fq_translate

#!/bin/env python2.7
#
#
# Just a cute little python script to switch between different fastq format (sanger, solexa, illumina 1,5 and 1,8)
# Slow, but does the job. Better than emboss (it actually works).
#
# Jordan Texier, Jan 2014
# DittmerLab, UNC




def stringOffset(string, n):
	"""Offsets a string's ascii values by an offset of 'n'"""
	charlist=[]

	for char in string:
		val=ord(char)
		newval=val + n
		newchar=chr(newval)
		charlist.append(newchar)

	return "".join(charlist)





# ------------------------------ #
# ============ MAIN ============ #
# ------------------------------ #

if __name__=="__main__":


	import sys
	import argparse
	
	# Create Parser, required and optional args. Then parse:
	parser = argparse.ArgumentParser(description='Changes the fastq format of a file',
						prog='fq_translate',
						usage='%(prog)s -i input_file -if input_format -of output_format [-o output_file]',
						formatter_class=lambda prog: argparse.HelpFormatter(prog,max_help_position = 50),
						add_help=False)
	
	required = parser.add_argument_group('Required Arguments')
	required.add_argument('-i', '--input', default='',  help='input file')
	required.add_argument('-if', '--in-format', default='', type=str, help='Input format ("sanger", "solexa", "illumina-1.5", "illumina-1.8")')
	required.add_argument('-of', '--out-format', default='', type=str, help='Output format (same as -if)')
	
	optional = parser.add_argument_group('Optional Arguments')
	optional.add_argument('-h', '--help', action='help')
	optional.add_argument('-o', '--output', default='',  help='output file, if empty print to console')


	args = vars(parser.parse_args())


	# Verify required args:
	if(args["input"] == ''):
		parser.print_help()
		sys.exit()
	
	# Find ASCII Offset:
	if(args["out_format"] == '' and args["in_format"] == ''):
		# Define default offset
		offset = 31
	else:
		if(args["out_format"] == '' or args["in_format"] == ''):
			print "Both input and output format must be defined:"
			parser.print_help()
			sys.exit()
		
		inAscii = 0
		if(args["in_format"] == '18' or args["in_format"] == 'illumina-1.8'
				or args["in_format"] == 'sanger'):
			inAscii = 33
		elif(args["in_format"] == '15' or args["in_format"] == 'illumina-1.5'
				or args["in_format"] == 'solexa'):
			inAscii = 64
		else:
			print "Input format not recognized:\n"
			print args["in_format"]
			print "Acceptable arguments:\n illumina-1.8,\n illumina-1.5,\n sanger, or\n solexa."
			parser.print_help()
			sys.exit()



		outAscii = 0
		if(args["out_format"] == '18' or args["out_format"] == 'illumina-1.8'
				or args["out_format"] == 'sanger'):
			outAscii = 33
		elif(args["out_format"] == '15' or args["out_format"] == 'illumina-1.5'
				or args["out_format"] == 'solexa'):
			outAscii = 64
		else:
			print "Output format not recognized\n"
			print "Acceptable arguments:\n illumina-1.8,\n illumina-1.5,\n sanger, or\n solexa."
			parser.print_help()
			sys.exit()

		offset = outAscii - inAscii



	# Begin translation: 	
	
	with open(args['input']) as infile:
		i=0
	
		if (args['output'] is ''):
			out = sys.stdout
		else:
			out = open(args['output'], 'a')

		for line in infile:
			if (i%4 == 3):
				newStr = stringOffset(line[:-1], offset)
				out.write(newStri + '\n')
			else: 
				out.write(line)

		infile.close()