def __displayAlignedDescriptorPair(self,
ccId,
descrRef,
buildTypeRef,
descrFit,
buildTypeFit,
title=None,
limitPerceptions=True):
oemfRef = OeMoleculeFactory()
oemfRef.setDescriptor(descrRef, buildTypeRef, ccId)
oemfRef.build(molBuildType=buildTypeRef,
limitPerceptions=limitPerceptions)
oeMolRef = oemfRef.getMol()
#
oemfFit = OeMoleculeFactory()
oemfFit.setDescriptor(descrFit, buildTypeFit, ccId)
oemfFit.build(molBuildType=buildTypeFit,
limitPerceptions=limitPerceptions)
oeMolFit = oemfFit.getMol()
#
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 buildOeBinaryMolCacheFromIndex(self,
filePath,
ccIdxD,
quietFlag=False,
fpTypeList=None,
limitPerceptions=False,
suppressHydrogens=False):
"""Build cache of OEGraphMol() objects from the input chemical component search index.
Args:
filePath (str): output cache file path
ccIdxD (dict): search index dictionary
quietFlag (bool, optional): suppress OE output. Defaults to False.
fpTypeList (list, optional): list of fingerprint types. Defaults to None.
limitPerceptions (bool, optional): suppress automatic chemical perceptions. Defaults to False.
suppressHydrogens (bool, optional): suppress explicit hydrogen count. Defaults to False.
Returns:
(int, int, list): chem comp success count, error count, chem comp identifier failure list
"""
failIdList = []
ccCount = 0
errCount = 0
startTime = time.time()
try:
ofs = oechem.oemolostream()
ofs.SetFormat(oechem.OEFormat_OEB)
if ofs.open(filePath):
oemf = OeMoleculeFactory()
if quietFlag:
oemf.setQuiet()
for searchCcId, ccIdx in ccIdxD.items():
oemf.setDescriptor(ccIdx["smiles"], "oe-iso-smiles",
searchCcId)
ok = oemf.build(molBuildType="oe-iso-smiles",
limitPerceptions=limitPerceptions)
if ok and fpTypeList:
fpOk = oemf.addFingerPrints(fpTypeList)
if not fpOk:
logger.info("Fingerprint generation fails for %r",
searchCcId)
if ok:
if not suppressHydrogens:
oemf.addExplicitHydrogens()
oemf.setSimpleAtomNames()
oeMol = oemf.getMol(
suppressHydrogens=suppressHydrogens)
oechem.OEWriteMolecule(ofs, oeMol)
ccCount += 1
if not ok:
# build failed incomplete component (e.g. missing atoms or bonds)
errCount += 1
failIdList.append(searchCcId)
else:
logger.error("Unable to open cache database %s", filePath)
errCount += 1
except Exception as e:
logger.exception("Failing with %s", str(e))
#
endTime = time.time()
logger.info("Completed operation at %s (%.4f seconds)",
time.strftime("%Y %m %d %H:%M:%S", time.localtime()),
endTime - startTime)
return ccCount, errCount, failIdList
def testCompareByBuildType(self):
"""Compare depictions constructed molecules with various builds from chemical defintions -
all build types 8769 (all)
connect - smiles 6743
model vs iso smiles 5937
ideal va iso smiles 7047
"""
doDepict = False
ccResultD = {}
genResultD = {}
smilesByBuildTypeD = {}
try:
ccMolD, ccIdxD = self.__getChemCompDefs()
#
limitPerceptions = True
# molBuildTypeL = ["model-xyz", "ideal-xyz", "connection-table", "oe-iso-smiles"]
molBuildTypeL = ["ideal-xyz", "oe-iso-smiles"]
#
startTime = time.time()
oefm = OeMoleculeFactory()
oefm.setQuiet()
for molBuildType in molBuildTypeL:
for ccId, idxD in ccIdxD.items():
ccObj = ccMolD[ccId]
# ----
ccIsoSmiles = idxD["oe-iso-smiles"]
ccSmiles = idxD["oe-smiles"]
# ----
tId = oefm.setChemCompDef(ccObj)
if not tId:
logger.info("Skipping bad component %r", ccId)
continue
self.assertEqual(tId, ccId)
ok = oefm.build(molBuildType=molBuildType,
limitPerceptions=limitPerceptions)
if not ok:
logger.info("Build using %r failed for %s",
molBuildType, ccId)
continue
# ------
oeMol = oefm.getGraphMol()
oeIsoSmiles = oefm.getIsoSMILES()
oeSmiles = oefm.getCanSMILES()
ccEq = oeIsoSmiles == ccIsoSmiles and oeSmiles == ccSmiles
#
oefmR = OeMoleculeFactory()
oefmR.setQuiet()
ccIdGen = ccId + "_gen"
oefmR.setDescriptor(oeIsoSmiles, "oe-iso-smiles", ccIdGen)
ok = oefmR.build(molBuildType="oe-iso-smiles",
