def testCache(self, minMatchPrimaryPercent=None, logSizes=False):
okC = True
logger.info("Reference cache lengths: refIdMap %d matchD %d refD %d", len(self.__refIdMapD), len(self.__matchD), len(self.__refD))
ok = bool(self.__refIdMapD and self.__matchD and self.__refD)
#
numRef = len(self.__refIdMapD)
countD = defaultdict(int)
logger.info("Match dictionary length %d", len(self.__matchD))
for _, mD in self.__matchD.items():
if "matched" in mD:
countD[mD["matched"]] += 1
logger.info("Reference length %d match length %d coverage %r", len(self.__refD), len(self.__matchD), countD.items())
if minMatchPrimaryPercent:
try:
okC = 100.0 * float(countD["primary"]) / float(numRef) > minMatchPrimaryPercent
except Exception:
okC = False
logger.info("Primary reference match percent test status %r", okC)
#
if logSizes:
logger.info(
"Pfam %.2f InterPro %.2f SIFTS %.2f GO %.2f EC %.2f RefIdMap %.2f RefMatchD %.2f RefD %.2f",
getObjSize(self.__pfP) / 1000000.0,
getObjSize(self.__ipP) / 1000000.0,
getObjSize(self.__ssP) / 1000000.0,
getObjSize(self.__goP) / 1000000.0,
getObjSize(self.__ecP) / 1000000.0,
getObjSize(self.__refIdMapD) / 1000000.0,
getObjSize(self.__matchD) / 1000000.0,
getObjSize(self.__refD) / 1000000.0,
)
return ok and okC
def testCache(self, minCount=None, logSizes=False):
if logSizes and self.__ccIdxD:
logger.info("ccIdxD (%.2f MB)",
getObjSize(self.__ccIdxD) / 1000000.0)
ok = self.__ccIdxD and len(
self.__ccIdxD
) >= minCount if minCount else self.__ccIdxD is not None
return ok
def testCache(self, minMatchPrimaryPercent=None, logSizes=False, minMissing=0):
"""Test the state of reference sequence data relative to proportion of matched primary sequence
in the primary data set.
Args:
minMatchPrimaryPercent (float, optional): minimal acceptable of matching primary accessions. Defaults to None.
logSizes (bool, optional): flag to log resource sizes. Defaults to False.
minMissing (int, optional): minimum acceptable missing matched reference Ids. Defaults to 0.
Returns:
bool: True for success or False otherwise
"""
try:
ok = bool(self.__matchD and self.__refD and self.__missingMatchIds <= minMissing)
logger.info("Reference cache lengths: matchD %d refD %d missing matches %d", len(self.__matchD), len(self.__refD), self.__missingMatchIds)
if ok:
return ok
except Exception as e:
logger.error("Failing with unexpected cache state %s", str(e))
return False
#
# -- The remaining check on the portion is not currently --
#
numRef = len(self.__matchD)
countD = defaultdict(int)
logger.info("Match dictionary length %d", len(self.__matchD))
for _, mD in self.__matchD.items():
if "matched" in mD:
countD[mD["matched"]] += 1
logger.info("Reference length %d match length %d coverage %r", len(self.__refD), len(self.__matchD), countD.items())
if minMatchPrimaryPercent:
try:
okC = 100.0 * float(countD["primary"]) / float(numRef) > minMatchPrimaryPercent
except Exception:
okC = False
logger.info("Primary reference match count test status %r", okC)
#
if logSizes:
logger.info(
"RefMatchD %.2f RefD %.2f",
getObjSize(self.__matchD) / 1000000.0,
getObjSize(self.__refD) / 1000000.0,
)
return ok and okC
def testCache(self, minCount=None, logSizes=False):
okC = bool(self.__refD)
if not okC:
return okC
logger.info("Reference data cache lengths: refD %d", len(self.__refD))
if minCount and len(self.__refD) < minCount:
return False
#
if logSizes:
logger.info("refD %.2f", getObjSize(self.__refD) / 1000000.0)
return True
def testCache(self, minMatch=None, logSizes=False):
self.getMatchData()
okC = bool(self.__matchD)
if not okC:
return okC
logger.info("Reference data cache lengths: matchD %d",
len(self.__matchD))
if minMatch and len(self.__matchD) < minMatch:
return False
#
if logSizes:
logger.info("PubChem MatchD %.2f",
getObjSize(self.__matchD) / 1000000.0)
return True
def testCache(self, minCount=None, logSizes=False):
if logSizes and self.__ccMolD:
logger.info("ccMolD object size %.2f MB", getObjSize(self.__ccMolD) / 1000000.0)
ok = self.__ccMolD and len(self.__ccMolD) >= minCount if minCount else self.__ccMolD is not None
return ok
def testCache(self, minCount=None, logSizes=False):
if logSizes and self.__searchIdx:
logger.info("searchIdxD (%.2f MB)", getObjSize(self.__searchIdx) / 1000000.0)
ok = self.__searchIdx and len(self.__searchIdx) >= minCount if minCount else self.__searchIdx is not None
return ok