def main():
if not oefastrocs.OEFastROCSIsGPUReady():
oechem.OEThrow.Info("No supported GPU available!")
return 0
args = getargs()
dbname = args.d
# read in database
ifs = oechem.oemolistream()
if not ifs.open(dbname):
oechem.OEThrow.Fatal("Unable to open '%s'" % dbname)
print("Opening database file %s ..." % dbname)
timer = oechem.OEWallTimer()
opts = oefastrocs.OEShapeDatabaseOptions()
opts.SetLimit(1)
dbase = oefastrocs.OEShapeDatabase()
moldb = oechem.OEMolDatabase()
if not moldb.Open(ifs):
oechem.OEThrow.Fatal("Unable to open '%s'" % dbname)
dots = oechem.OEThreadedDots(10000, 200, "conformers")
if not dbase.Open(moldb, dots):
oechem.OEThrow.Fatal("Unable to initialize OEShapeDatabase on '%s'" %
dbname)
dots.Total()
print("%f seconds to load database" % timer.Elapsed())
df = pd.read_csv(args.i)
res = []
for smile in tqdm(df.loc[:, 'smiles'].tolist()):
resn = len(res)
try:
q = FromString(smile)[0]
for score in dbase.GetSortedScores(q, 1):
res.append(score.GetTanimotoCombo())
break
except KeyboardInterrupt:
print("caught")
exit()
except:
res.append(np.nan)
if len(res) == resn:
res.append(np.nan)
df['fastroc'] = res
print(df.head)
df.to_csv(args.o, sep=',', index=False)
return 0
def run(self):
""" Open the database file and load it into the OEShapeDatabase """
timer = oechem.OEWallTimer()
sys.stderr.write("Opening database file %s ...\n" % self.dbname)
if not self.moldb.Open(self.dbname):
oechem.OEThrow.Fatal("Unable to open '%s'" % self.dbname)
dots = oechem.OEThreadedDots(10000, 200, "conformers")
if not self.shapedb.Open(self.moldb, dots):
oechem.OEThrow.Fatal(
"Unable to initialize OEShapeDatabase on '%s'" % self.dbname)
dots.Total()
sys.stderr.write("%s seconds to load database\n" % timer.Elapsed())
self.loadedEvent.set()
def main(argv=[__name__]):
itf = oechem.OEInterface()
oechem.OEConfigure(itf, InterfaceData)
oegraphsim.OEConfigureFingerPrint(
itf, oegraphsim.OEGetFPType(oegraphsim.OEFPType_Tree))
if not oechem.OEParseCommandLine(itf, argv):
return 1
ifname = itf.GetString("-in")
ffname = itf.GetString("-fpdb")
if oechem.OEGetFileExtension(ffname) != "fpbin":
oechem.OEThrow.Fatal(
"Fingerprint database file should have '.fpbin' file extension!")
idxfname = oechem.OEGetMolDatabaseIdxFileName(ifname)
if not os.path.exists(idxfname):
if not oechem.OECreateMolDatabaseIdx(ifname):
oechem.OEThrow.Warning("Unable to create %s molecule index file" %
idxfname)
oechem.OEThrow.Info("Using %s index molecule file" % idxfname)
moldb = oechem.OEMolDatabase()
if not moldb.Open(ifname):
oechem.OEThrow.Fatal("Cannot open molecule database file!")
nrmols = moldb.GetMaxMolIdx()
fptype = oegraphsim.OESetupFingerPrint(itf)
oechem.OEThrow.Info("Using fingerprint type %s" % fptype.GetFPTypeString())
opts = oegraphsim.OECreateFastFPDatabaseOptions(fptype)
opts.SetTracer(oechem.OEDots(100000, 1000, "fingerprints"))
oechem.OEThrow.Info("Generating fingerprints with %d threads" %
opts.GetNumProcessors())
timer = oechem.OEWallTimer()
if not oegraphsim.OECreateFastFPDatabaseFile(ffname, ifname, opts):
oechem.OEThrow.Fatal("Cannot create fingerprint database file!")
