def getSubSearchDb(self,
screenType="SMARTS",
numProc=1,
forceRefresh=False):
if not self.__ssDb or forceRefresh:
oeIo = OeIoUtils()
fp = os.path.join(self.__dirPath,
self.__getSubSearchFileName(screenType))
logger.info("Opening screened substructure search database %r", fp)
self.__ssDb = oeIo.loadOeSubSearchDatabase(fp,
screenType,
numProc=numProc)
return self.__ssDb
def __reload(self, **kwargs):
"""Reload the dictionary of OE molecules and related data artifacts for chemical component definitions.
Args:
molBuildType (str): coordinates to use in building OE molecules from CIF components (model, ideal or None)
limitPerceptions(bool): process input descriptors in essentially verbatim mode (default: True)
fpTypeList (list): fingerprint type (TREE,PATH,MACCS,CIRCULAR,LINGO)
screenTypeList (list): fast sub search screen type (MOLECULE, SMARTS, MDL, ... )
useCache (bool, optional): flag to use cached files. Defaults to True.
cachePath (str): path to the top cache directory. Defaults to '.'.
numProc (int): number processors to engage in screen substructure search database generation.
molLimit (int, optional): limiting number of molecules in data store (default: 0 no limit)
suppressHydrogens (bool, optional): flag to suppress explicit hydrogens in the OE data store.
Returns:
(dict): dictionary of constructed OE molecules
"""
useCache = kwargs.get("useCache", True)
cachePath = kwargs.get("cachePath", ".")
numProc = kwargs.get("numProc", 2)
molLimit = kwargs.get("molLimit", 0)
fpTypeList = kwargs.get("fpTypeList",
["TREE", "PATH", "MACCS", "CIRCULAR", "LINGO"])
# screenTypeList = kwargs.get("screenTypeList", ["SMARTS"])
screenTypeList = kwargs.get("screenTypeList", [])
molBuildType = kwargs.get("molBuildType", "model-xyz")
limitPerceptions = kwargs.get("limitPerceptions", False)
quietFlag = kwargs.get("quietFlag", True)
suppressHydrogens = kwargs.get("suppressHydrogens", False)
logSizes = kwargs.get("logSizes", False)
fpDbType = "STANDARD"
#
ccCount = 0
oeCount = 0
errCount = 0
failIdList = []
oeIo = OeIoUtils(quietFlag=quietFlag)
# --------
oeMolFilePath = os.path.join(self.__dirPath, self.__getOeMolFileName())
if not useCache or (useCache and not self.__mU.exists(oeMolFilePath)):
cmpKwargs = {
k: v
for k, v in kwargs.items()
if k not in ["cachePath", "useCache", "molLimit"]
}
ccmP = ChemCompMoleculeProvider(cachePath=cachePath,
useCache=True,
molLimit=molLimit,
**cmpKwargs)
ok = ccmP.testCache(minCount=molLimit, logSizes=logSizes)
ccObjD = ccmP.getMolD() if ok else {}
ccCount = len(ccObjD)
# -------
startTime = time.time()
oeCount, errCount, failIdList = oeIo.buildOeBinaryMolCache(
oeMolFilePath,
ccObjD,
molBuildType=molBuildType,
quietFlag=quietFlag,
fpTypeList=fpTypeList,
limitPerceptions=limitPerceptions,
suppressHydrogens=suppressHydrogens)
logger.info(
"Stored %d/%d OeMols (suppressH = %r) created with molBuildType %r (unconverted %d)",
oeCount, ccCount, suppressHydrogens, molBuildType, errCount)
if failIdList:
logger.info("%r failures %r", molBuildType, failIdList)
endTime = time.time()
logger.info("Constructed %d/%d cached oeMols (%.4f seconds)",
oeCount, ccCount, endTime - startTime)
# --------
oeMolDbFilePath = os.path.join(self.__dirPath,
self.__getOeMolDbFileName())
if not useCache or (useCache
and not self.__mU.exists(oeMolDbFilePath)):
startTime = time.time()
molCount = oeIo.createOeBinaryDatabaseAndIndex(
oeMolFilePath, oeMolDbFilePath)
endTime = time.time()
logger.info(
"Created and stored %d indexed OeMols in OE database format (%.4f seconds)",
molCount, endTime - startTime)
# --------
if fpDbType == "FAST":
for fpType in fpTypeList:
startTime = time.time()
# Fast FP search database file names
fpPath = os.path.join(self.__dirPath,
self.__getFastFpDbFileName(fpType))
if not useCache or (useCache and not self.__mU.exists(fpPath)):
ok = oeIo.createOeFingerPrintDatabase(oeMolDbFilePath,
fpPath,
fpType=fpType)
endTime = time.time()
logger.info(
"Created and stored %s fingerprint database (%.4f seconds)",
fpType, endTime - startTime)
# --------
if molBuildType in ["oe-iso-smiles"]:
for screenType in screenTypeList:
startTime = time.time()
fp = os.path.join(self.__dirPath,
self.__getSubSearchFileName(screenType))
if not useCache or (useCache and not self.__mU.exists(fp)):
ok = oeIo.createOeSubSearchDatabase(oeMolFilePath,
fp,
screenType=screenType,
numProc=numProc)
endTime = time.time()
logger.info(
"Constructed screened substructure database (status %r) with screenType %s (%.4f seconds)",
ok, screenType, endTime - startTime)
# ---------
ssDb = oeIo.loadOeSubSearchDatabase(fp,
screenType=screenType,
numProc=numProc)
ok = ssDb.NumMolecules() == oeCount
# ----------
return oeCount
def __reload(self, **kwargs):
"""Reload the dictionary of OE molecules and related data artifacts for chemical component definitions.
