def getSubSearchDb(self,
screenType="SMARTS",
numProc=1,
forceRefresh=False):
if not self.__ssDb or forceRefresh:
oeIo = OeIoUtils()
fp = os.path.join(self.__dirPath,
self.__getSubSearchFileName(screenType))
logger.info("Opening screened substructure search database %r", fp)
self.__ssDb = oeIo.loadOeSubSearchDatabase(fp,
screenType,
numProc=numProc)
return self.__ssDb
def __exhaustiveSubStructureSearch(self, numMols, **kwargs):
"""Exhaustive substructure search."""
try:
limitPerceptions = kwargs.get("limitPerceptions", False)
buildTypeList = kwargs.get("buildTypeList", ["oe-iso-smiles"])
oesmP, ccIdxD = self.__getSearchDataProviders(**kwargs)
oesU = OeSearchUtils(oesmP, fpTypeList=[])
oeioU = OeIoUtils()
#
for ccId, ccD in list(ccIdxD.items())[:numMols]:
matchCount = 0
mtS = set()
for buildType in buildTypeList:
if buildType in ccD:
oeMol = oeioU.descriptorToMol(
ccD[buildType],
buildType,
limitPerceptions=limitPerceptions,
messageTag=ccId + ":" + buildType)
if not oeMol:
logger.error(
"%s %s build query molecule build fails (skipping)",
ccId, buildType)
continue
# ----
startTime = time.time()
retStatus, mL = oesU.searchSubStructure(
oeMol, matchOpts="graph-strict")
if not retStatus:
logger.info("%s match fails for build type %s",
ccId, buildType)
elif not self.__resultContains(ccId, mL):
logger.info(
"%s failed match length %d build type %s in (%.4f seconds)",
ccId, len(mL), buildType,
time.time() - startTime)
elif self.__resultContains(ccId, mL):
mtS.update([m.ccId for m in mL])
matchCount += 1
self.assertTrue(retStatus)
self.assertTrue(self.__resultContains(ccId, mL))
if matchCount:
logger.info("%s MATCHES %d: %r", ccId, matchCount, mtS)
else:
logger.info("%s NO MATCHES", ccId)
# ----
return True
except Exception as e:
logger.exception("Failing with %s", str(e))
self.fail()
return False
def makeFiles(self, fmt="sdf"):
""" Create files (mol, mol2) for all public chemical components.
"""
try:
if fmt not in ["mol", "mol2", "mol2h", "sdf"]:
return False
if not self.__setLicense(self.__licensePath):
logger.error("Invalid license details - exiting")
return False
for ccId, oeMol in self.__oeMolD.items():
if self.__molBuildType == "ideal-xyz":
filePath = os.path.join(self.__fileDirPath, fmt, ccId[0], ccId + "_ideal." + fmt)
oeioU = OeIoUtils()
oeioU.write(filePath, oeMol, constantMol=True)
else:
filePath = os.path.join(self.__fileDirPath, fmt, ccId[0], ccId + "_model." + fmt)
oeioU = OeIoUtils()
oeioU.write(filePath, oeMol, constantMol=True)
return True
except Exception as e:
logger.exception("Failing with %s", str(e))
return False
def getFingerPrintDb(self, fpType, fpDbType="STANDARD", rebuild=False):
if fpType not in self.__fpDbD or rebuild:
oeIo = OeIoUtils()
fastFpDbPath = os.path.join(self.__dirPath,
self.__getFastFpDbFileName(fpType))
oeMolDbFilePath = os.path.join(self.__dirPath,
self.__getOeMolDbFileName())
fpDb = oeIo.loadOeFingerPrintDatabase(oeMolDbFilePath,
fastFpDbPath,
inMemory=True,
fpType=fpType,
fpDbType=fpDbType)
if fpDb:
self.__fpDbD[fpType] = fpDb
#
return self.__fpDbD[fpType]
def getOeMolD(self):
try:
if not self.__oeMolD:
startTime = time.time()
oeIo = OeIoUtils()
self.__oeMolD = oeIo.readOeBinaryMolCache(
os.path.join(self.__dirPath,
self.__getOeSearchMolFileName()))
logger.info(
"Loading OE binary molecule cache length %d (%.4f seconds)",
len(self.__oeMolD),
time.time() - startTime)
return self.__oeMolD
except Exception as e:
logger.exception("Failing with %s", str(e))
return None
def __toMolFile(self, oeMol, molfilePath, **kwargs):
"""Write the
Args:
oeMol (object): instance of an OE graph molecule
molfilePath (string): file path for molfile (type determined by extension)
Returns:
bool: True for success or False otherwise
"""
try:
_ = kwargs
oeio = OeIoUtils()
oeio.write(molfilePath, oeMol, constantMol=True)
return True
except Exception as e:
logger.exception("Failing with %s", str(e))
return False
def __getMol(self,
query,
queryType,
queryId,
limitPerceptions=False,
suppressHydrogens=True):
oeioU = OeIoUtils()
if queryType == "CC":
oeMol = self.__oesmP.getMol(query)
else:
oeMol = oeioU.descriptorToMol(query,
queryType,
limitPerceptions=limitPerceptions,
messageTag=queryId)
#
if suppressHydrogens:
oeMol = oeioU.suppressHydrogens(oeMol)
oeMol.SetTitle(queryId)
return oeMol
def testDepictSMILES(self):
"""Test case - create depiction from SMILES descriptor."""