oechem.OEThrow.Info("%5.2f secs to generate %d fingerprints" %
(timer.Elapsed(), nrmols))
return 0
def main(argv=[__name__]):
parser = argparse.ArgumentParser()
# positional arguments retaining backward compatibility
parser.add_argument(
'database',
help='File containing the database molecules to be search \
(format not restricted to *.oeb).')
parser.add_argument(
'query',
default=[],
nargs='+',
help='File containing the query molecule(s) to be search \
(format not restricted to *.oeb).')
parser.add_argument(
'--nHits',
dest='nHits',
type=int,
default=100,
help='Number of hits to return (default = number of database mols).')
parser.add_argument('--cutoff',
dest='cutoff',
type=float,
default=argparse.SUPPRESS,
help='Specify a cutoff criteria for scores.')
parser.add_argument(
'--tversky',
dest='tversky',
action='store_true',
default=argparse.SUPPRESS,
help='Switch to Tversky similarity scoring (default = Tanimoto).')
args = parser.parse_args()
dbname = args.database
if not oefastrocs.OEFastROCSIsGPUReady():
oechem.OEThrow.Info("No supported GPU available!")
return 0
# set options
opts = oefastrocs.OEShapeDatabaseOptions()
opts.SetLimit(args.nHits)
print("Number of hits set to %u" % opts.GetLimit())
if hasattr(args, 'cutoff') is not False:
opts.SetCutoff(args.cutoff)
print("Cutoff set to %f" % args.cutoff)
if hasattr(args, 'tversky') is not False:
opts.SetSimFunc(args.tversky)
print("Tversky similarity scoring set.")
# read in database
ifs = oechem.oemolistream()
if not ifs.open(dbname):
oechem.OEThrow.Fatal("Unable to open '%s'" % dbname)
print("\nOpening database file %s ..." % dbname)
timer = oechem.OEWallTimer()
dbase = oefastrocs.OEShapeDatabase()
moldb = oechem.OEMolDatabase()
if not moldb.Open(ifs):
oechem.OEThrow.Fatal("Unable to open '%s'" % dbname)
dots = oechem.OEThreadedDots(10000, 200, "conformers")
if not dbase.Open(moldb, dots):
oechem.OEThrow.Fatal("Unable to initialize OEShapeDatabase on '%s'" %
dbname)
dots.Total()
print("%f seconds to load database\n" % timer.Elapsed())
for qfname in args.query:
# read in query
qfs = oechem.oemolistream()
if not qfs.open(qfname):
oechem.OEThrow.Fatal("Unable to open '%s'" % qfname)
mcmol = oechem.OEMol()
if not oechem.OEReadMolecule(qfs, mcmol):
oechem.OEThrow.Fatal("Unable to read query from '%s'" % qfname)
qfs.rewind()
ext = oechem.OEGetFileExtension(qfname)
qmolidx = 0
while oechem.OEReadMolecule(qfs, mcmol):
# write out to file name based on molecule title
ofs = oechem.oemolostream()
moltitle = mcmol.GetTitle()
if len(moltitle) == 0:
moltitle = str(qmolidx)
ofname = moltitle + "_results." + ext
if not ofs.open(ofname):
oechem.OEThrow.Fatal("Unable to open '%s'" % argv[4])
print("Searching for %s of %s (%s conformers)" %
(moltitle, qfname, mcmol.NumConfs()))
qconfidx = 0
for conf in mcmol.GetConfs():
for score in dbase.GetSortedScores(conf, opts):
dbmol = oechem.OEMol()
dbmolidx = score.GetMolIdx()
if not moldb.GetMolecule(dbmol, dbmolidx):
