def __init__(self,
organisms,
ncpus=1,
evalue=1e-10,
recover=False,
prefix='',
matrix='BLOSUM80',
queue=Queue.Queue()):
CommonMultiProcess.__init__(self, ncpus, queue)
# Blast
self.organisms = list(organisms)
self.dbs = {}
self._prot2orgs = {}
self.out = []
self.evalue = float(evalue)
# TODO: implement recovery
self.recover = recover
#
self.results = {}
self._blast = Blaster()
self._pangenomeroom = None
self.prefix = prefix.rstrip('_')
self.matrix = matrix
self._already = set()
# Results
self.orthologs = {}
self.core = []
self.accessory = []
self.unique = []
def __init__(self,
query,
target,
ncpus=1,
evalue=1e-50,
buildDB=True,
bbh=True,
recover=False,
queue=Queue.Queue()):
CommonMultiProcess.__init__(self, ncpus, queue)
# Blast
self.query = query
if buildDB:
self.target = target
self.db = None
else:
self.target = None
self.db = target
self.out = []
self.evalue = float(evalue)
self.bbh = bool(bbh)
self.recover = recover
self.ncpus = int(ncpus)
self._kohits = []
self.results = {}
self._keggroom = None
self._blast = Blaster()
def serialBBH(self):
orthindex = 1
self._maxsubstatus = len(self._prot2orgs)
for org in self.organisms:
seqs = [seq for seq in SeqIO.parse(open(org), 'fasta')]
# Iterate over each protein
for seq in seqs:
self._substatus += 1
self.updateStatus(sub=True)
# Log some info, might be useful for
# long running jobs
logger.debug('Running orthology prediction for protein %d/%d' %
(self._substatus, self._maxsubstatus))
logger.debug('Organism: %s, Protein: %s' % (org, seq.id))
if seq.id in self._already:
continue
orthname = self.prefix + str(orthindex)
orgsincluded = [org]
self.orthologs[orthname] = [seq.id]
query = '>%s\n%s\n' % (seq.id, str(seq.seq))
self.initiateParallel()
# Iterate over each other organism
for otherorg in self.organisms:
if org == otherorg:
continue
# Go fot it!
if len(seq) < 30:
short = True
else:
short = False
uniqueid = self.getUniqueID()
# Multi process
obj = RunBBH(query,
seq.id,
self.dbs[org],
self.dbs[otherorg],
otherorg,
self.evalue,
self.matrix,
short=short,
uniqueid=uniqueid,
useDisk=False)
self._paralleltasks.put(obj)
# Poison pill to stop the workers
self.addPoison()
while True:
if self.killed:
logger.debug('Exiting for a kill signal')
return
while not self._parallelresults.empty():
if self.killed:
logger.debug('Exiting for a kill signal')
return
result = self._parallelresults.get()
if not result[2]:
logger.error(
'An error occurred for BBH on query %s' %
seq.id + ' and target %s' % result[1])
return False
if result[0] and result[0] not in self._already:
self.orthologs[orthname].append(result[0])
orgsincluded.append(result[1])
self._already.add(result[0])
if self.isTerminated():
break
self.sleeper.sleep(0.01)
while not self._parallelresults.empty():
if self.killed:
logger.debug('Exiting for a kill signal')
return
result = self._parallelresults.get()
if not result[2]:
logger.error('An error occurred for BBH on query %s' %
seq.id + ' and target %s' % result[1])
return False
if result[0] and result[0] not in self._already:
self.orthologs[orthname].append(result[0])
orgsincluded.append(result[1])
self._already.add(result[0])
self.killParallel()
if len(orgsincluded) < len(self.organisms):
logger.debug('Additional search on missing organisms for' +
' ortholog %s' % orthname)
for otherprotein in self.orthologs[orthname]:
if otherprotein == seq.id:
continue
neworg = self._prot2orgs[otherprotein]
if neworg == org:
continue
searcher = Blaster(useDisk=False)
searcher.retrieveFromDB(self.dbs[neworg], otherprotein)
query = searcher.retrieved
self.initiateParallel()
for evenneworg in self.organisms:
if evenneworg in orgsincluded:
continue
# Go fot it!
if len(seq) < 30:
short = True
else:
short = False
uniqueid = self.getUniqueID()
# Multi process
obj = RunBBH(query,
otherprotein,
self.dbs[neworg],
self.dbs[evenneworg],
evenneworg,
self.evalue,
self.matrix,
short=short,
uniqueid=uniqueid,
useDisk=False)
self._paralleltasks.put(obj)
# Poison pill to stop the workers
self.addPoison()
while True:
if self.killed:
logger.debug('Exiting for a kill signal')
return
while not self._parallelresults.empty():
if self.killed:
logger.debug('Exiting for a kill signal')
return
result = self._parallelresults.get()
if not result[2]:
logger.error(
'An error occurred for BBH on query %s'
% seq.id +
' and target %s' % result[1])
return False
if result[0] and result[0] not in self._already:
self.orthologs[orthname].append(result[0])
orgsincluded.append(result[1])
self._already.add(result[0])
if self.isTerminated():
break
self.sleeper.sleep(0.01)
while not self._parallelresults.empty():
if self.killed:
logger.debug('Exiting for a kill signal')
return
result = self._parallelresults.get()
if not result[2]:
logger.error(
'An error occurred for BBH on query %s' %
seq.id + ' and target %s' % result[1])
return False
if result[0] and result[0] not in self._already:
self.orthologs[orthname].append(result[0])
orgsincluded.append(result[1])
self._already.add(result[0])
self.killParallel()
orthindex += 1
return True
