#!/usr/bin/python
# script for computing overlap signatures from a bowtie output
# Christophe Antoniewski <drosofff@gmail.com>
# Usage piRNAsignature.py <1:input> <2:format of input> <3:minsize query> <4:maxsize query> <5:minsize target> <6:maxsize target>
#			  <7:minscope> <8:maxscope> <9:output> <10:bowtie index> <11:procedure option> <12: graph (global or lattice)>
#			  <13: R code>

import sys, subprocess
from smRtools import *
from collections import defaultdict # test whether it is required

if sys.argv[11] == "--extract_index":
  if sys.argv[2] == "tabular":
    Genome = HandleSmRNAwindows (sys.argv[1],"tabular",sys.argv[10],"bowtieIndex")
  elif sys.argv[2] == "sam":
    Genome = HandleSmRNAwindows (sys.argv[1],"sam",sys.argv[10],"bowtieIndex")
  else:
    Genome = HandleSmRNAwindows (sys.argv[1],"bam",sys.argv[10],"bowtieIndex")
else:
  if sys.argv[2] == "tabular":
    Genome = HandleSmRNAwindows (sys.argv[1],"tabular",sys.argv[10],"fastaSource") 
  elif sys.argv[2] == "sam":
    Genome = HandleSmRNAwindows (sys.argv[1],"sam",sys.argv[10],"fastaSource")
  else:
    Genome = HandleSmRNAwindows (sys.argv[1],"bam",sys.argv[10],"fastaSource")
# this decisional tree may be simplified if sam and bam inputs are treated the same way by pysam

# replace objDic by Genome.instanceDict or... objDic = Genome.instanceDict
objDic = Genome.instanceDict

minquery = int(sys.argv[3])
maxquery = int(sys.argv[4])
mintarget = int(sys.argv[5])
maxtarget = int(sys.argv[6])
minscope = int(sys.argv[7])
maxscope = int(sys.argv[8]) + 1
general_frequency_table = dict ([(i,0) for i in range(minscope,maxscope)])
general_percent_table = dict ([(i,0) for i in range(minscope,maxscope)])
OUT = open (sys.argv[9], "w")

if sys.argv[12] == "global":
  ###### for normalized summing of local_percent_table(s)
  readcount_dic = {}
  Total_read_in_objDic = 0
  for item in objDic:
    readcount_dic[item] = objDic[item].readcount(minquery, maxquery)
    Total_read_in_objDic += readcount_dic[item]
  ######
  for x in (objDic):
    local_frequency_table = objDic[x].signature( minquery, maxquery, mintarget, maxtarget, range(minscope,maxscope) )
    local_percent_table = objDic[x].hannon_signature( minquery, maxquery, mintarget, maxtarget, range(minscope,maxscope) )
    try:
      for overlap in local_frequency_table.keys():
        general_frequency_table[overlap] = general_frequency_table.get(overlap, 0) + local_frequency_table[overlap]
    except:
      pass
    try:
      for overlap in local_percent_table.keys():
        general_percent_table[overlap] = general_percent_table.get(overlap, 0) + (1./Total_read_in_objDic*readcount_dic[x]*local_percent_table[overlap])
    except:
      pass
  print >> OUT, "overlap\tnum of pairs\tprobability"
  for classe in sorted(general_frequency_table):
    print >> OUT, "%i\t%i\t%f" % (classe, general_frequency_table[classe], general_percent_table[classe])

else:
  print >> OUT, "overlap\tnum of pairs\tprobability\titem"
  for x in (objDic):
    local_frequency_table = objDic[x].signature( minquery, maxquery, mintarget, maxtarget, range(minscope,maxscope) )
    local_percent_table = objDic[x].hannon_signature( minquery, maxquery, mintarget, maxtarget, range(minscope,maxscope) )
    for classe in range(minscope,maxscope):
      print >> OUT, "%i\t%i\t%f\t%s" % (classe, local_frequency_table[classe], local_percent_table[classe], x)

OUT.close()

## Run the R script that is defined in the xml using the Rscript binary provided with R.
R_command="Rscript "+ sys.argv[13]
process = subprocess.Popen(R_command.split())