):
descrCurs.executemany("insert into %s values (%s)" % (options.descrTableName, descrQuery), descrRows)
descrRows = []
descrConn.Commit()
if len(pharm2DRows):
fpCurs.executemany("insert into %s values (?,?,?)" % options.pharm2DTableName, pharm2DRows)
pharm2DRows = []
fpConn.Commit()
if len(gobbi2DRows):
fpCurs.executemany("insert into %s values (?,?,?)" % options.gobbi2DTableName, gobbi2DRows)
gobbi2DRows = []
fpConn.Commit()
if len(morganRows):
fpCurs.executemany("insert into %s values (?,?,?)" % options.morganFpTableName, morganRows)
morganRows = []
fpConn.Commit()
if not options.silent:
logger.info("Finished.")
from rdkit import Chem
from rdkit import RDConfig
import time, cPickle, sys, gzip
from rdkit.RDLogger import logger
logger = logger()
logger.info('reading smarts')
qs = []
smas = []
for line in file(RDConfig.RDDataDir + '/SmartsLib/RLewis_smarts.txt',
'r').readlines():
if line[0] == '#':
continue
line = line.split(' ')
p = Chem.MolFromSmarts(line[0])
if not p:
print >> sys.stderr, line[0]
continue
smas.append(line[0])
qs.append(p)
logger.info('reading target counts')
refFps = cPickle.loads(gzip.open('fps.1000.counts.pkl.gz', 'rb').read())
fps = []
logger.info('reading mols:')
ms = cPickle.loads(gzip.open('mols.1000.pkl.gz', 'rb').read())
t1 = time.time()
nFail = 0
for i, m in enumerate(ms):
fp = [0] * len(qs)
from rdkit import Chem
from rdkit import RDConfig
import time,cPickle,sys,gzip
from rdkit.RDLogger import logger
logger = logger()
logger.info('reading smarts')
qs = []
smas = []
for line in file(RDConfig.RDDataDir+'/SmartsLib/RLewis_smarts.txt','r').readlines():
if line[0] == '#':
continue
line = line.split(' ')
p = Chem.MolFromSmarts(line[0])
if not p:
print >>sys.stderr,line[0]
continue
smas.append(line[0])
qs.append(p)
logger.info('reading target counts')
refFps = cPickle.loads(gzip.open('fps.1000.counts.pkl.gz','rb').read())
fps = []
logger.info('reading mols:')
ms = cPickle.loads(gzip.open('mols.1000.pkl.gz','rb').read())
t1 = time.time()
nFail=0
for i,m in enumerate(ms):
fp = [0]*len(qs)
for j,q in enumerate(qs):
# propField is the name of the property (from the SD file) you want to use
# as the "activity"
propField = 'chemical_shift_1'
# similarity threshold for a pair to be considered interesting.
# (i.e. pairs with a similarity below this value will not be
# added to the output.
similarityThreshold = 0.5
if __name__ == '__main__':
suppl = Chem.SDMolSupplier(sys.argv[1])
outF = file(sys.argv[2], 'w+')
data = []
logger.info('reading molecules and generating fingeprints')
for i, mol in enumerate(suppl):
if not mol:
continue
smi = Chem.MolToSmiles(mol, True)
nm = mol.GetProp(nameField)
property = float(mol.GetProp(propField))
fp = GetMolFingerprint(mol, maxPathLength)
data.append((nm, smi, property, fp))
logger.info(' got %d molecules' % len(data))
logger.info('calculating pairs')
pairs = []
for i in range(len(data)):
for j in range(i + 1, len(data)):
from chemgrams import get_arpa_vocab, KenLMDeepSMILESLanguageModel, DeepSMILESLanguageModelUtils, \
LanguageModelMCTSWithPUCTTerminating
from chemgrams.qedscorer import QEDScorer
from rdkit.RDLogger import logger
from rdkit import rdBase
rdBase.DisableLog('rdApp.error')
rdBase.DisableLog('rdApp.warning')
logger = logger()
THIS_DIR = os.path.dirname(os.path.abspath(__file__))
if __name__ == '__main__':
logger.info("loading language model...")
