def testSubStructureSearchWithFingerPrint(self):
oemp = OeMoleculeProvider(**self.__myKwargs)
#
ok = oemp.testCache()
ccmP = ChemCompIndexProvider(**self.__myKwargs)
ccIdxD = ccmP.getIndex()
ok = ccmP.testCache(minCount=self.__minCount)
self.assertTrue(ok)
minFpScore = 0.40
maxFpResults = 50
numMols = 20
matchOpts = "graph-relaxed"
oesU = OeSearchUtils(oemp, fpTypeList=self.__fpTypeList)
# ----
startTime = time.time()
for ccId, _ in list(ccIdxD.items())[:numMols]:
for fpType in self.__fpTypeList:
oeMol = oemp.getMol(ccId)
retStatus, mL = oesU.searchSubStructureWithFingerPrint(
oeMol,
fpType,
minFpScore,
maxFpResults,
matchOpts=matchOpts)
self.assertTrue(retStatus)
self.assertTrue(self.__resultContains(ccId, mL))
logger.info("%s fingerprints search on %d in (%.4f seconds)",
len(self.__fpTypeList), numMols,
time.time() - startTime)
def testFingerPrintSearch(self):
oemp = OeMoleculeProvider(**self.__myKwargs)
# This will reload the oe binary cache.
oeMol = oemp.getMol("004")
self.assertGreaterEqual(len(list(oeMol.GetAtoms())), 12)
#
ok = oemp.testCache()
ccmP = ChemCompIndexProvider(**self.__myKwargs)
ccIdxD = ccmP.getIndex()
ok = ccmP.testCache(minCount=self.__minCount)
self.assertTrue(ok)
minScore = 0.50
maxResults = 50
numMols = 50
oesU = OeSearchUtils(oemp, fpTypeList=self.__fpTypeList)
# ----
startTime = time.time()
for ccId, _ in list(ccIdxD.items())[:numMols]:
for fpType in self.__fpTypeList:
oeMol = oemp.getMol(ccId)
retStatus, mL = oesU.searchFingerPrints(
oeMol,
fpType=fpType,
minFpScore=minScore,
maxFpResults=maxResults)
self.assertTrue(retStatus)
self.assertTrue(self.__resultContains(ccId, mL))
# self.assertGreaterEqual(len(mL), 1)
logger.info("%s fingerprints search on %d in (%.4f seconds)",
len(self.__fpTypeList), numMols,
time.time() - startTime)
def updateDescriptors(self, useCache=True):
ccidxP = ChemCompIndexProvider(
ccUrlTarget=self.__ccUrlTarget,
birdUrlTarget=self.__birdUrlTarget,
cachePath=self.__cachePath,
useCache=useCache,
molLimit=None,
ccFileNamePrefix=self.__ccFileNamePrefix,
)
ok = ccidxP.testCache()
if ok:
ccIdList = ccidxP.getIdList()
curIdList = list(self.__descrD.keys())
updIdList = list(set(ccIdList) - set(curIdList))
if updIdList:
logger.info(
"Updating Chemaxon descriptors for (%d) components",
len(updIdList))
uD = self.__fetchDescriptors(updIdList,
ccidxP,
chunkSize=self.__chunkSize)
self.__descrD.update(uD)
descrFilePath = self.getIndexFilePath()
tS = datetime.datetime.now().isoformat()
vS = datetime.datetime.now().strftime("%Y-%m-%d")
self.__version = vS
dD = {"created": tS, "version": vS, "smiles": self.__descrD}
ok = self.__mU.doExport(descrFilePath,
dD,
fmt="json",
indent=3)
#
return ok
def __getSearchDataProviders(self, **kwargs):
oesmP = OeSearchMoleculeProvider(**kwargs)
ok = oesmP.testCache()
ccIdxP = ChemCompIndexProvider(**kwargs)
ok = ccIdxP.testCache()
self.assertTrue(ok)
ccIdxD = ccIdxP.getIndex()
return oesmP, ccIdxD
def testSubStructureSearch(self):
oemp = OeMoleculeProvider(**self.__myKwargs)
ok = oemp.testCache()
ccmP = ChemCompIndexProvider(**self.__myKwargs)
ccIdxD = ccmP.getIndex()
ok = ccmP.testCache(minCount=self.__minCount)
self.assertTrue(ok)
oesU = OeSearchUtils(oemp, fpTypeList=self.__fpTypeList)