try:
imagePath = os.path.join(self.__workPath, "benzene-from-smi.svg")
oeio = OeIoUtils()
oeMol = oeio.smilesToMol("c1ccccc1")
oed = OeDepict()
oed.setMolTitleList([("benzene", oeMol, "Title for benzene")])
oed.setDisplayOptions(labelAtomName=False,
labelAtomCIPStereo=True,
labelBondCIPStereo=True,
labelAtomIndex=False,
labelBondIndex=False,
bondDisplayWidth=1.0)
oed.setGridOptions(rows=1, cols=1)
oed.prepare()
oed.write(imagePath)
except Exception as e:
logger.exception("Failing with %s", str(e))
self.fail()
def testSubStructureSearchScreenedFiltered(self):
myKwargs = {
"cachePath": self.__cachePath,
"useCache": True,
"fpTypeList": self.__fpTypeList,
"ccFileNamePrefix": "cc-filtered",
"oeFileNamePrefix": "oe-filtered",
"molBuildType": "oe-iso-smiles",
"limitPerceptions": False,
}
oeioU = OeIoUtils()
oemp = OeMoleculeProvider(**myKwargs)
ok = oemp.testCache()
ccmP = ChemCompIndexProvider(**myKwargs)
ccIdxD = ccmP.getIndex()
ok = ccmP.testCache(minCount=self.__minCount)
self.assertTrue(ok)
oesU = OeSearchUtils(oemp,
screenType=self.__screenType,
numProc=self.__numProc)
numMols = 5000
missL = []
for ccId, ccD in list(ccIdxD.items())[:numMols]:
# ----
startTime = time.time()
if "oe-smiles" not in ccD:
continue
logger.info("Search %s %r", ccId, ccD["oe-smiles"])
oeQMol = oeioU.smartsToQmol(ccD["oe-smiles"])
retStatus, mL = oesU.searchSubStructureScreened(oeQMol,
maxMatches=100)
logger.info("%s (status=%r)match length %d in (%.4f seconds)",
ccId, retStatus, len(mL),
time.time() - startTime)
if not self.__resultContains(ccId, mL):
missL.append(ccId)
# self.assertGreaterEqual(len(mL), 1)
# ----
logger.info("Missed searches (%d) %r", len(missL), missL)
def testIoOps(self):
"""Test IO operation on generated related molecules"""
try:
oeIoU = OeIoUtils()
mU = MarshalUtil()
mU.mkdir(self.__molfileDirPath)
ccMolD = self.__getChemCompDefs()
oemf = OeMoleculeFactory()
for ccId, ccObj in list(ccMolD.items())[:10]:
# ----
tId = oemf.setChemCompDef(ccObj)
self.assertEqual(tId, ccId)
relatedIdxD = oemf.buildRelated(limitPerceptions=False)
logger.info("%s generated %d molecular forms", ccId,
len(relatedIdxD))
for sId, idxD in relatedIdxD.items():
logger.info("sId %r smiles %r", sId, idxD["smiles"])
mol2Path = os.path.join(self.__molfileDirPath,
sId + ".mol2")
oeMol = oeIoU.descriptorToMol(idxD["smiles"],
"oe-iso-smiles",
limitPerceptions=False,
messageTag=None)
oeIoU.write(mol2Path,
oeMol,
constantMol=True,
addSdTags=True)
sdfPath = os.path.join(self.__molfileDirPath, sId + ".mol")
oeMol = oeIoU.descriptorToMol(idxD["smiles"],
"oe-iso-smiles",
limitPerceptions=False,
messageTag=None)
oeIoU.write(sdfPath,
oeMol,
constantMol=True,
addSdTags=True)
# ----
except Exception as e:
logger.exception("Failing with %s", str(e))
self.fail()
def alignMoleculePair(self,
refIdentifier,
refIdentifierType,
fitIdentifier,
fitIdentifierType,
imagePath=None,
**kwargs):
"""Create aligned depiction for a target molecule InChI, SMILES descriptors or PDB identifier."""
try:
imagePath = imagePath if imagePath else self.__makeImagePath()
oeio = OeIoUtils()
ccsw = ChemCompSearchWrapper()
oesmP = ccsw.getSearchMoleculeProvider()
# ---
if refIdentifierType.lower() in ["smiles"]:
oeMolRef = oeio.smilesToMol(refIdentifier)
elif refIdentifierType.lower() in ["inchi"]:
oeMolRef = oeio.inchiToMol(refIdentifier)
elif refIdentifierType.lower() in ["identifierpdb"]:
oeMolRef = oesmP.getMol(refIdentifier)
#
if fitIdentifierType.lower() in ["smiles"]:
oeMolFit = oeio.smilesToMol(fitIdentifier)
elif fitIdentifierType.lower() in ["inchi"]:
oeMolFit = oeio.inchiToMol(fitIdentifier)
elif fitIdentifierType.lower() in ["identifierpdb"]:
oeMolFit = oesmP.getMol(fitIdentifier)
# ---
logger.info("oeMolRef atoms %r", oeMolRef.NumAtoms())
logger.info("oeMolFit atoms %r", oeMolFit.NumAtoms())
displayIdRef = "Ref"
displayIdFit = "Fit"
ok = self.__depictAlignedPair(oeMolRef, displayIdRef, oeMolFit,
displayIdFit, imagePath, **kwargs)
return imagePath if ok else None
except Exception as e:
logger.exception("Failing with %s", str(e))
return None
def testDepictOneSDF(self):
"""Test case - get, read, build OE molecule from SDF file, and depict the molecule."""