print(
"Unable to retrieve molecule '%u' from the database"
% dbmolidx)
continue
mol = oechem.OEGraphMol(
dbmol.GetConf(oechem.OEHasConfIdx(score.GetConfIdx())))
oechem.OESetSDData(mol, "QueryConfidx", "%s" % qconfidx)
oechem.OESetSDData(mol, "ShapeTanimoto",
"%.4f" % score.GetShapeTanimoto())
oechem.OESetSDData(mol, "ColorTanimoto",
"%.4f" % score.GetColorTanimoto())
oechem.OESetSDData(mol, "TanimotoCombo",
"%.4f" % score.GetTanimotoCombo())
score.Transform(mol)
oechem.OEWriteMolecule(ofs, mol)
qconfidx += 1
print("%s conformers processed" % qconfidx)
print("Wrote results to %s\n" % ofname)
qmolidx += 1
return 0
def main(argv=[__name__]):
itf = oechem.OEInterface()
oechem.OEConfigure(itf, InterfaceData)
defopts = oegraphsim.OEFPDatabaseOptions(10,
oegraphsim.OESimMeasure_Tanimoto)
oegraphsim.OEConfigureFPDatabaseOptions(itf, defopts)
oegraphsim.OEConfigureFPDatabaseMemoryType(itf)
if not oechem.OEParseCommandLine(itf, argv):
return 0
qfname = itf.GetString("-query")
mfname = itf.GetString("-molfname")
ffname = itf.GetString("-fpdbfname")
ofname = itf.GetString("-out")
# initialize databases
timer = oechem.OEWallTimer()
timer.Start()
ifs = oechem.oemolistream()
if not ifs.open(qfname):
oechem.OEThrow.Fatal("Cannot open input file!")
query = oechem.OEGraphMol()
if not oechem.OEReadMolecule(ifs, query):
oechem.OEThrow.Fatal("Cannot read query molecule!")
moldb = oechem.OEMolDatabase()
if not moldb.Open(mfname):
oechem.OEThrow.Fatal("Cannot open molecule database!")
memtype = oegraphsim.OEGetFPDatabaseMemoryType(itf)
fpdb = oegraphsim.OEFastFPDatabase(ffname, memtype)
if not fpdb.IsValid():
oechem.OEThrow.Fatal("Cannot open fingerprint database!")
nrfps = fpdb.NumFingerPrints()
memtypestr = fpdb.GetMemoryTypeString()
ofs = oechem.oemolostream()
if not ofs.open(ofname):
oechem.OEThrow.Fatal("Cannot open output file!")
if not oegraphsim.OEAreCompatibleDatabases(moldb, fpdb):
oechem.OEThrow.Fatal("Databases are not compatible!")
oechem.OEThrow.Info("%5.2f sec to initialize databases" % timer.Elapsed())
fptype = fpdb.GetFPTypeBase()
oechem.OEThrow.Info("Using fingerprint type %s" % fptype.GetFPTypeString())
opts = oegraphsim.OEFPDatabaseOptions()
oegraphsim.OESetupFPDatabaseOptions(opts, itf)
# search fingerprint database
timer.Start()
scores = fpdb.GetSortedScores(query, opts)
oechem.OEThrow.Info("%5.2f sec to search %d fingerprints %s" %
(timer.Elapsed(), nrfps, memtypestr))
timer.Start()
nrhits = 0
hit = oechem.OEGraphMol()
for si in scores:
if moldb.GetMolecule(hit, si.GetIdx()):
nrhits += 1
oechem.OESetSDData(hit, "Similarity score", "%.2f" % si.GetScore())
oechem.OEWriteMolecule(ofs, hit)
oechem.OEThrow.Info("%5.2f sec to write %d hits" %
(timer.Elapsed(), nrhits))
return 0
def main(argv=[__name__]):
if len(argv) < 3:
oechem.OEThrow.Usage("%s <database> [<queries> ... ]" % argv[0])
if not oefastrocs.OEFastROCSIsGPUReady():
oechem.OEThrow.Info("No supported GPU available!")
