#!/usr/bin/env python
# -*- coding: utf-8 -*-

"""
uses line by line matching files for DE matrix and GO groups to
provide DE values for each GO entry portionate to the number of GO 
groups for that particular sequence"

Usage:
% GO_DE.py matrix.txt GO.txt

  Written by jason macrander """

import sys
import re
import collections

#Filament_write = open("Filament.txt",'w')
gastric_cirri_write = open("gastric_cirri.txt",'w')
Tentacle_write = open("Tentacle.txt",'w')

try:
    from Bio import SeqIO
except:
    print "This program requires the Biopython library"
    sys.exit(0)

try:
	matrix_file = sys.argv[1] #Inputs the matrix file
	GO_file = sys.argv[2] #inputs the GO file
except:
	print __doc__
	sys.exit(0)


GO = {}
with open (GO_file) as n:
	for line in n:
		if line != (""):
			line = line.strip()
			line = line.split("\t")
			ID = line[0]
			print ID
			GO_IDs = line[1].split("; ")
			GO[ID] = GO.get(ID,[]) + list(set(GO_IDs))

#print GO

matrix = {}
#Fentry = collections.defaultdict(list)
GCentry = collections.defaultdict(list)
Tentry = collections.defaultdict(list)
with open(matrix_file, 'r') as m:
		for line in m:
			if line != (""):
				line = line.strip()
				line = line.split("\t")
				transcript = line[0]
				ID = line[3] 
				#Filament = float(line[3])
				gastric_cirri= float(line[1])
				Tentacle = float(line[2])
				for key, value in GO.iteritems():
					if key == ID:
						#Filament_value = Filament/len(value)						
						gastric_cirri_value = gastric_cirri/len(value)
						Tentacle_value = Tentacle/len(value)
						for item in value:
							#Fentry[item].append(Filament_value)
							GCentry[item].append(gastric_cirri_value)
							Tentry[item].append(Tentacle_value)


#print Fentry
#for key,value in Fentry.iteritems():
#	if sum(value) > 0:
#		#print "%s\t%f" % (key,sum(value))
#		Filament_write.write("%s\t%f\n" % (key,sum(value)))
#print "Filament done"

for key,value in GCentry.iteritems():
	if sum(value) > 0:
		#print "%s\t%f" % (key,sum(value))
		gastric_cirri_write.write("%s\t%f\n" % (key,sum(value)))
print "gastric_cirri done"

for key,value in Tentry.iteritems():
	if sum(value) > 0:
		#print "%s\t%f" % (key,sum(value))
		Tentacle_write.write("%s\t%f\n" % (key,sum(value)))
print "Tentacle done"

#print matrix







#with open(matrix_file, 'r') as m:
#		for line in m:
#			if line != (""):
#				line = line.strip()
#				line = line.split("\t")
#				transcript = line[0]
#				entry = {"ID": line[4], 
#				"Filament":line[1],
#				"gastric_cirri":line[2],
#				"Tentacle":line[3]
#				}
#				matrix[transcript] = entry
#
##print matrix
#