def initFromLines(self, lines):
import re
spaces = re.compile("[\ \t]+")
feats = []
rads = []
for lineNum, line in enumerate(lines):
txt = line.split("#")[0].strip()
if txt:
splitL = spaces.split(txt)
if len(splitL) < 5:
logger.error(
"Input line %d only contains %d fields, 5 are required. Read failed." % (lineNum, len(splitL))
)
return
fName = splitL[0]
try:
xP = float(splitL[1])
yP = float(splitL[2])
zP = float(splitL[3])
rad = float(splitL[4])
except ValueError:
logger.error("Error parsing a number of line %d. Read failed." % (lineNum))
return
feats.append(ChemicalFeatures.FreeChemicalFeature(fName, fName, Geometry.Point3D(xP, yP, zP)))
rads.append(rad)
self._initializeFeats(feats, rads)
def initFromLines(self, lines):
import re
spaces = re.compile('[\ \t]+')
feats = []
rads = []
for lineNum, line in enumerate(lines):
txt = line.split('#')[0].strip()
if txt:
splitL = spaces.split(txt)
if len(splitL) < 5:
logger.error(
'Input line %d only contains %d fields, 5 are required. Read failed.'
% (lineNum, len(splitL)))
return
fName = splitL[0]
try:
xP = float(splitL[1])
yP = float(splitL[2])
zP = float(splitL[3])
rad = float(splitL[4])
except ValueError:
logger.error(
'Error parsing a number of line %d. Read failed.' %
(lineNum))
return
feats.append(
ChemicalFeatures.FreeChemicalFeature(
fName, fName, Geometry.Point3D(xP, yP, zP)))
rads.append(rad)
self._initializeFeats(feats, rads)
def checkConstraints(mol,recConf,feat,filt):
# identify what constraint we have to test:
mC = mol.GetConformer()
matchId = feat.GetAtomIds()[0]
# all values in the filters are "anded" so we need to loop over them
for i in range(2,len(filt),3):
if filt[i]=="Distance":
# get the partners
dist = mC.GetAtomPosition(matchId).Distance(recConf.GetAtomPosition(filt[1][0]))
#!!!POTENTIAL ERROR!!! Identified by SamudBe1 on 04AUG2014 18:44 PDT
#!!!POTENTIAL ERROR!!! In the following line, the code should be "dist>=filt[i+1]" AND "dist<filt[i+2]"
if (dist<=filt[i+1]) or (dist>filt[i+2]):
return False
elif filt[i]=="Angle":
# changed to be able to handle multiple neighbour atoms for rec
# the first entry in filt is the matching atom - then a list of all others are neighbours
# and the same is now also added for the ligand (ie multiple neighbour atoms)
# get the matching atom neighbours - the ligans don't get the H's read in - so we can use
# that list directly
fitsRecAngle = False
fitsLigAngle = False
# first the receptor angles - there are usually less nieghbours for rec atoms
for neighIdx in filt[1][1]:
# here we are looping over all possible combinations and check at the very end if
# both angles are ok
# if an angle is already ok we can skip the second calculation
# get the vectors
l1 = mC.GetAtomPosition(matchId)-recConf.GetAtomPosition(filt[1][0])
l2 = recConf.GetAtomPosition(neighIdx)-recConf.GetAtomPosition(filt[1][0])
angle = math.degrees(l1.AngleTo(l2))
# old version: angle = math.degrees(math.acos(l1.DotProduct(l2)/(l1.Length()*l2.Length())))
if (angle>filt[i+1]) and (angle<filt[i+2]):
fitsRecAngle = True
break
if not fitsRecAngle:
return False
# now we check on the ligands
neighbAtmIdx = [a.GetIdx() for a in mol.GetAtomWithIdx(matchId).GetNeighbors()]
for idx in neighbAtmIdx:
l1 = mC.GetAtomPosition(idx)-mC.GetAtomPosition(matchId)
l2 = recConf.GetAtomPosition(filt[1][0])-mC.GetAtomPosition(matchId)
angle = math.degrees(l1.AngleTo(l2))
# old version: angle = math.degrees(math.acos(l1.DotProduct(l2)/(l1.Length()*l2.Length())))
if (angle>filt[i+1]) and (angle<filt[i+2]):
fitsLigAngle = True
break
if not fitsLigAngle:
return False
else:
logger.error("Requesting a constraint that is not defined %s" % filt[2])
return False