try:
imagePath = os.path.join(self.__workPath, "benzene-from-smi.svg")
sdfPath = os.path.join(self.__dataPath, "ATP.sdf")
oeio = OeIoUtils()
oeMolL = oeio.fileToMols(sdfPath)
#
oed = OeDepict()
oed.setMolTitleList([("ATP", oeMolL[0], "Title for ATP")])
oed.setDisplayOptions(labelAtomName=True,
labelAtomCIPStereo=True,
labelBondCIPStereo=True,
labelAtomIndex=False,
labelBondIndex=False,
bondDisplayWidth=0.5)
oed.setGridOptions(rows=1, cols=1)
oed.prepare()
oed.write(imagePath)
except Exception as e:
logger.exception("Failing with %s", str(e))
self.fail()
def depictMolecule(self,
identifier,
identifierType,
imagePath=None,
**kwargs):
"""Create depiction from InChI, SMILES descriptors or PDB identifier."""
try:
imagePath = imagePath if imagePath else self.__makeImagePath()
oeio = OeIoUtils()
if identifierType.lower() in ["smiles"]:
oeMol = oeio.smilesToMol(identifier)
elif identifierType.lower() in ["inchi"]:
oeMol = oeio.inchiToMol(identifier)
elif identifierType.lower() in ["identifierpdb"]:
ccsw = ChemCompSearchWrapper()
oesmP = ccsw.getSearchMoleculeProvider()
oeMol = oesmP.getMol(identifier)
#
ok = self.__depictOne(oeMol, imagePath, **kwargs)
return imagePath if ok else None
except Exception as e:
logger.exception("Failing with %s", str(e))
return None
def __displayAlignedDescriptorPair(self,
ccId,
descrRef,
buildTypeRef,
descrFit,
buildTypeFit,
title=None,
limitPerceptions=True):
oeioU = OeIoUtils()
oeMolRef = oeioU.descriptorToMol(descrRef,
buildTypeRef,
limitPerceptions=limitPerceptions,
messageTag=ccId + ":" + buildTypeRef)
oeMolFit = oeioU.descriptorToMol(descrFit,
buildTypeFit,
limitPerceptions=limitPerceptions,
messageTag=ccId + ":" + buildTypeFit)
#
oed = OeDepictMCSAlignPage()
oed.setSearchType(sType="graph-relaxed", minAtomMatchFraction=0.50)
oed.setDisplayOptions(labelAtomName=True,
labelAtomCIPStereo=True,
labelAtomIndex=False,
labelBondIndex=False,
highlightStyleFit="ballAndStickInverse",
bondDisplayWidth=0.5)
oed.setRefMol(oeMolRef, ccId)
oed.setFitMol(oeMolFit, ccId)
myTitle = title if title else buildTypeRef + "-" + buildTypeFit
imgPath = os.path.join(self.__workPath, myTitle + "-" + ccId + ".svg")
logger.info("Using image path %r", imgPath)
aML = oed.alignPair(imagePath=imgPath)
if aML:
logger.info("%s aligned image path %r", ccId, imgPath)
for (rCC, rAt, tCC, tAt) in aML:
logger.debug("%5s %-5s %5s %-5s", rCC, rAt, tCC, tAt)
def __exhaustiveSubStructureSearch(self, numMols, **kwargs):
"""Exhaustive substructure search."""
try:
limitPerceptions = kwargs.get("limitPerceptions", False)
buildTypeList = kwargs.get("buildTypeList", ["oe-iso-smiles"])
oesmP, ccIdxD = self.__getSearchDataProviders(**kwargs)
oesU = OeSearchUtils(oesmP, fpTypeList=[])
oeioU = OeIoUtils()
#
for ccId, ccD in list(ccIdxD.items())[:numMols]:
for buildType in buildTypeList:
if buildType in ccD:
oeMol = oeioU.descriptorToMol(
ccD[buildType],
buildType,
limitPerceptions=limitPerceptions,
messageTag=ccId + ":" + buildType)
if not oeMol:
continue
# ----
startTime = time.time()
retStatus, mL = oesU.searchSubStructure(
oeMol, matchOpts="graph-strict")
if not self.__resultContains(ccId, mL):
logger.info(
"%s match length %d build type %s in (%.4f seconds)",
ccId, len(mL), buildType,
time.time() - startTime)
self.assertTrue(retStatus)
self.assertTrue(self.__resultContains(ccId, mL))
# ----
return True
except Exception as e:
logger.exception("Failing with %s", str(e))
self.fail()
return False
def __fingerPrintScores(self, numMols, **kwargs):
maxFpResults = kwargs.get("maxResults", 50)
limitPerceptions = kwargs.get("limitPerceptions", True)
fpTypeCuttoffList = kwargs.get("fpTypeCuttoffList", [("TREE", 0.6)])
buildTypeList = kwargs.get("buildTypeList", ["oe-iso-smiles"])
doDisplay = kwargs.get("doDisplay", False)
failedIdList = kwargs.get("failedIdList", [])
#
oesmP, ccIdxD = self.__getSearchDataProviders(**kwargs)
oesU = OeSearchUtils(oesmP,
fpTypeList=[tup[0] for tup in fpTypeCuttoffList])
oeioU = OeIoUtils()