return 0
dbname = argv[1]
# read in database
ifs = oechem.oemolistream()
if not ifs.open(dbname):
oechem.OEThrow.Fatal("Unable to open '%s'" % dbname)
print("Opening database file %s ..." % dbname)
timer = oechem.OEWallTimer()
dbase = oefastrocs.OEShapeDatabase()
moldb = oechem.OEMolDatabase()
if not moldb.Open(ifs):
oechem.OEThrow.Fatal("Unable to open '%s'" % dbname)
dots = oechem.OEThreadedDots(10000, 200, "conformers")
if not dbase.Open(moldb, dots):
oechem.OEThrow.Fatal("Unable to initialize OEShapeDatabase on '%s'" % dbname)
dots.Total()
print("%f seconds to load database" % timer.Elapsed())
for qfname in argv[2:]:
# read in query
qfs = oechem.oemolistream()
if not qfs.open(qfname):
oechem.OEThrow.Fatal("Unable to open '%s'" % qfname)
query = oechem.OEGraphMol()
if not oechem.OEReadMolecule(qfs, query):
oechem.OEThrow.Fatal("Unable to read query from '%s'" % qfname)
ext = oechem.OEGetFileExtension(qfname)
base = qfname[:-(len(ext) + 1)]
# write out everthing to a similary named file
ofs = oechem.oemolostream()
ofname = base + "_results." + ext
if not ofs.open(ofname):
oechem.OEThrow.Fatal("Unable to open '%s'" % argv[4])
print("Searching for %s" % qfname)
numHits = moldb.NumMols()
for score in dbase.GetSortedScores(query, numHits):
dbmol = oechem.OEMol()
molidx = score.GetMolIdx()
if not moldb.GetMolecule(dbmol, molidx):
print("Unable to retrieve molecule '%u' from the database" % molidx)
continue
mol = oechem.OEGraphMol(dbmol.GetConf(oechem.OEHasConfIdx(score.GetConfIdx())))
oechem.OESetSDData(mol, "ShapeTanimoto", "%.4f" % score.GetShapeTanimoto())
oechem.OESetSDData(mol, "ColorTanimoto", "%.4f" % score.GetColorTanimoto())
oechem.OESetSDData(mol, "TanimotoCombo", "%.4f" % score.GetTanimotoCombo())
score.Transform(mol)
oechem.OEWriteMolecule(ofs, mol)
print("Wrote results to %s" % ofname)
return 0
def GetBestOverlays(self, querymolstr, options, iformat, oformat):
""" Return a string of the format specified by 'oformat'
containing nhits overlaid confomers using querymolstr as the
query interpretted as iformat.
querymolstr - a string containing a molecule to use as the query
options - an instance of OEShapeDatabaseOptions
iformat - a string representing the file extension to parse the querymolstr as.
Note: old clients could be passing .sq files, so
iformat == '.oeb' will try to interpret the file as
a .sq file.
oformat - file format to write the results as
"""
timer = oechem.OEWallTimer()
# make sure to wait for the load to finish
blocking = True
loaded = self.IsLoaded(blocking)
assert loaded
if iformat.startswith(".sq"):
query = ReadShapeQuery(querymolstr)
else:
# read in query
qfs = oechem.oemolistream()
qfs = SetupStream(qfs, iformat)
if not qfs.openstring(querymolstr):
raise ValueError("Unable to open input molecule string")
query = oechem.OEGraphMol()
if not oechem.OEReadMolecule(qfs, query):
if iformat == ".oeb": # could be an old client trying to send a .sq file.