limitPerceptions=limitPerceptions)
if not ok:
logger.info("Build using %r failed for %s",
molBuildType, ccIdGen)
continue
# ------
#
# oeMolGen = oefmR.getGraphMol()
oeIsoSmilesGen = oefmR.getIsoSMILES()
oeSmilesGen = oefmR.getCanSMILES()
genEq = oeIsoSmiles == oeIsoSmilesGen and oeSmiles == oeSmilesGen
smilesByBuildTypeD.setdefault(ccId, {}).setdefault(
molBuildType, []).append(oeIsoSmilesGen)
#
logger.debug("%s buildType %s ccEq %r genEq %r", ccId,
molBuildType, ccEq, genEq)
if not ccEq:
ccResultD.setdefault(molBuildType, []).append(ccId)
if not genEq:
genResultD.setdefault(molBuildType, []).append(ccId)
if doDepict:
pS = "-limited" if limitPerceptions else ""
imagePath = os.path.join(
self.__workPath,
ccId + "-%s%s.svg" % (molBuildType, pS))
oed = OeDepict()
title = ""
oed.setMolTitleList([(ccId, oeMol, title)])
oed.setDisplayOptions(labelAtomName=False,
labelAtomCIPStereo=True,
labelAtomIndex=False,
labelBondIndex=False,
cellBorders=False,
bondDisplayWidth=0.5)
oed.setGridOptions(rows=1, cols=1)
oed.prepare()
oed.write(imagePath)
logger.info(
"Completed comparing %d molecules in %d builds (%.4f seconds)",
len(ccIdxD), len(molBuildTypeL),
time.time() - startTime)
#
#
for molBuildType in molBuildTypeL:
if molBuildType in genResultD:
logger.info("GEN %s (%d) %r", molBuildType,
len(genResultD[molBuildType]),
genResultD[molBuildType])
numDiff = 0
for ccId, btD in smilesByBuildTypeD.items():
tS = set()
for molBuildType, sL in btD.items():
tS.add(sL[0])
if len(tS) > 1:
numDiff += 1
logger.debug("%s diff smiles (%d) %r", ccId, len(tS), tS)
logger.info("Components with inconsistent SMILES %d", numDiff)
#
except Exception as e:
logger.exception("Failing with %s", str(e))
self.fail()
def testRoundTrip(self):
"""Round trip smiles comparisons -"""
try:
ccMolD = self.__getChemCompDefs()
# useCache = True
# quietFlag = False
# molBuildTypeL = ["model-xyz", "ideal-xyz", None]
# molBuildTypeL = [None]
buildTypeRef = "oe-iso-smiles"
oemf1 = OeMoleculeFactory()
oemf2 = OeMoleculeFactory()
#
for ccId, ccObj in ccMolD.items():
# ----
ccIt = iter(PdbxChemCompIt(ccObj))
cc = next(ccIt)
formula = cc.getFormulaWithCharge()
# ccId = cc.getId()
ccName = cc.getName()
ifCharge = cc.getFormalChargeAsInt()
isAmbiguous = cc.getAmbiguousFlag() in ["Y", "y"]
isCurrent = cc.getReleaseStatus() in ["REL"]
logger.debug("%s name %r formula %r charge %d", ccId, ccName,
formula, ifCharge)
# ----
ccId = oemf1.setChemCompDef(ccObj)
ok = oemf1.build(molBuildType=buildTypeRef,
limitPerceptions=False)
if not ok:
logger.info(
"Build using %r failed for %s (ambiguous flag %r current %r)",
buildTypeRef, ccId, isAmbiguous, isCurrent)
#
isDiff = False
#
if isDiff:
genIsoSmi = oemf1.getCanSMILES()
oemf2 = OeMoleculeFactory()
oemf2.setDescriptor(genIsoSmi, "oe-iso-smiles", ccId)
oemf2.build(molBuildType="oe-iso-smiles",
limitPerceptions=False)
regenIsoSmi = oemf2.getIsoSMILES()
if genIsoSmi != regenIsoSmi:
logger.info(
"%s regenerated ISOSMILES differ \n -- INP: %s\n -- OUT: %s",
ccId, genIsoSmi, regenIsoSmi)
oed = OeDepictMCSAlignPage()
oed.setDisplayOptions(
labelAtomName=True,
labelAtomCIPStereo=True,
labelAtomIndex=False,
labelBondIndex=False,
labelBondCIPStereo=True,
highlightStyleFit="ballAndStickInverse",
highLightNotMatchColorRef="pink",
bondDisplayWidth=0.5,
)
oed.setRefMol(oemf1.getGraphMol(), ccId)
oed.setFitMol(oemf2.getGraphMol(), ccId)
imgPath = os.path.join(
self.__workPath,
"compare-assigned-" + ccId + "-calc-" + ccId + ".svg")
logger.info("Using image path %r", imgPath)
aML = oed.alignPair(imagePath=imgPath)
if aML:
for (rCC, rAt, tCC, tAt) in aML:
logger.info("%5s %-5s %5s %-5s", rCC, rAt, tCC,
tAt)
else:
logger.debug("%s matched all cases", ccId)
except Exception as e:
logger.exception("Failing with %s", str(e))
self.fail()
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)