# This will reload the oe binary cache.
oeMol = oesmP.getMol("004")
self.assertGreaterEqual(len(list(oeMol.GetAtoms())), 12)
#
missedFpD = {}
missedBuildD = {}
numMols = min(len(ccIdxD), numMols) if numMols else len(ccIdxD)
logger.info("Begin finger print score search on %d molecules", numMols)
# ----
startTime = time.time()
# for ccId, ccD in list(ccIdxD.items())[:numMols]:
for ii, ccId in enumerate(failedIdList[:numMols]):
ccD = ccIdxD[ccId]
for buildType in buildTypeList:
if buildType in ccD:
oeMol = oeioU.descriptorToMol(
ccD[buildType],
buildType,
limitPerceptions=limitPerceptions,
messageTag=ccId + ":" + buildType)
if not oeMol:
logger.debug("%s build failed for %s - skipping", ccId,
buildType)
continue
maxHits = 0
minHits = maxFpResults
selfHit = False
#
startTime1 = time.time()
for fpType, minFpScore in fpTypeCuttoffList:
retStatus, mL = oesU.getFingerPrintScores(
oeMol, fpType, minFpScore, maxFpResults)
self.assertTrue(retStatus)
logger.debug("%s fpType %r hits %d", ccId, fpType,
len(mL))
maxHits = max(maxHits, len(mL))
minHits = min(minHits, len(mL))
matchedSelf = self.__resultContains(ccId, mL)
selfHit = selfHit or matchedSelf
if not matchedSelf:
missedFpD.setdefault(ccId, []).append(
(buildType, fpType, len(mL)))
#
if not selfHit:
missedBuildD.setdefault(ccId, []).append(buildType)
#
if maxHits < 1 or not selfHit:
logger.info(
"%s MISSED for buildType %r min hits %d max hits %d (%.4f seconds)",
ccId, buildType, minHits, maxHits,
time.time() - startTime1)
else:
logger.debug(
"%s MATCHED for buildType %r min hits %d max hits %d (%.4f seconds)",
ccId, buildType, minHits, maxHits,
time.time() - startTime1)
else:
logger.debug("%s missing descriptor %r", ccId, buildType)
if ii % 100 == 0:
logger.info(
"Completed %d of %d missed count %d in (%.4f seconds)", ii,
len(failedIdList), len(missedBuildD),
time.time() - startTime)
# ------
for ccId, bTL in missedBuildD.items():
logger.info("%s missed all fptypes: buildtype list %r", ccId, bTL)
if ccId in missedFpD:
logger.info("%s unmatched by fpTypes %r", ccId, missedFpD[ccId])
#
if doDisplay:
for ccId, bTL in missedBuildD.items():
idxD = ccIdxD[ccId]
if "oe-iso-smiles" in idxD:
for bT in bTL:
self.__displayAlignedDescriptorPair(
ccId,
idxD["oe-iso-smiles"],
"oe-iso-smiles",
idxD[bT],
bT,
title=None,
limitPerceptions=True)
logger.info("%s fingerprints search on %d in (%.4f seconds)",
len(fpTypeCuttoffList), numMols,
time.time() - startTime)
def __sssWithFingerPrintFromDescriptor(self, numMols, **kwargs):
maxFpResults = kwargs.get("maxResults", 50)
limitPerceptions = kwargs.get("limitPerceptions", False)
fpTypeCuttoffList = kwargs.get("fpTypeCuttoffList", [("TREE", 0.6)])
buildTypeList = kwargs.get("buildTypeList", ["oe-iso-smiles"])
doDisplay = kwargs.get("doDisplay", False)
#
oesmP, ccIdxD = self.__getSearchDataProviders(**kwargs)
oesU = OeSearchUtils(oesmP,
fpTypeList=[tup[0] for tup in fpTypeCuttoffList])
oeioU = OeIoUtils()
# This will reload the oe binary cache.
oeMol = oesmP.getMol("004")
self.assertGreaterEqual(len(list(oeMol.GetAtoms())), 12)
# matchOpts = "graph-relaxed"
matchOpts = "graph-strict"
missTupL = []
missedD = {}
missedFpD = {}
numMols = min(len(ccIdxD), numMols) if numMols else len(ccIdxD)
logger.info(
"Begin substructure search w/ finger print filter on %d molecules",
numMols)
# ----
startTime = time.time()
for (
ii,
ccId,
) in enumerate(list(ccIdxD.keys())[:numMols]):
ccD = ccIdxD[ccId]
for buildType in buildTypeList:
if buildType in ccD:
startTime1 = time.time()
oeMol = oeioU.descriptorToMol(
ccD[buildType],
buildType,
limitPerceptions=limitPerceptions,
messageTag=ccId + ":" + buildType)
if not oeMol:
logger.debug("%s build failed for %s - skipping", ccId,
buildType)
continue
maxHits = 0
minHits = maxFpResults
selfHit = False
for fpType, minFpScore in fpTypeCuttoffList:
retStatus, mL = oesU.searchSubStructureWithFingerPrint(
oeMol,
fpType,
minFpScore,
maxFpResults,
matchOpts=matchOpts)
self.assertTrue(retStatus)
logger.debug("%s fpType %r hits %d", ccId, fpType,
len(mL))
maxHits = max(maxHits, len(mL))
minHits = min(minHits, len(mL))
matchedSelf = self.__resultContains(ccId, mL)
selfHit = selfHit or matchedSelf
if not matchedSelf:
missedFpD.setdefault(ccId, []).append(
(buildType, fpType, len(mL)))
if not selfHit:
missedD.setdefault(ccId, []).append(buildType)
if maxHits < 1 or not selfHit:
logger.info(
"%s (%r) buildType %r min hits %d max hits %d (%.4f seconds)",
ccId, selfHit, buildType, minHits, maxHits,
time.time() - startTime1)
else:
logger.debug("%s missing descriptor %r", ccId, buildType)
if ii % 100 == 0:
logger.info("Completed %d of %d missed count %d", ii, numMols,
len(missedD))
#
for ccId, missL in missedD.items():
logger.info("%s missed list %r", ccId, missL)
if ccId in missedFpD:
logger.info("%s unmatched for fpTypes %r", ccId,
missedFpD[ccId])
# ----
if doDisplay:
mD = {}
for missTup in missTupL:
mD.setdefault(missTup[0], []).append(missTup[1])
for ccId, buildTypeL in mD.items():
idxD = ccIdxD[ccId]
if "oe-iso-smiles" in idxD:
for buildType in buildTypeL:
self.__displayAlignedDescriptorPair(
ccId,
idxD["oe-iso-smiles"],
"oe-iso-smiles",
idxD[buildType],
buildType,
title=None,
limitPerceptions=True)
logger.info("%s fingerprints search on %d in (%.4f seconds)",
len(fpTypeCuttoffList), numMols,
time.time() - startTime)
return True
def __reload(self, **kwargs):
"""Reload the dictionary of OE molecules and related data artifacts for chemical component definitions.
