class CoreCreator:
"""Top level class for making bins"""
def __init__(self, dbFileName):
self._pm = ProfileManager(dbFileName)
self._dbFileName = dbFileName
def loadProfile(self, timer, minLength):
return self._pm.loadData(timer,
minLength=minLength,
loadMarkers=True,
loadBins=False)
def run(self,
timer,
minLength,
minSize,
minPts,
savedDistsPrefix="",
keepDists=False,
force=False):
# check that the user is OK with nuking stuff...
if not force and not self._pm.promptOnOverwrite():
return
profile = self.loadProfile(timer,
minLength=minLength
)
if savedDistsPrefix=="":
savedDistsPrefix = self._dbFileName+".dists"
cacher = FileCacher(savedDistsPrefix)
ce = ClassificationClusterEngine(profile,
minPts=minPts,
minSize=minSize,
cacher=cacher,
)
ce.makeBins(timer,
out_bins=profile.binIds,
out_reach_order=profile.reachOrder,
out_reach_dists=profile.reachDists
)
# Now save all the stuff to disk!
print "Saving bins"
self._pm.setReachabilityOrder(profile)
self._pm.setBinAssignments(profile, nuke=True)
print " %s" % timer.getTimeStamp()
# Remove created files
if not keepDists:
try:
cacher.cleanup()
except:
raise
class CoreCreator:
"""Top level class for making bins"""
def __init__(self, dbFileName):
self._pm = ProfileManager(dbFileName)
self._dbFileName = dbFileName
def loadProfile(self, timer, minLength):
return self._pm.loadData(timer,
minLength=minLength,
loadMarkers=True,
loadBins=False)
def run(self,
timer,
minLength,
minSize,
minPts,
savedDistsPrefix="",
keepDists=False,
force=False):
# check that the user is OK with nuking stuff...
if not force and not self._pm.promptOnOverwrite():
return
profile = self.loadProfile(timer, minLength=minLength)
if savedDistsPrefix == "":
savedDistsPrefix = self._dbFileName + ".dists"
cacher = FileCacher(savedDistsPrefix)
ce = ClassificationClusterEngine(
profile,
minPts=minPts,
minSize=minSize,
cacher=cacher,
)
ce.makeBins(timer,
out_bins=profile.binIds,
out_reach_order=profile.reachOrder,
out_reach_dists=profile.reachDists)
# Now save all the stuff to disk!
print "Saving bins"
self._pm.setReachabilityOrder(profile)
self._pm.setBinAssignments(profile, nuke=True)
print " %s" % timer.getTimeStamp()
# Remove created files
if not keepDists:
try:
cacher.cleanup()
except:
raise
class BinImporter:
"""Used for importing bin assignments"""
def __init__(self,
dbFileName):
self._pm = ProfileManager(dbFileName)
def loadProfile(self, timer):
return self._pm.loadData(timer)
def importBinAssignments(self,
timer,
infile,
separator):
"""Parse assignment file for bin contigs"""
profile = self.loadProfile(timer)
br = BinReader()
# looks like cid->bid
contig_bins = {}
try:
with open(infile, "r") as f:
try:
(con_names, con_bins) = br.parse(f, separator)
(_, con_bid) = np.unique(con_bins, return_inverse=True)
con_bid += 1 # bid zero is unbinned
contig_bins = dict(zip(con_names, con_bid))
except:
print "Error parsing bin assignments"
raise
except:
print "Could not parse bin assignment file:",infile,sys.exc_info()[0]
raise
# now get the internal indices for contigs
for (i, cid) in enumerate(profile.contigNames):
try:
profile.binIds[i] = contig_bins[cid]
except KeyError:
pass
# Now save all the stuff to disk!
print "Saving bins"
self._pm.setBinAssignments(profile, nuke=True)
print " %s" % timer.getTimeStamp()
class BinImporter:
"""Used for importing bin assignments"""
def __init__(self, dbFileName):
self._pm = ProfileManager(dbFileName)
def loadProfile(self, timer):
return self._pm.loadData(timer)
def importBinAssignments(self, timer, infile, separator):
"""Parse assignment file for bin contigs"""
profile = self.loadProfile(timer)
br = BinReader()
# looks like cid->bid
contig_bins = {}
try:
with open(infile, "r") as f:
try:
(con_names, con_bins) = br.parse(f, separator)
(_, con_bid) = np.unique(con_bins, return_inverse=True)
con_bid += 1 # bid zero is unbinned
contig_bins = dict(zip(con_names, con_bid))
except:
print "Error parsing bin assignments"
raise
except:
print "Could not parse bin assignment file:", infile, sys.exc_info(
)[0]
raise
# now get the internal indices for contigs
for (i, cid) in enumerate(profile.contigNames):
try:
profile.binIds[i] = contig_bins[cid]
except KeyError:
pass
# Now save all the stuff to disk!
print "Saving bins"
self._pm.setBinAssignments(profile, nuke=True)
print " %s" % timer.getTimeStamp()
