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 __fingerPrintSearch(self, numMols, **kwargs):
maxFpResults = kwargs.get("maxFpResults", 50)
limitPerceptions = kwargs.get("limitPerceptions", False)
fpTypeCuttoffList = kwargs.get("fpTypeCuttoffList", [("TREE", 0.6)])
buildTypeList = kwargs.get("buildTypeList", ["oe-iso-smiles"])
#
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 search on %d molecules", numMols)
# ----
startTime = time.time()
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
selfHit = False
for fpType, minFpScore in fpTypeCuttoffList:
retStatus, mL = oesU.searchFingerPrints(
oeMol,
fpType=fpType,
minFpScore=minFpScore,
maxFpResults=maxFpResults)
self.assertTrue(retStatus)
#
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)
# ------
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])
# ----
logger.info("%s fingerprints search on %d in (%.4f seconds)",
len(fpTypeCuttoffList), numMols,
time.time() - startTime)
# ----
return True