Args:
limitPerceptions(bool): process input descriptors in essentially verbatim mode (default: True)
fpTypeList (list): fingerprint type (TREE,PATH,MACCS,CIRCULAR,LINGO)
screenTypeList (list): fast sub search screen type (MOLECULE, SMARTS, MDL, ... )
useCache (bool, optional): flag to use cached files. Defaults to True.
cachePath (str): path to the top cache directory. Defaults to '.'.
numProc (int): number processors to engage in screen substructure search database generation.
suppressHydrogens (bool, optional): flag to suppress explicit hydrogens in the OE data store.
molLimit (int):
Returns:
(bool) : True for success or False othewise
"""
try:
useCache = kwargs.get("useCache", True)
cachePath = kwargs.get("cachePath", ".")
numProc = kwargs.get("numProc", 2)
molLimit = kwargs.get("molLimit", None)
fpTypeList = kwargs.get(
"fpTypeList", ["TREE", "PATH", "MACCS", "CIRCULAR", "LINGO"])
# screenTypeList = kwargs.get("screenTypeList", ["SMARTS"])
screenTypeList = kwargs.get("screenTypeList", None)
limitPerceptions = kwargs.get("limitPerceptions", False)
suppressHydrogens = kwargs.get("suppressHydrogens", False)
quietFlag = kwargs.get("quietFlag", True)
logSizes = kwargs.get("logSizes", False)
fpDbType = "STANDARD"
buildScreenedDb = True
#
oeCount = 0
errCount = 0
failIdList = []
oeIo = OeIoUtils(quietFlag=quietFlag)
# --------
oeSearchMolFilePath = os.path.join(self.__dirPath,
self.__getOeSearchMolFileName())
if not useCache or (useCache
and not self.__mU.exists(oeSearchMolFilePath)):
cmpKwargs = {
k: v
for k, v in kwargs.items()
if k not in ["cachePath", "useCache", "molLimit"]
}
ccsiP = ChemCompSearchIndexProvider(cachePath=cachePath,
useCache=True,
molLimit=molLimit,
**cmpKwargs)
ok = ccsiP.testCache(minCount=molLimit, logSizes=logSizes)
# ----
ccIdxD = ccsiP.getIndex() if ok else {}
idxCount = len(ccIdxD)
# ------- JDW OE mol construction here -----
startTime = time.time()
oeCount, errCount, failIdList = oeIo.buildOeBinaryMolCacheFromIndex(
oeSearchMolFilePath,
ccIdxD,
quietFlag=quietFlag,
fpTypeList=fpTypeList,
limitPerceptions=limitPerceptions,
suppressHydrogens=suppressHydrogens)
if failIdList:
logger.info("failures %r", failIdList)
endTime = time.time()
logger.info(
"Constructed %d/%d cached oeMols (unconverted %d) (%.4f seconds)",
oeCount, idxCount, errCount, endTime - startTime)
# --------
oeMolDbFilePath = os.path.join(self.__dirPath,
self.__getOeMolDbFileName())
if not useCache or (useCache
and not self.__mU.exists(oeMolDbFilePath)):
startTime = time.time()
molCount = oeIo.createOeBinaryDatabaseAndIndex(
oeSearchMolFilePath, oeMolDbFilePath)
endTime = time.time()
logger.info(
"Created and stored %d indexed oeMols in OE database format (%.4f seconds)",
molCount, endTime - startTime)
# --------
if fpDbType == "FAST":
for fpType in fpTypeList:
startTime = time.time()
# Fast FP search database file names
fpPath = os.path.join(self.__dirPath,
self.__getFastFpDbFileName(fpType))
if not useCache or (useCache
and not self.__mU.exists(fpPath)):
ok = oeIo.createOeFingerPrintDatabase(oeMolDbFilePath,
fpPath,
fpType=fpType)
endTime = time.time()
logger.info(
"Created and stored %s fingerprint database (%.4f seconds)",
fpType, endTime - startTime)
# --------
if buildScreenedDb and screenTypeList:
for screenType in screenTypeList:
startTime = time.time()
fp = os.path.join(self.__dirPath,
self.__getSubSearchFileName(screenType))
if not useCache or (useCache and not self.__mU.exists(fp)):
ok = oeIo.createOeSubSearchDatabase(
oeSearchMolFilePath,
fp,
screenType=screenType,
numProc=numProc)
endTime = time.time()
logger.info(
"Constructed screened substructure database (status %r) with screenType %s (%.4f seconds)",
ok, screenType, endTime - startTime)
# ---------
ssDb = oeIo.loadOeSubSearchDatabase(
fp, screenType=screenType, numProc=numProc)
ok = ssDb.NumMolecules() == oeCount
# ----------
#
return True
except Exception as e:
logger.exception("Failing with %s", str(e))
return False