def makeHits(self,
queueURL, # the queue we'll be sending work to
commsPort, # port we'll be asked for progress etc on
portRange, # the total number of concurrent pairs we can handle
sgeBaseDir, # where to write SGE scripts to
workingDir, # where tmp files will be stored
batches=[], # the order to do pairs in
hitCache=1000 # number of hits to cache before writing to database
):
"""process any specified outstanding pairs"""
print ">> DBG " + datetime.datetime.fromtimestamp(time.time()).strftime('%Y-%m-%d %H:%M:%S')
print "Processing outstanding pairs"
# set tmp dirs
self.sgeBaseDir = sgeBaseDir
self.workingDir = workingDir
# get hold of the server queue
self.queueManager = SGE(queueURL)
VI = ViewInterface(self.dbFileName)
contig_headers = {}
if batches != []:
for batch in batches:
pairs = VI.getOutstandingPairs(batch=batch)
if len(pairs) > 0:
contig_headers = VI.getContigHeaders()
self.processOutstandingPairs(pairs, portRange, contig_headers, hitCache, batch=batch)
else:
pairs = VI.getOutstandingPairs()
if len(pairs) > 0:
contig_headers = VI.getContigHeaders()
self.processOutstandingPairs(pairs, portRange, contig_headers, hitCache)
def __init__(self, dbFileName, ani=95, batches=[]):
# get an interface to the DB
VI = ViewInterface(dbFileName)
# build the condition we want to select on and get the hit data
C = Condition("ani_comp", "<=", ani)
if len(batches) > 0:
bc = Condition("batch", "=", batches[0])
for batch_num in range(1, len(batches)):
bc = Condition(bc, "or", Condition("batch", "=", batches[batch_num]))
C = Condition(C, "and", bc)
seqs = VI.getHitData(C)