Args:
limitPerceptions(bool): process input descriptors in essentially verbatim mode (default: True)
fpTypeList (list): fingerprint type (TREE,PATH,MACCS,CIRCULAR,LINGO)
screenTypeList (list): fast sub search screen type (MOLECULE, SMARTS, MDL, ... )
useCache (bool, optional): flag to use cached files. Defaults to True.
cachePath (str): path to the top cache directory. Defaults to '.'.
numProc (int): number processors to engage in screen substructure search database generation.
suppressHydrogens (bool, optional): flag to suppress explicit hydrogens in the OE data store.
molLimit (int):
Returns:
(bool) : True for success or False othewise
"""
try:
useCache = kwargs.get("useCache", True)
cachePath = kwargs.get("cachePath", ".")
numProc = kwargs.get("numProc", 2)
molLimit = kwargs.get("molLimit", None)
fpTypeList = kwargs.get(
"fpTypeList", ["TREE", "PATH", "MACCS", "CIRCULAR", "LINGO"])
# screenTypeList = kwargs.get("screenTypeList", ["SMARTS"])
screenTypeList = kwargs.get("screenTypeList", None)
limitPerceptions = kwargs.get("limitPerceptions", False)
suppressHydrogens = kwargs.get("suppressHydrogens", False)
quietFlag = kwargs.get("quietFlag", True)
logSizes = kwargs.get("logSizes", False)
fpDbType = "STANDARD"
buildScreenedDb = True
#
oeCount = 0
errCount = 0
failIdList = []
oeIo = OeIoUtils(quietFlag=quietFlag)
# --------
oeSearchMolFilePath = os.path.join(self.__dirPath,
self.__getOeSearchMolFileName())
if not useCache or (useCache
and not self.__mU.exists(oeSearchMolFilePath)):
cmpKwargs = {
k: v
for k, v in kwargs.items()
if k not in ["cachePath", "useCache", "molLimit"]
}
ccsiP = ChemCompSearchIndexProvider(cachePath=cachePath,
useCache=True,
molLimit=molLimit,
**cmpKwargs)
ok = ccsiP.testCache(minCount=molLimit, logSizes=logSizes)
# ----
ccIdxD = ccsiP.getIndex() if ok else {}
idxCount = len(ccIdxD)
# ------- JDW OE mol construction here -----
startTime = time.time()
oeCount, errCount, failIdList = oeIo.buildOeBinaryMolCacheFromIndex(
oeSearchMolFilePath,
ccIdxD,
quietFlag=quietFlag,
fpTypeList=fpTypeList,
limitPerceptions=limitPerceptions,
suppressHydrogens=suppressHydrogens)
if failIdList:
logger.info("failures %r", failIdList)
endTime = time.time()
logger.info(
"Constructed %d/%d cached oeMols (unconverted %d) (%.4f seconds)",
oeCount, idxCount, errCount, endTime - startTime)
# --------
oeMolDbFilePath = os.path.join(self.__dirPath,
self.__getOeMolDbFileName())
if not useCache or (useCache
and not self.__mU.exists(oeMolDbFilePath)):
startTime = time.time()
molCount = oeIo.createOeBinaryDatabaseAndIndex(
oeSearchMolFilePath, oeMolDbFilePath)
endTime = time.time()
logger.info(
"Created and stored %d indexed oeMols in OE database format (%.4f seconds)",
molCount, endTime - startTime)
# --------
if fpDbType == "FAST":
for fpType in fpTypeList:
startTime = time.time()
# Fast FP search database file names
fpPath = os.path.join(self.__dirPath,
self.__getFastFpDbFileName(fpType))
if not useCache or (useCache
and not self.__mU.exists(fpPath)):
ok = oeIo.createOeFingerPrintDatabase(oeMolDbFilePath,
fpPath,
fpType=fpType)
endTime = time.time()
logger.info(
"Created and stored %s fingerprint database (%.4f seconds)",
fpType, endTime - startTime)
# --------
if buildScreenedDb and screenTypeList:
for screenType in screenTypeList:
startTime = time.time()
fp = os.path.join(self.__dirPath,
self.__getSubSearchFileName(screenType))
if not useCache or (useCache and not self.__mU.exists(fp)):
ok = oeIo.createOeSubSearchDatabase(
oeSearchMolFilePath,
fp,
screenType=screenType,
numProc=numProc)
endTime = time.time()
logger.info(
"Constructed screened substructure database (status %r) with screenType %s (%.4f seconds)",
ok, screenType, endTime - startTime)
# ---------
ssDb = oeIo.loadOeSubSearchDatabase(
fp, screenType=screenType, numProc=numProc)
ok = ssDb.NumMolecules() == oeCount
# ----------
#
return True
except Exception as e:
logger.exception("Failing with %s", str(e))
return False
def buildSearchFiles(self, **kwargs):
"""Build cif, sdf (optional), and mol2 files for components in the chemical component search index.