# we only reach this position if the ligand matches all queries
return True
def SaveState(self, fileName):
""" Writes this calculator off to a file so that it can be easily loaded later
**Arguments**
- fileName: the name of the file to be written
"""
try:
f = open(fileName, 'wb+')
except Exception:
logger.error('cannot open output file %s for writing' % (fileName))
return
pickle.dump(self, f)
f.close()
def SaveState(self, fileName):
""" Writes this calculator off to a file so that it can be easily loaded later
**Arguments**
- fileName: the name of the file to be written
"""
try:
f = open(fileName, 'wb+')
except Exception:
logger.error('cannot open output file %s for writing' % (fileName))
return
cPickle.dump(self, f)
f.close()
if options.outF!='-':
options.outF = file(options.outF,'w+')
else:
options.outF = sys.stdout
# get the ph4 Scoring parameters
try:
options.ph4DescOutFile = open(options.ph4DescOutFile,'w')
except:
parser.error('Error opening You need to have a ph4DescOutFile file')
try:
f = open(options.ph4desc,'r')
except IOError, err:
logger.error(err)
featureOrder=[]
splitL = []
Ph4_Init_value={}
Ph4_Descriptors={}
Ph4_location={}
Ph4_Type={}
Ph4_radius={}
Ph4_FeatureType={}
for line in f:
line=line.rstrip("\n")
#(Name,Init_value,Type,radius,x,y,z,Feature_Type)=line.split("\t")
splitL = line.split(" ")
morganRows = []
fpConn.Commit()
if not options.silent:
logger.info('Finished.')
if __name__=='__main__':
options,args = parser.parse_args()
if options.loadMols:
if len(args)!=1:
parser.error('please provide a filename argument')
dataFilename = args[0]
try:
dataFile = open(dataFilename,'r')
except IOError:
logger.error('input file %s does not exist'%(dataFilename))
sys.exit(0)
dataFile=None
if not options.outDir:
prefix = os.path.splitext(dataFilename)[0]
options.outDir=prefix
if not os.path.exists(options.outDir):
try:
os.mkdir(options.outDir)
except:
logger.error('could not create output directory %s'%options.outDir)
sys.exit(1)
if 1:
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!')
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!')
# get the input data
# the receptor
try:
rec = Chem.MolFromMol2File(options.recF,removeHs=False)
# setup the conformer for the receptor
recConf = rec.GetConformer()
except:
print ("Problem reading receptor from %s" % options.recF)
print "Error:", sys.exc_info()[0]
quit()
try:
f = open(options.ph4F,'r')
except IOError, err:
logger.error(err)
ph4Filters=[]
for line in f:
line=line.rstrip("\n")
splitL=line.split(" ")
# modify the types from str to the relevant types
splitL[1]=int(splitL[1])
# I am not going to worry about "Represntation Errors"!
for i in range(2,len(splitL),3):
splitL[i+1]=float(splitL[i+1])
splitL[i+2]=float(splitL[i+2])
ph4Filters.append(splitL)
if options.scoreFilter:
try:
f = open(options.scoreF,'r')
logger.info(f'testing {nm} queries')
t1 = time.time()
nPossible = 0
nTested = 0
nFound = 0
nErrors = 0
for i, fragfp in enumerate(qfps):
for j, mfp in enumerate(mfps):
nPossible += 1
if args.validateResults:
matched = mols[j].HasSubstructMatch(qs[i])
fpMatch = DataStructs.AllProbeBitsMatch(fragfp, mfp)
if fpMatch:
nTested += 1
if matched:
nFound += 1
if not fpMatch:
nErrors += 1
logger.error(f"ERROR: mol {j} query {i}")
else:
if DataStructs.AllProbeBitsMatch(fragfp, mfp):
nTested += 1
if mols[j].HasSubstructMatch(qs[i]):
nFound += 1
t2 = time.time()
ts.append(t2 - t1)
logger.info(
f'Results{len(ts)}: {t2-t1 : .2f} seconds. {nTested} tested ({nTested/nPossible :.4f} of total), {nFound} found, {nFound/nTested : .2f} accuracy. {nErrors} errors.'
)
print(f"| {rdkit.__version__} | {' | '.join(['%.1f' % x for x in ts])} |")