# find_substitution.py
# Author: Sami Khuri
# Last updated: December 28, 2017
# Purpose: To locate the positions where 2 similar sequences differ
# Program uses the functions len() and min() and user defined function
#         substitution() to form a list of indices at which the 2 seqs differ
#

def substitution(a, b):
    """ Creates a list of indices for which the values in seqs a and b differ """
    x1 = len(a)
    x2 = len(b)
    z = [ ]
    end = min(x1, x2)
    for p in range(0, end):
        if a[p] != b[p]:
            z = z + [p]
    return z

# main program

handle_Protein_wildtype = open("Protein_wildtype.txt","r")
seq_wildtype = handle_Protein_wildtype.read()
seq_wildtype = seq_wildtype.replace(" ","")
seq_wildtype = seq_wildtype.lower()

handle_Protein_mutant = open("Protein_mutant.txt","r")
seq_mutant = handle_Protein_mutant.read()
seq_mutant = seq_mutant.replace(" ","")
seq_mutant = seq_mutant.lower()

my_list = substitution(seq_wildtype,seq_mutant)


# Output findings in tabular fashion (not zero-based)
print("Postion" + "\tFirst" + "\tSecond")
print("-------" + "\t-----" + "\t------")
for pos in my_list:
    print(str(pos+1) + "\t" + seq_wildtype[pos] + "\t" + seq_mutant[pos])