## BioBrick compatibility tool v 0.9b 01.06.12

def readseq (f) :
	lseq = f.readlines()
	s = ""
	k = 0
				
	while k < len(lseq) :
			
		if lseq[k][0] == ">" :
			lseq.remove(lseq[k])
		lseq[k] = lseq[k].replace("\n", "")
		k += 1
	#print lseq	
	
	for i in lseq :
		s += i
	s = s.replace("\W", "")
	s = s.replace(" ", "")
	#print s
			
	return s
	
def sites6 (s) :
	index = {}
	cons = ["GAATTC", "TCTAGA", "ACTAGT", "CTGCAG"]
	i=0
	
	while i < (len(s)-5) : 
		if s[i:i+6] in cons :
			index[s[i:i+6]] = i
		i += 1
	return index 
	
def sites8 (s) :
	index8 = {}
	cons = ["GCGGCCGC"]
	i=0
	
	while i < (len(s)-7) : 
		if s[i:i+8] in cons :
			index8[s[i:i+8]] = i
		i += 1
	return index8 	
	
def prefsuffix(seq) :
	pATG = "gaattcgcggccgcttctag"
	p = "gaattcgcggccgcttctagag" 
	s = "tactagtagcggccgctgcag"
	s1 = ""
	
	if seq[0:3] == "ATG" :
		s1 = pATG + seq + s	
	else : 
		s1 = p + seq + s
	
	return s1


f = open("./seq.fasta")

s = readseq(f)

six = sites6(s)
eight = sites8(s)

f=open('./output.txt', 'w')
restrictionsites = { "GAATTC":"EcoRI", "TCTAGA":"XbaI", "ACTAGT":"SpeI", "CTGCAG":"PstI", "GCGGCCGC":"NotI"}
		
if (not six) and (not eight) :
	f.write(prefsuffix(s) + "\n")
	print "The sequence is OK. Prefix and suffix have been added. Check the output.txt file in the folder of this script to get the results. "
else:
	for i in six.keys():	
		out = "There's a " + i +" ("+restrictionsites[i]+") at position " + str(six[i]) + ".\n"
		f.write(out)		
	for i in eight.keys():
		out = "There's a " + i +" ("+restrictionsites[i]+") at position " + str(eight[i]) + ".\n"
		f.write(out)			
	print "There are some forbidden restriction sites. Check the output.txt file in the folder of this script to get the results. "
	
f.close()