parseContigs

#!/usr/bin/python
'''
Parse the results of a CONTIGuator mapping
They should be present as distinct directories in a folder
The folder should have this format: DRAFT_COMPLETE_*
The user can provide a Threshold of genomes in which the contig has to mapped
to be assigned to a particular replicon.

Inputs: IN_DIR MAP_THRESHOLD
Outputs: a series of tab files
CONTIG  REPLICON
'''

import sys

if len(sys.argv) < 3:
    print 'USAGE: parseContigs IN_DIR MAP_THRESHOLD'
    sys.exit(1)

indir, mapt = sys.argv[1:3]
mapt = int(mapt)

import os
dir_1 = os.listdir(indir)
Draft = {}

for i in dir_1:
	try:
		draft, complete, maptag, replicon = i.split('_')
	except:                                  
		try:    
			draft, complete, replicon = i.split('_')
		except:
			continue
	try:
		f = os.path.join(indir, i, 'MappedContigs.txt')
		for l in open(f):
			contig, number = l.split('\t')
			#print draft, replicon, contig
			Draft[draft] = Draft.get(draft, {})
			Draft[draft][contig] = Draft[draft].get(contig, {})
			Draft[draft][contig][replicon] = Draft[draft][contig].get(replicon, 0)
			Draft[draft][contig][replicon] +=1
	except:
		f = os.path.join(indir, i, 'UnMappedContigs.txt')
		for k in open(f):
			contig, number = k.split('\t')
			#print draft, replicon, contig
			Draft[draft] = Draft.get(draft, {})
			Draft[draft][contig] = Draft[draft].get(contig, {})
			Draft[draft][contig][replicon] = Draft[draft][contig].get(replicon, 0)
			Draft[draft][contig][replicon] +=1
for draft in Draft:
	draft_1 = draft + '.tab'
	f = open(draft_1, 'w')
	for contig in Draft[draft]:
		keys = Draft[draft][contig].keys ()
		if len(keys)==1:
			if keys == ['UnMappedContigs']:
				f.write(contig+'\t'+"UnMapped"+"\n")
			else:
				f.write(contig+'\t'+keys[0]+"\n")
		elif len(keys)==2 and 'UnMappedContigs' in keys:
			keys.remove('UnMappedContigs')
			f.write(contig+'\t'+keys[0]+"\n")
		elif len(keys)>=2:
			b=False
			for replicon in Draft[draft][contig]:
				if Draft[draft][contig][replicon]>=mapt:
					b=True
					break
			if b is True:
				if replicon == ('UnMappedContigs'):
					f.write(contig+'\t'+"UnMapped"+"\n")
				else:
					f.write(contig+'\t'+replicon+"\n")
			else:
				f.write(contig+'\t'+"UnMapped"+"\n")
	f.close()