def sortSequencesByIncreasingLength( inFile, outFile, verbose=0 ):
"""
Save sequences in 'inFile' into 'outFile' sorted by their length in increasing order.
"""
if verbose > 0:
print "sort sequences by increasing length"
sys.stdout.flush()
if not os.path.exists( inFile ):
print "ERROR: file '%s' doesn't exist" % ( inFile )
sys.exit(1)
# read each seq one by one
# save them in distinct temporary files
# with their length in the name
inFileHandler = open( inFile, "r" )
bs = Bioseq()
countSeq = 0
while True:
bs.read( inFileHandler )
if bs.header == None:
break
countSeq += 1
tmpFile = "%ibp_%inb" % ( bs.getLength(), countSeq )
bs.save( tmpFile )
if verbose > 1:
print "%s (%i bp) saved in '%s'" % ( bs.header, bs.getLength(), tmpFile )
bs.header = ""
bs.sequence = ""
inFileHandler.close()
# sort temporary file names
# concatenate them into the output file
if os.path.exists( outFile ):
os.remove( outFile )
lFiles = glob.glob( "*bp_*nb" )
lFiles.sort( key=lambda s:int(s.split("bp_")[0]) )
for fileName in lFiles:
cmd = "cat %s >> %s" % ( fileName, outFile )
returnValue = os.system( cmd )
if returnValue != 0:
print "ERROR while concatenating '%s' with '%s'" % ( fileName, outFile )
sys.exit(1)
os.remove( fileName )
return 0
