orf.py

#!/usr/bin/python
import sys

rna={}
f=open('rna_table.txt','rU')
for line in f:
    x=line.split()
    codons=x[0::2]
    acids=x[1::2]
    rna_new=dict(zip(codons,acids))
    rna.update(rna_new)

f.close()

f=open('rosalind_orf.txt','rU')
import string
import sys

value,key,s="","",{}
maxt, max="",0.0

with open("rosalind_orf.txt","rU") as f:
    for x in f:
        if x[0]!=">":
            value=value+x.strip()
        else:
            if key!="": s[key]=value
            key=x[1:].strip()

s=""
sc=""

# crude translation to RNA from DNA
for x in value:
    if x=="T":
        s+="U"
        sc+="A"
    if x=="A":
        s+=x
        sc+="U"
    if x=="G":
        s+=x
        sc+="C"
    if x=="C":
        s+=x
        sc+="G"

proteins={}
p=""
for offset in range(0,3):
    for i in range(offset,len(s)-3,3):
        if s[i:i+3]=="AUG":
            p="M"
            for j in range(i+3,len(s)-3,3):
                if rna[s[j:j+3]]=='Stop':
                    proteins[p]=1
                    p=""
                    break
                p=p+rna[s[j:j+3]]

s=sc[::-1]
p=""
for offset in range(0,3):
    for i in range(offset,len(s)-3,3):
        if s[i:i+3]=="AUG":
            p="M"
            for j in range(i+3,len(s)-3,3):
                if rna[s[j:j+3]]=='Stop':
                    proteins[p]=1
                    p=""
                    break
                p=p+rna[s[j:j+3]]

for x in proteins.keys():
    print x