numMols = 10
for ccId, _ in list(ccIdxD.items())[:numMols]:
# ----
startTime = time.time()
oeMol = oemp.getMol(ccId)
retStatus, mL = oesU.searchSubStructure(oeMol, matchOpts="relaxed")
logger.info("%s match length %d in (%.4f seconds)", ccId, len(mL),
time.time() - startTime)
self.assertTrue(retStatus)
self.assertTrue(self.__resultContains(ccId, mL))
def __testBuildMoleculeCacheFiles(self, **kwargs):
"""Test build chemical component cache files from the input component dictionaries"""
molLimit = kwargs.get("molLimit", None)
useCache = kwargs.get("useCache", True)
logSizes = kwargs.get("logSizes", False)
ccFileNamePrefix = kwargs.get("ccFileNamePrefix", "cc")
ccUrlTarget = kwargs.get("ccUrlTarget", None)
birdUrlTarget = kwargs.get("birdUrlTarget", None)
#
ccidxP = ChemCompIndexProvider(ccUrlTarget=ccUrlTarget,
birdUrlTarget=birdUrlTarget,
cachePath=self.__cachePath,
useCache=useCache,
molLimit=molLimit,
ccFileNamePrefix=ccFileNamePrefix)
ok = ccidxP.testCache(minCount=molLimit, logSizes=logSizes)
self.assertTrue(ok)
logger.info(" ******* Completed operation ******** ")
#
return ccidxP
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 __getChemCompDefs(self):
ccMolD = {}
ccIdxD = {}
try:
ccmP = ChemCompMoleculeProvider(**self.__myKwargs)
ok = ccmP.testCache()
ccMolD = ccmP.getMolD()
ccmP = ChemCompIndexProvider(**self.__myKwargs)
ccIdxD = ccmP.getIndex()
ok = ccmP.testCache(minCount=10)
self.assertTrue(ok)
except Exception as e:
logger.exception("Failing with %s", str(e))
return ccMolD, ccIdxD
def __buildChemCompIndex(self, **kwargs):
"""Build chemical component cache files from the input component dictionaries"""
try:
molLimit = kwargs.get("molLimit", None)
useCache = not kwargs.get("rebuildChemIndices", False)
logSizes = kwargs.get("logSizes", False)
ccFileNamePrefix = kwargs.get("ccFileNamePrefix", "cc-full")
ccUrlTarget = kwargs.get("ccUrlTarget", None)
birdUrlTarget = kwargs.get("birdUrlTarget", None)
cachePath = kwargs.get("cachePath", self.__cachePath)
#
ccidxP = ChemCompIndexProvider(ccUrlTarget=ccUrlTarget,
birdUrlTarget=birdUrlTarget,
cachePath=cachePath,
useCache=useCache,
molLimit=molLimit,
ccFileNamePrefix=ccFileNamePrefix)
ok = ccidxP.testCache(minCount=molLimit, logSizes=logSizes)
return ok, ccidxP if ok else None
except Exception as e:
logger.exception("Failing with %s", str(e))
#
return False, None
def buildDescriptors(self):
descrFilePath = self.getIndexFilePath()
ccidxP = ChemCompIndexProvider(
ccUrlTarget=self.__ccUrlTarget,
birdUrlTarget=self.__birdUrlTarget,
cachePath=self.__cachePath,
useCache=True,
molLimit=self.__molLimit,
ccFileNamePrefix=self.__ccFileNamePrefix,
)
ok = ccidxP.testCache()
if ok:
ccIdList = ccidxP.getIdList()
self.__descrD = self.__fetchDescriptors(ccIdList,
ccidxP,
chunkSize=self.__chunkSize)
tS = datetime.datetime.now().isoformat()
vS = datetime.datetime.now().strftime("%Y-%m-%d")
self.__version = vS
dD = {"created": tS, "version": vS, "smiles": self.__descrD}
ok = self.__mU.doExport(descrFilePath, dD, fmt="json", indent=3)
logger.info("Stored %s descriptors for %d components (status=%r) ",
descrFilePath, len(self.__descrD), ok)
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)