Exclude ions or other extraneous molecules lacking bonds.
Args:
ccUrlTarget (str): locator for source chemical component dictionary (default: full public dictionary)
birdUrlTarget (str): locator for source BIRD dictionary (default: full public dictionary)
limitPerceptions (bool): restrict automatic perceptions in OE molecular build operations (default: False)
numProc (int): number of processors
useCache (bool): use existing resource file where possible (default: True)
molLimit (str): limit the number to ingested chemical compont (default: None)
quietFlag (bool): suppress output in OE library operations (default: True)
Returns:
(int): number molfiles generated
"""
cachePath = self.__cachePath
ccUrlTarget = kwargs.get("ccUrlTarget", None)
birdUrlTarget = kwargs.get("birdUrlTarget", None)
molLimit = kwargs.get("molLimit", None)
quietFlag = kwargs.get("quietFlag", True)
fpTypeList = kwargs.get("fpTypeList", [])
screenTypeList = kwargs.get("screenTypeList", [])
ccFileNamePrefix = "cc-%s" % self.__prefix if self.__prefix else "cc-full"
oeFileNamePrefix = "oe-%s" % self.__prefix if self.__prefix else "oe-cc-full"
numProc = kwargs.get("numProc", 2)
minCount = kwargs.get("minCount", 0)
useCache = kwargs.get("useCache", True)
useSdf = kwargs.get("useSdf", True)
useMol2 = kwargs.get("useMol2", False)
limitPerceptions = kwargs.get("limitPerceptions", False)
logSizes = False
#
startTime = time.time()
ccmP = ChemCompMoleculeProvider(cachePath=cachePath,
useCache=useCache,
ccFileNamePrefix=ccFileNamePrefix,
ccUrlTarget=ccUrlTarget,
birdUrlTarget=birdUrlTarget,
molLimit=molLimit)
ok = ccmP.testCache(minCount=minCount, logSizes=logSizes)
logger.info(
"Completed chemical component provider load %r (%.4f seconds)", ok,
time.time() - startTime)
#
startTime = time.time()
oesmp = OeSearchMoleculeProvider(
ccUrlTarget=ccUrlTarget,
birdUrlTarget=birdUrlTarget,
cachePath=cachePath,
ccFileNamePrefix=ccFileNamePrefix,
oeFileNamePrefix=oeFileNamePrefix,
useCache=useCache,
quietFlag=quietFlag,
fpTypeList=fpTypeList,
screenTypeList=screenTypeList,
numProc=numProc,
molLimit=molLimit,
limitPerceptions=limitPerceptions,
)
ok = oesmp.testCache()
logger.info("Completed OE molecule provider load %r (%.4f seconds)",
ok,
time.time() - startTime)
#
startTime = time.time()
ccSIdxP = ChemCompSearchIndexProvider(
cachePath=cachePath,
useCache=useCache,
ccFileNamePrefix=ccFileNamePrefix,
limitPerceptions=limitPerceptions,
numProc=numProc)
ok = ccSIdxP.testCache()
logger.info(
"Completed chemical component search index load %r (%.4f seconds)",
ok,
time.time() - startTime)
#
ccSIdx = ccSIdxP.getIndex() if ccSIdxP and ok else {}
logger.info("Search index status %r index length %d", ok, len(ccSIdx))
#
ccIdD = {}
mU = MarshalUtil()
oeU = OeIoUtils(dirPath=cachePath)
numMols = 0
searchFileDirPath = self.getSearchDirFilePath()
pathTupList = []
for sId in ccSIdx:
ccId = sId.split("|")[0]
# standard CIF definition
if ccId not in ccIdD:
cifPath = os.path.join(searchFileDirPath, ccId[0], ccId,
ccId + ".cif")
if not (useCache and mU.exists(cifPath)):
ccMol = ccmP.getMol(ccId)
if not self.__checkCif(ccMol):
continue
mU.doExport(cifPath, [ccMol], fmt="mmcif")
#
oeMol = oesmp.getMol(sId)
if not self.__checkOeMol(oeMol):
continue
#
# Sanity checks on the generated OE molecule
#
cifPath = os.path.join(searchFileDirPath, ccId[0], ccId,
sId + ".cif")
if sId != ccId and not (useCache and mU.exists(cifPath)):
oeccU = OeChemCompUtils()
ok = oeccU.addOeMol(sId,
oeMol,
missingModelXyz=True,
writeIdealXyz=False)
if ok:
oeccU.write(cifPath)
if useSdf:
molFilePath = os.path.join(searchFileDirPath, ccId[0], ccId,
sId + ".sdf")
if not (useCache and mU.exists(molFilePath)):
ok = oeU.write(molFilePath,
oeMol,
constantMol=False,
addSdTags=True)
if ok:
pathTupList.append((sId, molFilePath, "sdf"))
#
if useMol2:
mol2FilePath = os.path.join(searchFileDirPath, ccId[0], ccId,
sId + ".mol2")
if not (useCache and mU.exists(mol2FilePath)):
oeU.write(mol2FilePath,
oeMol,
constantMol=False,
addSdTags=True)
if ok:
pathTupList.append((sId, mol2FilePath, "mol2"))
numMols += 1
#
self.__storePathList(pathTupList)
return numMols
def __reload(self, **kwargs):
"""Reload the dictionary of OE molecules and related data artifacts for chemical component definitions.