vocab = get_arpa_vocab(
'../resources/chemts_250k_deepsmiles_klm_6gram_190414.arpa')
lm = KenLMDeepSMILESLanguageModel(
'../resources/chemts_250k_deepsmiles_klm_6gram_190414.klm', vocab)
num_simulations = 100000
width = 24
max_depth = 100
start_state = ["<s>"]
c = 5
qedscorer = QEDScorer()
all_smiles = {}
from rdkit import Chem
from rdkit.Chem import AllChem
from rdkit.RDLogger import logger
logger = logger()
tests = [1] * 1001
if len(sys.argv) > 1:
tests = [0] * 1001
tests[1] = 1
for x in sys.argv[1:]:
x = int(x)
tests[x] = 1
ts = []
mols = []
lines = gzip.open('../Data/znp.50k.smi.gz', 'rt').readlines()
logger.info('mols from smiles')
nMols = 0
nBad = 0
t1 = time.time()
for line in lines:
line = line.strip().split(' ')
m = Chem.MolFromSmiles(line[0])
if m:
nMols += 1
mols.append(m)
else:
nBad += 1
t2 = time.time()
logger.info('Results1: %.2f seconds, %d passed, %d failed' % (t2 - t1, nMols, nBad))
ts.append(t2 - t1)
def RunSearch(options, queryFilename):
global sigFactory
if options.similarityType == 'AtomPairs':
fpBuilder = FingerprintUtils.BuildAtomPairFP
simMetric = DataStructs.DiceSimilarity
dbName = os.path.join(options.dbDir, options.pairDbName)
fpTableName = options.pairTableName
fpColName = options.pairColName
elif options.similarityType == 'TopologicalTorsions':
fpBuilder = FingerprintUtils.BuildTorsionsFP
simMetric = DataStructs.DiceSimilarity
dbName = os.path.join(options.dbDir, options.torsionsDbName)
fpTableName = options.torsionsTableName
fpColName = options.torsionsColName
elif options.similarityType == 'RDK':
fpBuilder = FingerprintUtils.BuildRDKitFP
simMetric = DataStructs.FingerprintSimilarity
dbName = os.path.join(options.dbDir, options.fpDbName)
fpTableName = options.fpTableName
if not options.fpColName:
options.fpColName = 'rdkfp'
fpColName = options.fpColName
elif options.similarityType == 'Pharm2D':
fpBuilder = FingerprintUtils.BuildPharm2DFP
simMetric = DataStructs.DiceSimilarity
dbName = os.path.join(options.dbDir, options.fpDbName)
fpTableName = options.pharm2DTableName
if not options.fpColName:
options.fpColName = 'pharm2dfp'
fpColName = options.fpColName
FingerprintUtils.sigFactory = BuildSigFactory(options)
elif options.similarityType == 'Gobbi2D':
from rdkit.Chem.Pharm2D import Gobbi_Pharm2D
fpBuilder = FingerprintUtils.BuildPharm2DFP
simMetric = DataStructs.TanimotoSimilarity
dbName = os.path.join(options.dbDir, options.fpDbName)
fpTableName = options.gobbi2DTableName
if not options.fpColName:
options.fpColName = 'gobbi2dfp'
fpColName = options.fpColName
FingerprintUtils.sigFactory = Gobbi_Pharm2D.factory
elif options.similarityType == 'Morgan':
fpBuilder = FingerprintUtils.BuildMorganFP
simMetric = DataStructs.DiceSimilarity
dbName = os.path.join(options.dbDir, options.morganFpDbName)
fpTableName = options.morganFpTableName
fpColName = options.morganFpColName
extraArgs = {}
if options.similarityMetric == 'tanimoto':
simMetric = DataStructs.TanimotoSimilarity
elif options.similarityMetric == 'dice':
simMetric = DataStructs.DiceSimilarity
elif options.similarityMetric == 'tversky':
simMetric = DataStructs.TverskySimilarity
extraArgs['tverskyA'] = options.tverskyA
extraArgs['tverskyB'] = options.tverskyB
if options.smilesQuery:
mol = Chem.MolFromSmiles(options.smilesQuery)
if not mol:
logger.error('could not build query molecule from smiles "%s"' %
options.smilesQuery)
sys.exit(-1)
options.queryMol = mol
elif options.smartsQuery:
mol = Chem.MolFromSmarts(options.smartsQuery)
if not mol:
logger.error('could not build query molecule from smarts "%s"' %
options.smartsQuery)
sys.exit(-1)
options.queryMol = mol
if options.outF == '-':
outF = sys.stdout
elif options.outF == '':
outF = None
else:
outF = open(options.outF, 'w+')
molsOut = False
if options.sdfOut:
molsOut = True
