# create_pwm.py
# Author: Neha Bhagwat
# Last updated: December 31, 2017
# Purpose: Program to create a position weight matrix with Laplace pseudocount
# Program uses the Python function open() and the Python methods read(), write(),
#     close(), split(), append() and log()
#
from __future__ import division
import math

# Take Donor_MOG.txt as input
input_file = open("Donor_MOG.txt","r")

input_data = input_file.read()
lines = input_data.split('\r\n')

input_file.close()

# Initialize 4 lists: a, c, g, and t will 9 zeros in each
a = 9*[0]
c = 9*[0]
g = 9*[0]
t = 9*[0]

# Count of the occurrences of each character at a given position
for line in lines:
    for i in range(9):
        if line[i] == 'A':
            a[i] = a[i]+1
        elif line[i] == 'C':
            c[i] = c[i]+1
        elif line[i] == 'G':
            g[i] = g[i]+1
        else:
            t[i] = t[i]+1

# Compute the 36 log-odd scores (log of observed/expected) where expected is:
# 0.28 for a's and t's, and 0.22 for c's and g's.
# Pseudocount is 1 [aka Laplace pseudocount]
# The log is taken in base 2
for i in range(9):
    a[i] = round(math.log(((a[i] + 1)/11)/0.28,2),3)
    c[i] = round(math.log(((c[i] + 1)/11)/0.22,2),3)
    g[i] = round(math.log(((g[i] + 1)/11)/0.22,2),3)
    t[i] = round(math.log(((t[i] + 1)/11)/0.28,2),3)

# Write the 36 values of the entries of the PWM into a file Donor_MOG_matrix.txt,
# Donor_MOG_matrix.txt should have 4 rows with 9 values on each line
output_file = open("Donor_MOG_matrix.txt","w")
for i in range(9):
    output_file.write(str(a[i]) + "\t")
output_file.write("\n")
for i in range(9):
    output_file.write(str(c[i]) + "\t")
output_file.write("\n")
for i in range(9):
    output_file.write(str(g[i]) + "\t")
output_file.write("\n")
for i in range(9):
    output_file.write(str(t[i]) + "\t")
output_file.write("\n")

output_file.close()