Args:
molBuildType (str): coordinates to use in building OE molecules from CIF components (model, ideal or None)
limitPerceptions(bool): process input descriptors in essentially verbatim mode (default: True)
fpTypeList (list): fingerprint type (TREE,PATH,MACCS,CIRCULAR,LINGO)
screenTypeList (list): fast sub search screen type (MOLECULE, SMARTS, MDL, ... )
useCache (bool, optional): flag to use cached files. Defaults to True.
cachePath (str): path to the top cache directory. Defaults to '.'.
numProc (int): number processors to engage in screen substructure search database generation.
molLimit (int, optional): limiting number of molecules in data store (default: 0 no limit)
suppressHydrogens (bool, optional): flag to suppress explicit hydrogens in the OE data store.
Returns:
(dict): dictionary of constructed OE molecules
"""
useCache = kwargs.get("useCache", True)
cachePath = kwargs.get("cachePath", ".")
numProc = kwargs.get("numProc", 2)
molLimit = kwargs.get("molLimit", 0)
fpTypeList = kwargs.get("fpTypeList",
["TREE", "PATH", "MACCS", "CIRCULAR", "LINGO"])
# screenTypeList = kwargs.get("screenTypeList", ["SMARTS"])
screenTypeList = kwargs.get("screenTypeList", [])
molBuildType = kwargs.get("molBuildType", "model-xyz")
limitPerceptions = kwargs.get("limitPerceptions", False)
quietFlag = kwargs.get("quietFlag", True)
suppressHydrogens = kwargs.get("suppressHydrogens", False)
logSizes = kwargs.get("logSizes", False)
fpDbType = "STANDARD"
#
ccCount = 0
oeCount = 0
errCount = 0
failIdList = []
oeIo = OeIoUtils(quietFlag=quietFlag)
# --------
oeMolFilePath = os.path.join(self.__dirPath, self.__getOeMolFileName())
if not useCache or (useCache and not self.__mU.exists(oeMolFilePath)):
cmpKwargs = {
k: v
for k, v in kwargs.items()
if k not in ["cachePath", "useCache", "molLimit"]
}
ccmP = ChemCompMoleculeProvider(cachePath=cachePath,
useCache=True,
molLimit=molLimit,
**cmpKwargs)
ok = ccmP.testCache(minCount=molLimit, logSizes=logSizes)
ccObjD = ccmP.getMolD() if ok else {}
ccCount = len(ccObjD)
# -------
startTime = time.time()
oeCount, errCount, failIdList = oeIo.buildOeBinaryMolCache(
oeMolFilePath,
ccObjD,
molBuildType=molBuildType,
quietFlag=quietFlag,
fpTypeList=fpTypeList,
limitPerceptions=limitPerceptions,
suppressHydrogens=suppressHydrogens)
logger.info(
"Stored %d/%d OeMols (suppressH = %r) created with molBuildType %r (unconverted %d)",
oeCount, ccCount, suppressHydrogens, molBuildType, errCount)
if failIdList:
logger.info("%r failures %r", molBuildType, failIdList)
endTime = time.time()
logger.info("Constructed %d/%d cached oeMols (%.4f seconds)",
oeCount, ccCount, endTime - startTime)
# --------
oeMolDbFilePath = os.path.join(self.__dirPath,
self.__getOeMolDbFileName())
if not useCache or (useCache
and not self.__mU.exists(oeMolDbFilePath)):
startTime = time.time()
molCount = oeIo.createOeBinaryDatabaseAndIndex(
oeMolFilePath, oeMolDbFilePath)
endTime = time.time()
logger.info(
"Created and stored %d indexed OeMols in OE database format (%.4f seconds)",
molCount, endTime - startTime)
# --------
if fpDbType == "FAST":
for fpType in fpTypeList:
startTime = time.time()
# Fast FP search database file names
fpPath = os.path.join(self.__dirPath,
self.__getFastFpDbFileName(fpType))
if not useCache or (useCache and not self.__mU.exists(fpPath)):
ok = oeIo.createOeFingerPrintDatabase(oeMolDbFilePath,
fpPath,
fpType=fpType)
endTime = time.time()
logger.info(
"Created and stored %s fingerprint database (%.4f seconds)",
fpType, endTime - startTime)
# --------
if molBuildType in ["oe-iso-smiles"]:
for screenType in screenTypeList:
startTime = time.time()
fp = os.path.join(self.__dirPath,
self.__getSubSearchFileName(screenType))
if not useCache or (useCache and not self.__mU.exists(fp)):
ok = oeIo.createOeSubSearchDatabase(oeMolFilePath,
fp,
screenType=screenType,
numProc=numProc)
endTime = time.time()
logger.info(
"Constructed screened substructure database (status %r) with screenType %s (%.4f seconds)",
ok, screenType, endTime - startTime)
# ---------
ssDb = oeIo.loadOeSubSearchDatabase(fp,
screenType=screenType,
numProc=numProc)
ok = ssDb.NumMolecules() == oeCount
# ----------
return oeCount
def testSssWithFingerPrintFromDescriptor(self):
oemp = OeMoleculeProvider(**self.__myKwargs)
ok = oemp.testCache()
ccmP = ChemCompIndexProvider(**self.__myKwargs)
ccIdxD = ccmP.getIndex()
ok = ccmP.testCache(minCount=self.__minCount)