if options.sdfOut == '-':
sdfOut = sys.stdout
else:
sdfOut = open(options.sdfOut, 'w+')
else:
sdfOut = None
if options.smilesOut:
molsOut = True
if options.smilesOut == '-':
smilesOut = sys.stdout
else:
smilesOut = open(options.smilesOut, 'w+')
else:
smilesOut = None
if queryFilename:
try:
tmpF = open(queryFilename, 'r')
except IOError:
logger.error('could not open query file %s' % queryFilename)
sys.exit(1)
if options.molFormat == 'smiles':
func = GetMolsFromSmilesFile
elif options.molFormat == 'sdf':
func = GetMolsFromSDFile
if not options.silent:
msg = 'Reading query molecules'
if fpBuilder: msg += ' and generating fingerprints'
logger.info(msg)
probes = []
i = 0
nms = []
for nm, smi, mol in func(queryFilename, None, options.nameProp):
i += 1
nms.append(nm)
if not mol:
logger.error('query molecule %d could not be built' % (i))
probes.append((None, None))
continue
if fpBuilder:
probes.append((mol, fpBuilder(mol)))
else:
probes.append((mol, None))
if not options.silent and not i % 1000:
logger.info(" done %d" % i)
else:
probes = None
conn = None
idName = options.molIdName
ids = None
names = None
molDbName = os.path.join(options.dbDir, options.molDbName)
molIdName = options.molIdName
mConn = DbConnect(molDbName)
cns = [(x.lower(), y)
for x, y in mConn.GetColumnNamesAndTypes('molecules')]
idCol, idTyp = cns[0]
if options.propQuery or options.queryMol:
conn = DbConnect(molDbName)
curs = conn.GetCursor()
if options.queryMol:
if not options.silent: logger.info('Doing substructure query')
if options.propQuery:
where = 'where %s' % options.propQuery
else:
where = ''
if not options.silent:
curs.execute('select count(*) from molecules %(where)s' %
locals())
nToDo = curs.fetchone()[0]
join = ''
doSubstructFPs = False
fpDbName = os.path.join(options.dbDir, options.fpDbName)
if os.path.exists(fpDbName) and not options.negateQuery:
curs.execute("attach database '%s' as fpdb" % (fpDbName))
try:
curs.execute('select * from fpdb.%s limit 1' %
options.layeredTableName)
except:
pass
else:
doSubstructFPs = True
join = 'join fpdb.%s using (%s)' % (
options.layeredTableName, idCol)
query = LayeredOptions.GetQueryText(options.queryMol)
if query:
if not where:
where = 'where'
else:
where += ' and'
where += ' ' + query
cmd = 'select %(idCol)s,molpkl from molecules %(join)s %(where)s' % locals(
)
curs.execute(cmd)
row = curs.fetchone()
nDone = 0
ids = []
while row:
id, molpkl = row
if not options.zipMols:
m = _molFromPkl(molpkl)
else:
m = Chem.Mol(zlib.decompress(molpkl))
matched = m.HasSubstructMatch(options.queryMol)
if options.negateQuery:
matched = not matched
if matched:
ids.append(id)
nDone += 1
if not options.silent and not nDone % 500:
if not doSubstructFPs:
logger.info(
' searched %d (of %d) molecules; %d hits so far' %
(nDone, nToDo, len(ids)))
else:
logger.info(
' searched through %d molecules; %d hits so far' %
(nDone, len(ids)))
row = curs.fetchone()
if not options.silent and doSubstructFPs and nToDo:
nFiltered = nToDo - nDone
logger.info(
' Fingerprint screenout rate: %d of %d (%%%.2f)' %
(nFiltered, nToDo, 100. * nFiltered / nToDo))
elif options.propQuery:
if not options.silent: logger.info('Doing property query')
propQuery = options.propQuery.split(';')[0]
curs.execute(
'select %(idCol)s from molecules where %(propQuery)s' %
locals())
ids = [x[0] for x in curs.fetchall()]
if not options.silent:
logger.info('Found %d molecules matching the query' % (len(ids)))
t1 = time.time()
if probes:
if not options.silent: logger.info('Finding Neighbors')
conn = DbConnect(dbName)
cns = conn.GetColumnNames(fpTableName)
curs = conn.GetCursor()
if ids:
ids = [(x, ) for x in ids]
curs.execute(
'create temporary table _tmpTbl (%(idCol)s %(idTyp)s)' %
locals())
curs.executemany('insert into _tmpTbl values (?)', ids)
join = 'join _tmpTbl using (%(idCol)s)' % locals()
