def dbLongestSequences( num, inFileName, outFileName="", verbose=0, minThresh=0 ):
bsDB = BioseqDB( inFileName )
if verbose > 0:
print "nb of input sequences: %i" % ( bsDB.getSize() )
if outFileName == "":
outFileName = inFileName + ".best" + str(num)
outFile = open( outFileName, "w" )
if bsDB.getSize()==0:
return 0
num = int(num)
if verbose > 0:
print "keep the %i longest sequences" % ( num )
if minThresh > 0:
print "with length > %i bp" % ( minThresh )
sys.stdout.flush()
# retrieve the length of each input sequence
tmpLSeqLgth = []
seqNum = 0
for bs in bsDB.db:
seqNum += 1
tmpLSeqLgth.append( bs.getLength() )
if verbose > 1:
print "%d seq %s : %d bp" % ( seqNum, bs.header[0:40], bs.getLength() )
sys.stdout.flush()
# sort the lengths
tmpLSeqLgth.sort()
tmpLSeqLgth.reverse()
# select the longest
lSeqLgth = []
for i in xrange( 0, min(num,len(tmpLSeqLgth)) ):
if tmpLSeqLgth[i] >= minThresh:
lSeqLgth.append( tmpLSeqLgth[i] )
if verbose > 0:
print "selected max length: %i" % ( max(lSeqLgth) )
print "selected min length: %i" % ( min(lSeqLgth) )
sys.stdout.flush()
# save the longest
inFile = open( inFileName )
seqNum = 0
nbSave = 0
for bs in bsDB.db:
seqNum += 1
if bs.getLength() >= min(lSeqLgth) and bs.getLength() >= minThresh:
bs.write( outFile )
if verbose > 1:
print "%d seq %s : saved !" % ( seqNum, bs.header[0:40] )
sys.stdout.flush()
nbSave += 1
if nbSave == num:
break
inFile.close()
outFile.close()
if verbose > 0:
print nbSave, "saved sequences in ", outFileName
sys.stdout.flush()
return 0