self.assertTrue(ok)
limitPerceptions = False
# minFpScore = 0.5
maxFpResults = 50
matchOpts = "graph-relaxed"
numMols = 20
oeioU = OeIoUtils()
oesU = OeSearchUtils(oemp, fpTypeList=self.__fpTypeList)
missTupL = []
missedD = {}
missedFpD = {}
# ----
startTime = time.time()
for ccId, ccD in list(ccIdxD.items())[:numMols]:
for buildType in [
"oe-iso-smiles", "oe-smiles", "acdlabs-smiles",
"cactvs-iso-smiles", "cactvs-smiles", "inchi"
]:
if buildType in ccD:
logger.debug("Search %s %r", ccId, ccD[buildType])
if buildType in ["inchi"]:
oemf = OeMoleculeFactory()
oemf.setDescriptor(ccD["inchi"], "inchi", ccId)
ok = oemf.build(molBuildType="inchi",
limitPerceptions=limitPerceptions)
if not ok:
logger.info("%s build failed with InChI %r", ccId,
ccD["inchi"])
else:
oeMol = oemf.getMol()
if oemf.getInChI() != ccD["inchi"]:
logger.info(
"%s regenerated InChI differs\n%r\n%s",
ccId, ccD["inchi"], oemf.getInChI())
else:
oeMol = oeioU.smilesToMol(
ccD[buildType], limitPerceptions=limitPerceptions)
if not oeMol:
continue
maxHits = 0
minHits = maxFpResults
selfHit = False
for fpType, minFpScore in self.__fpTypeCuttoffList:
retStatus, mL = oesU.searchSubStructureWithFingerPrint(
oeMol,
fpType,
minFpScore,
maxFpResults,
matchOpts=matchOpts)
self.assertTrue(retStatus)
logger.debug("%s fpType %r hits %d", ccId, fpType,
len(mL))
maxHits = max(maxHits, len(mL))
minHits = min(minHits, len(mL))
matchedSelf = self.__resultContains(ccId, mL)
selfHit = selfHit or matchedSelf
if not matchedSelf:
missedFpD.setdefault(ccId, []).append(
(buildType, fpType, len(mL)))
if not selfHit:
missedD.setdefault(ccId, []).append(buildType)
logger.info("%s (%r) buildType %r min hits %d max hits %d",
ccId, selfHit, buildType, minHits, maxHits)
else:
logger.info("%s missing descriptor %r", ccId, buildType)
#
for ccId, missL in missedD.items():
logger.info("%s missed list %r", ccId, missL)
if ccId in missedFpD:
logger.info("%s unmatched for fpTypes %r", ccId,
missedFpD[ccId])
# ----
doDepict = False
if doDepict:
mD = {}
for missTup in missTupL:
mD.setdefault(missTup[0], []).append(missTup[1])
for ccId, buildTypeL in mD.items():
idxD = ccIdxD[ccId]
if "oe-iso-smiles" in idxD:
for buildType in buildTypeL:
self.__displayAlignedDescriptorPair(
ccId,
idxD["oe-iso-smiles"],
"oe-iso-smiles",
idxD[buildType],
buildType,
title=None,
limitPerceptions=True)
logger.info("%s fingerprints search on %d in (%.4f seconds)",
len(self.__fpTypeList), numMols,
time.time() - startTime)
def __getMiscFile(self, filePath, suppressHydrogens=False, importType="2D", title=None, largestPart=False):
"""Fetch a miscellaneous chemical file (ccPath) and build OE molecules
for comparison.
"""
try:
oeioU = OeIoUtils()
oeMolL = oeioU.fileToMols(filePath, use3D=importType == "3D", largestPart=largestPart)
logger.info("Read (%d) from %s ", len(oeMolL), filePath)
oeMol = oeMolL[0]
ccId = title if title else oeMol.GetTitle()
if title:
oeMol.SetTitle(ccId)
#
oemf = OeMoleculeFactory()
if not self.__verbose:
oemf.setQuiet()
oemf.setOeMol(oeMol, ccId)
#
fD = oemf.getOeMoleculeFeatures()
if self.__verbose:
logger.info(" Title = %s", title)
logger.info(" Title OEMF = %s", oemf.getTitle())
logger.info(" SMILES = %s", oemf.getCanSMILES())
logger.info(" SMILES (stereo) = %s", oemf.getIsoSMILES())
logger.info(" Formula (Hill) = %s", oemf.getFormula())
logger.info(" InChI key = %s", oemf.getInChIKey())
logger.info(" InChI = %s", oemf.getInChI())
#
ccId = oemf.getTitle()
if suppressHydrogens:
tMol = oemf.getGraphMolSuppressH()
else:
tMol = oemf.getMol()
molXyzL = []
if importType == "3D":
for atm in tMol.GetAtoms():
xyzL = oechem.OEFloatArray(3)
tMol.GetCoords(atm, xyzL)
molXyzL.append(
ComponentAtomDetails(
atIdx=atm.GetIdx(),
atNo=atm.GetAtomicNum(),
atName=atm.GetName(),
atType=atm.GetType(),
x=xyzL[0],
y=xyzL[1],
z=xyzL[2],
atFormalCharge=atm.GetFormalCharge(),
)
)
fD = {}
fD = {
"Formula": oemf.getFormula(),
"SMILES": oemf.getCanSMILES(),
"SMILES_STEREO": oemf.getIsoSMILES(),
"InChI": oemf.getInChI(),
"InChIKey": oemf.getInChIKey(),
"xyz": molXyzL,
}
for atm in tMol.GetAtoms():
xyzL = oechem.OEFloatArray(3)
tMol.GetCoords(atm, xyzL)
if self.__verbose:
logger.debug("atom %s %s %s %s %r", atm.GetIdx(), atm.GetAtomicNum(), atm.GetName(), atm.GetType(), xyzL)
fD["OEMOL"] = tMol
return (ccId, tMol, fD)
except Exception as e:
logger.exception("Failing with %s", str(e))
return None, None, None