else:
join = ''
if cns[0].lower() != idCol.lower():
# backwards compatibility to the days when mol tables had a guid and
# the fps tables did not:
curs.execute("attach database '%(molDbName)s' as mols" % locals())
curs.execute("""
select %(idCol)s,%(fpColName)s from %(fpTableName)s join
(select %(idCol)s,%(molIdName)s from mols.molecules %(join)s)
using (%(molIdName)s)
""" % (locals()))
else:
curs.execute(
'select %(idCol)s,%(fpColName)s from %(fpTableName)s %(join)s'
% locals())
def poolFromCurs(curs, similarityMethod):
row = curs.fetchone()
while row:
id, pkl = row
fp = DepickleFP(pkl, similarityMethod)
yield (id, fp)
row = curs.fetchone()
topNLists = GetNeighborLists(probes,
options.topN,
poolFromCurs(curs,
options.similarityType),
simMetric=simMetric,
simThresh=options.simThresh,
**extraArgs)
uniqIds = set()
nbrLists = {}
for i, nm in enumerate(nms):
topNLists[i].reverse()
scores = topNLists[i].GetPts()
nbrNames = topNLists[i].GetExtras()
nbrs = []
for j, nbrGuid in enumerate(nbrNames):
if nbrGuid is None:
break
else:
uniqIds.add(nbrGuid)
nbrs.append((nbrGuid, scores[j]))
nbrLists[(i, nm)] = nbrs
t2 = time.time()
if not options.silent:
logger.info('The search took %.1f seconds' % (t2 - t1))
if not options.silent: logger.info('Creating output')
curs = mConn.GetCursor()
ids = list(uniqIds)
ids = [(x, ) for x in ids]
curs.execute('create temporary table _tmpTbl (%(idCol)s %(idTyp)s)' %
locals())
curs.executemany('insert into _tmpTbl values (?)', ids)
curs.execute(
'select %(idCol)s,%(molIdName)s from molecules join _tmpTbl using (%(idCol)s)'
% locals())
nmDict = {}
for guid, id in curs.fetchall():
nmDict[guid] = str(id)
ks = list(nbrLists.keys())
ks.sort()
if not options.transpose:
for i, nm in ks:
nbrs = nbrLists[(i, nm)]
nbrTxt = options.outputDelim.join([nm] + [
'%s%s%.3f' % (nmDict[id], options.outputDelim, score)
for id, score in nbrs
])
if outF: print(nbrTxt, file=outF)
else:
labels = [
'%s%sSimilarity' % (x[1], options.outputDelim) for x in ks
]
if outF: print(options.outputDelim.join(labels), file=outF)
for i in range(options.topN):
outL = []
for idx, nm in ks:
nbr = nbrLists[(idx, nm)][i]
outL.append(nmDict[nbr[0]])
outL.append('%.3f' % nbr[1])
if outF: print(options.outputDelim.join(outL), file=outF)
else:
if not options.silent: logger.info('Creating output')
curs = mConn.GetCursor()
ids = [(x, ) for x in set(ids)]
curs.execute('create temporary table _tmpTbl (%(idCol)s %(idTyp)s)' %
locals())
curs.executemany('insert into _tmpTbl values (?)', ids)
molIdName = options.molIdName
curs.execute(
'select %(idCol)s,%(molIdName)s from molecules join _tmpTbl using (%(idCol)s)'
% locals())
nmDict = {}
for guid, id in curs.fetchall():
nmDict[guid] = str(id)
if outF: print('\n'.join(nmDict.values()), file=outF)
if molsOut and ids:
molDbName = os.path.join(options.dbDir, options.molDbName)
cns = [x.lower() for x in mConn.GetColumnNames('molecules')]
if cns[-1] != 'molpkl':
cns.remove('molpkl')
cns.append('molpkl')
curs = mConn.GetCursor()
#curs.execute('create temporary table _tmpTbl (guid integer)'%locals())
#curs.executemany('insert into _tmpTbl values (?)',ids)
cnText = ','.join(cns)
curs.execute(
'select %(cnText)s from molecules join _tmpTbl using (%(idCol)s)' %
locals())
row = curs.fetchone()
molD = {}
while row:
row = list(row)
m = _molFromPkl(row[-1])
guid = row[0]
nm = nmDict[guid]
if sdfOut:
m.SetProp('_Name', nm)
print(Chem.MolToMolBlock(m), file=sdfOut)
for i in range(1, len(cns) - 1):
pn = cns[i]
pv = str(row[i])
print >> sdfOut, '> <%s>\n%s\n' % (pn, pv)
print('$$$$', file=sdfOut)
if smilesOut:
smi = Chem.MolToSmiles(m, options.chiralSmiles)
if smilesOut:
print('%s %s' % (smi, str(row[1])), file=smilesOut)
row = curs.fetchone()
if not options.silent: logger.info('Done!')