query = ReadShapeQuery(querymolstr)
else:
raise ValueError(
"Unable to read a molecule from the string of format '%s'"
% iformat)
ofs = oechem.oemolostream()
ofs = SetupStream(ofs, oformat)
if not ofs.openstring():
raise ValueError("Unable to openstring for output")
# do we only want shape based results?
# this is a "Write" lock to be paranoid and not overload the GPU
self.rwlock.AcquireWriteLock()
try:
# do search
scores = self.shapedb.GetSortedScores(query, options)
sys.stderr.write("%f seconds to do search\n" % timer.Elapsed())
finally:
self.rwlock.ReleaseWriteLock()
timer.Start()
# write results
for score in scores:
mcmol = oechem.OEMol()
if not self.moldb.GetMolecule(mcmol, score.GetMolIdx()):
oechem.OEThrow.Warning(
"Can't retrieve molecule %i from the OEMolDatabase, "
"skipping..." % score.GetMolIdx())
continue
# remove hydrogens to make output smaller, this also
# ensures OEPrepareFastROCSMol will have the same output
oechem.OESuppressHydrogens(mcmol)
mol = oechem.OEGraphMol(
mcmol.GetConf(oechem.OEHasConfIdx(score.GetConfIdx())))
oechem.OECopySDData(mol, mcmol)
if options.GetSimFunc() == oefastrocs.OEShapeSimFuncType_Tanimoto:
oechem.OESetSDData(mol, "ShapeTanimoto",
"%.4f" % score.GetShapeTanimoto())
oechem.OESetSDData(mol, "ColorTanimoto",
"%.4f" % score.GetColorTanimoto())
oechem.OESetSDData(mol, "TanimotoCombo",
"%.4f" % score.GetTanimotoCombo())
else:
oechem.OESetSDData(mol, "ShapeTversky",
"%.4f" % score.GetShapeTversky())
oechem.OESetSDData(mol, "ColorTversky",
"%.4f" % score.GetColorTversky())
oechem.OESetSDData(mol, "TverskyCombo",
"%.4f" % score.GetTverskyCombo())
if options.GetInitialOrientation(
) != oefastrocs.OEFastROCSOrientation_Inertial:
oechem.OEAddSDData(
mol, "Opt. Starting Pos.",
GetAltStartsString(options.GetInitialOrientation()))
score.Transform(mol)
oechem.OEWriteMolecule(ofs, mol)
output = ofs.GetString()
sys.stderr.write("%f seconds to write hitlist\n" % timer.Elapsed())
sys.stderr.flush()
ofs.close()
return output
def QueryResults(self):
""" Return the best nhits results of these servers. """
timer = oechem.OEWallTimer()
thrdpool = LaunchFunctionThreadPool(ShapeServer.QueryResults)
for server in self.shapeservers:
thrdpool.AddThread(server)
data = []
for oebdata in thrdpool.GetResults():
data.append(oebdata.data)
sys.stderr.write("%f seconds to get results back" % timer.Elapsed())
data = b"".join(data)
if not data:
sys.stderr.write("Possible query error, no data returned \
by any of the downstream servers")
return ""
timer.Start()
# read in from OEB strings
ifs = oechem.oemolistream()
ifs = SetupStream(ifs, self.oformat)
if not ifs.openstring(data):
sys.stderr.write(
"Unable to open OEB string from downstream server")
return ""
mols = [oechem.OEGraphMol(mol) for mol in ifs.GetOEGraphMols()]
def GetScoreToCmp(mol):
if oechem.OEHasSDData(mol, "ShapeTanimoto"):
# sort by shape tanimoto
if oechem.OEHasSDData(mol, "TanimotoCombo"):
return float(oechem.OEGetSDData(mol, "TanimotoCombo"))
return float(oechem.OEGetSDData(mol, "ShapeTanimoto"))
else:
# sort by shape tversky
if oechem.OEHasSDData(mol, "TverskyCombo"):
return float(oechem.OEGetSDData(mol, "TverskyCombo"))
return float(oechem.OEGetSDData(mol, "ShapeTversky"))
mols.sort(key=GetScoreToCmp)
mols.reverse()
# write back out to an OEB string
ofs = oechem.oemolostream()
ofs = SetupStream(ofs, self.oformat)
ofs.openstring()
nhits = self.nhits
if not nhits:
nhits = len(mols)
for mol in mols[:nhits]:
oechem.OEWriteMolecule(ofs, mol)
sys.stderr.write("%f seconds to collate hitlist" % timer.Elapsed())
return Binary(ofs.GetString())
def main(argv=[__name__]):
itf = oechem.OEInterface()
oechem.OEConfigure(itf, InterfaceData)
defopts = oegraphsim.OEFPDatabaseOptions(10,
oegraphsim.OESimMeasure_Tanimoto)
oegraphsim.OEConfigureFPDatabaseOptions(itf, defopts)
oegraphsim.OEConfigureFingerPrint(
itf, oegraphsim.OEGetFPType(oegraphsim.OEFPType_Tree))
if not oechem.OEParseCommandLine(itf, argv):
return 0
qfname = itf.GetString("-query")
mfname = itf.GetString("-molfname")
ofname = itf.GetString("-out")
# initialize databases
timer = oechem.OEWallTimer()
timer.Start()
ifs = oechem.oemolistream()
if not ifs.open(qfname):
oechem.OEThrow.Fatal("Cannot open input file!")
