def __fingerPrintScores(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)
#
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 (
ii,
ccId,
) in enumerate(list(ccIdxD.keys())[: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 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", ii, numMols,
len(missedBuildD))
# ------
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)
return True
def testFingerPrintScores(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
maxResults = 100
numMols = 20
oeioU = OeIoUtils()
oesU = OeSearchUtils(oemp, fpTypeList=self.__fpTypeList)
missedFpD = {}
missedBuildD = {}
# ----
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"]:
# oeMol = oeioU.inchiToMol(ccD[buildType], limitPerceptions=limitPerceptions)
#
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 = maxResults
selfHit = False
#
for fpType, minFpScore in self.__fpTypeCuttoffList:
retStatus, mL = oesU.getFingerPrintScores(
oeMol, fpType, minFpScore, maxResults)
self.assertTrue(retStatus)
logger.info("%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)
logger.info("%s buildType %r min hits %d max hits %d",
ccId, buildType, minHits, maxHits)
else:
logger.info("%s missing descriptor %r", ccId, buildType)
#
for ccId, bTL in missedBuildD.items():
logger.info("%s missed build type list %r", ccId, bTL)
if ccId in missedFpD:
logger.info("%s unmatched fpTypes %r", ccId, missedFpD[ccId])
#
doDepict = False
if doDepict:
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(self.__fpTypeList), numMols,
time.time() - startTime)