from rdkit.Chem import AllChem
from rdkit.RDLogger import logger
logger = logger()
tests = [1] * 1001
if len(sys.argv) > 1:
tests = [0] * 1001
tests[1] = 1
for x in sys.argv[1:]:
x = int(x)
tests[x] = 1
ts = []
mols = []
lines = gzip.open("../Data/znp.50k.smi.gz", "rb").readlines()
logger.info("mols from smiles")
nMols = 0
nBad = 0
t1 = time.time()
for line in lines:
line = line.strip().split(" ")
m = Chem.MolFromSmiles(line[0])
if m:
nMols += 1
mols.append(m)
else:
nBad += 1
t2 = time.time()
logger.info("Results1: %.2f seconds, %d passed, %d failed" % (t2 - t1, nMols, nBad))
ts.append(t2 - t1)
parser = OptionParser("distance predict", version='%prog', option_class=MyOption)
parser.add_option('--maxPathLength', '--max', default=8, type=int,
help='maximum length path for the fingerprint')
parser.add_option('--similarityThreshold', '--sim', default=[0.9], type='floatlist',
help='threshold for similarity')
parser.add_option('--numNeighbors', '--num', '-n', '-k', default=50, type=int,
help='number of neighbors to consider')
parser.add_option('--neighborsFile', '--nbrs', default='',
help='name of an output file to hold the neighbor lists')
parser.add_option('--scan', default=False, action="store_true")
if __name__ == '__main__':
options, args = parser.parse_args()
outF = file(args[-1], 'w+')
logger.info('reading training molecules and generating fingerprints')
suppl = Chem.SDMolSupplier(args[0])
train = []
for i, mol in enumerate(suppl):
if not mol:
continue
smi = Chem.MolToSmiles(mol, True)
nm = mol.GetProp(nameField)
property = float(mol.GetProp(propField))
fp = GetMolFingerprint(mol, options.maxPathLength)
train.append((nm, smi, fp, property))
logger.info(' got %d molecules' % len(train))
if len(args) > 2:
suppl = Chem.SDMolSupplier(args[1])
haveTest = True
logger = logger()
tests = [1] * 1001
if len(sys.argv) > 1:
tests = [0] * 1001
for x in sys.argv[1:]:
x = int(x)
tests[x] = 1
ts = []
mols = []
if tests[0]:
lines = gzip.open(data('znp.50k.smi.gz'), 'rt').readlines()
logger.info('mols from smiles')
nMols = 0
nBad = 0
t1 = time.time()
for line in lines:
line = line.strip().split(' ')
m = Chem.MolFromSmiles(line[0])
if m:
nMols += 1
mols.append(m)
else:
nBad += 1
t2 = time.time()
logger.info('Results1: %.2f seconds, %d passed, %d failed' %
(t2 - t1, nMols, nBad))
from rdkit.Chem import Recap
from rdkit.RDLogger import logger
logger = logger()
tests = [1] * 1001
if len(sys.argv) > 1:
tests = [0] * 1001
tests[1] = 1
for x in sys.argv[1:]:
x = int(x)
tests[x] = 1
ts = []
sdData = gzip.open("../Data/mols.1000.sdf.gz").read()
logger.info("mols from sdf")
suppl = Chem.SDMolSupplier()
suppl.SetData(sdData)
mols = []
nMols = 0
nBad = 0
t1 = time.time()
for m in suppl:
if m:
nMols += 1
mols.append(m)
else:
nBad += 1
t2 = time.time()
logger.info("Results1: %.2f seconds, %d passed, %d failed" % (t2 - t1, nMols, nBad))
ts.append(t2 - t1)