def testRoundTripOps(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)
tMolL = oeIoU.fileToMols(mol2Path)
#
nextMol2Path = os.path.join(self.__molfileDirPath,
sId + "-next.mol2")
oeIoU.write(nextMol2Path,
tMolL[0],
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)
#
tMolL = oeIoU.fileToMols(sdfPath)
nextSdfPath = os.path.join(self.__molfileDirPath,
sId + "-next.sdf")
oeIoU.write(nextSdfPath,
tMolL[0],
constantMol=True,
addSdTags=True)
# ----
except Exception as e:
logger.exception("Failing with %s", str(e))
self.fail()
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
def matchByDescriptor(self,
descriptor,
descriptorType,
matchOpts="graph-relaxed",
searchId=None):
"""Return graph match (w/ finger print pre-filtering) and finger print search results for the
input desriptor.
Args:
descriptor (str): molecular descriptor (SMILES, InChI)
descriptorType (str): descriptor type (SMILES, InChI
matchOpts (str, optional): graph match criteria (graph-relaxed, graph-relaxed-stereo, graph-strict,
fingerprint-similarity, Defaults to "graph-relaxed")
searchId (str, optional): search identifier for logging. Defaults to None.
Returns:
(statusCode, list, list): status, graph match and finger match lists of type (MatchResults)
-100 descriptor processing error
-200 search execution error
0 search execution success
"""
ssL = fpL = []
retStatus = False
statusCode = -200
try:
fpTypeCuttoffD = self.__configD["oesmpKwargs"][
"fpTypeCuttoffD"] if "fpTypeCuttoffD" in self.__configD[
"oesmpKwargs"] else {}
maxFpResults = self.__configD["oesmpKwargs"][
"maxFpResults"] if "maxFpResults" in self.__configD[
"oesmpKwargs"] else 50
limitPerceptions = self.__configD["oesmpKwargs"][
"limitPerceptions"] if "limitPerceptions" in self.__configD[
"oesmpKwargs"] else False
#
searchId = searchId if searchId else "query"
messageTag = searchId + ":" + descriptorType
oeioU = OeIoUtils()
oeMol = oeioU.descriptorToMol(descriptor,
descriptorType,
limitPerceptions=limitPerceptions,
messageTag=messageTag)
oeMol = oeioU.suppressHydrogens(oeMol)
if not oeMol:
logger.warning("descriptor type %r molecule build fails: %r",
descriptorType, descriptor)
return self.__statusDescriptorError, ssL, fpL
#
retStatus, ssL, fpL = self.__oesU.searchSubStructureAndFingerPrint(
oeMol,
list(fpTypeCuttoffD.items())[:2],
maxFpResults,
matchOpts=matchOpts)
statusCode = 0 if retStatus else self.__searchError
except Exception as e:
logger.exception("Failing with %s", str(e))
#
return statusCode, ssL, fpL
def __subStructureSearchScreened(self, numMols, **kwargs):
#
buildTypeList = kwargs.get("buildTypeList", ["oe-iso-smiles"])
screenTypeList = kwargs.get("screenTypeList", ["SMARTS"])
oesmP, ccIdxD = self.__getSearchDataProviders(**kwargs)
for screenType in screenTypeList:
oesU = OeSearchUtils(oesmP,
screenType=screenType,
numProc=self.__numProc)
oeioU = OeIoUtils()
#
missL = []
numMols = min(len(ccIdxD), numMols) if numMols else len(ccIdxD)
for (
ii,
ccId,
) in enumerate(list(ccIdxD.keys())[:numMols]):
ccD = ccIdxD[ccId]
for buildType in buildTypeList:
if buildType in ccD:
if screenType == "SMARTS":
smiles = oeioU.descriptorToSmiles(ccD[buildType],
buildType,
messageTag=ccId +
":" + buildType)
oeQMol = oeioU.descriptorToMol(smiles,
"SMARTS",
messageTag=ccId +
":" + buildType)
else:
oeQMol = oeioU.descriptorToQMol(ccD[buildType],
"SMARTS",
messageTag=ccId +
":" + buildType)
if not oeQMol:
logger.debug("%s build failed for %s - skipping",
ccId, buildType)
continue
# ----
startTime = time.time()
retStatus, mL = oesU.searchSubStructureScreened(
oeQMol, maxMatches=100)
if retStatus:
logger.debug(
"%s - %s - %s (status=%r) match length %d in (%.4f seconds)",
ccId, buildType, screenType, retStatus,
len(mL),
time.time() - startTime)
if not self.__resultContains(ccId, mL):
missL.append((ccId, buildType, screenType))
# ----
if ii % 100 == 0:
logger.info("Completed %d of %d missed count %d", ii,
numMols, len(missL))
logger.info("Screen %r missed searches (%d) %r", screenType,
len(missL), missL)
return True
def subStructSearchByDescriptor(self,
descriptor,
descriptorType,
matchOpts="sub-struct-graph-relaxed",
searchId=None):
"""Return graph match (w/ finger print pre-filtering) and finger print search results for the
input desriptor.
Args:
descriptor (str): molecular descriptor (SMILES, InChI)
descriptorType (str): descriptor type (SMILES, InChI)
matchOpts (str, optional): graph match criteria (sub-struct-graph-relaxed, sub-struct-graph-relaxed-stereo,
sub-struct-graph-strict). Defaults to "sub-struct-graph-relaxed".
searchId (str, optional): search identifier for logging. Defaults to None.
Returns:
(statusCode, list, list): status, substructure search results of type (MatchResults), empty list placeholder
-100 descriptor processing error
-200 search execution error
0 search execution success
"""
ssL = []
retStatus = False
statusCode = -200
try:
limitPerceptions = self.__configD["oesmpKwargs"][
"limitPerceptions"] if "limitPerceptions" in self.__configD[
"oesmpKwargs"] else False
numProc = self.__configD["oesmpKwargs"][
"numProc"] if "numProc" in self.__configD["oesmpKwargs"] else 4
#
searchId = searchId if searchId else "query"
messageTag = searchId + ":" + descriptorType
oeioU = OeIoUtils()
oeMol = oeioU.descriptorToMol(descriptor,
descriptorType,
limitPerceptions=limitPerceptions,
messageTag=messageTag)
oeMol = oeioU.suppressHydrogens(oeMol)
if not oeMol:
logger.warning("descriptor type %r molecule build fails: %r",
descriptorType, descriptor)
return self.__statusDescriptorError, ssL, []
#
ccIdL = self.__oesubsU.prefilterIndex(oeMol,
self.__siIdxP,
matchOpts=matchOpts)
retStatus, ssL = self.__oesubsU.searchSubStructure(
oeMol, ccIdList=ccIdL, matchOpts=matchOpts, numProc=numProc)
statusCode = 0 if retStatus else self.__searchError
except Exception as e:
logger.exception("Failing with %s", str(e))
#
return statusCode, ssL, []
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 __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 __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 __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 __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 __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)