query = oechem.OEGraphMol()
if not oechem.OEReadMolecule(ifs, query):
oechem.OEThrow.Fatal("Cannot read query molecule!")
moldb = oechem.OEMolDatabase()
if not moldb.Open(mfname):
oechem.OEThrow.Fatal("Cannot open molecule database!")
ofs = oechem.oemolostream()
if not ofs.open(ofname):
oechem.OEThrow.Fatal("Cannot open output file!")
fptype = oegraphsim.OESetupFingerPrint(itf)
oechem.OEThrow.Info("Using fingerprint type %s" % fptype.GetFPTypeString())
fpdb = oegraphsim.OEFPDatabase(fptype)
emptyfp = oegraphsim.OEFingerPrint()
emptyfp.SetFPTypeBase(fptype)
nrmols = moldb.GetMaxMolIdx()
mol = oechem.OEGraphMol()
for idx in range(0, nrmols):
if moldb.GetMolecule(mol, idx):
fpdb.AddFP(mol)
else:
fpdb.AddFP(emptyfp)
nrfps = fpdb.NumFingerPrints()
oechem.OEThrow.Info("%5.2f sec to initialize databases" % timer.Elapsed())
opts = oegraphsim.OEFPDatabaseOptions()
oegraphsim.OESetupFPDatabaseOptions(opts, itf)
# search fingerprint database
timer.Start()
scores = fpdb.GetSortedScores(query, opts)
oechem.OEThrow.Info("%5.2f sec to search %d fingerprints" %
(timer.Elapsed(), nrfps))
timer.Start()
hit = oechem.OEGraphMol()
for si in scores:
if moldb.GetMolecule(hit, si.GetIdx()):
oechem.OEWriteMolecule(ofs, hit)
oechem.OEThrow.Info("%5.2f sec to write %d hits" %
(timer.Elapsed(), opts.GetLimit()))
return 0
fpdb = oegraphsim.OEFPDatabase(oegraphsim.OEFPType_Path)
emptyfp = oegraphsim.OEFingerPrint()
emptyfp.SetFPTypeBase(fpdb.GetFPTypeBase())
mol = oechem.OEGraphMol()
for idx in range(0, nrmols):
if moldb.GetMolecule(mol, idx):
fpdb.AddFP(mol)
else:
fpdb.AddFP(emptyfp)
nrfps = fpdb.NumFingerPrints()
timer = oechem.OEWallTimer()
while True:
# read query SMILES from stdin
sys.stdout.write("Enter SMILES> ")
line = sys.stdin.readline()
line = line.rstrip()
if len(line) == 0:
sys.exit(0)
# parse query
query = oechem.OEGraphMol()
if not oechem.OESmilesToMol(query, line):
oechem.OEThrow.Warning("Invalid SMILES string")