fq_split

#!/bin/env python2.7
# Small script splitting a genome fasta into ordered, overlapping reads


def exit(str):
	print str
	parser.print_help()
	sys.exit()


def getRead(lines, pos, length):
	"""Return a read of determined length starting at position"""

	read=''
	linelength = len(lines[0]) - 1
	
	linenum = pos / linelength
	start = pos % linelength


	i = 0
	while (len(read) < length):
		if ((linenum + i) >= len(lines)):
			return read	
		
		read = read + lines[linenum + i][start:-1]
		start = 0
		i += 1
	
	if (len(read) > length):
		read = read[:length]

	return read


def getSize(lines):
	N = len(lines)
	linelength = len(lines[0])

	lastlength = len(lines[N - 1])
	return (N - 1)*linelength + lastlength



if __name__=="__main__":
	
	import sys
	import argparse

	parser = argparse.ArgumentParser(description = 'Splits a fasta genome file into ordered reades with overlap.',
						prog='fq_split', usage = '%(prog)s -i input_file [optional arguments]',
						formatter_class=lambda prog: argparse.HelpFormatter(prog, max_help_position = 50, width = 200),
						add_help = False)

	required = parser.add_argument_group('Required Arguments')
	required.add_argument('-i', '--input', default = '', help = 'input file')
	
	optional = parser.add_argument_group('Optional Arguemnts')
	optional.add_argument('-l', '--read-length', default = 60, type = int, help = 'readlength (60)')
	optional.add_argument('-ovl', '--overlap', default = 12, type = int, help = 'reads overlap (12)')
	optional.add_argument('-o', '--output', default = '', help = 'output file (stdout)')
	optional.add_argument('-H', '--headers', action = 'store_true', help = 'display headers in output (False)')
	optional.add_argument('-h', '--help', action='help')

	args = vars( parser.parse_args() )

	# Verify input validity:
	if (args['input'] is ''):
		exit("No inputs provided")
	if (args['overlap'] < 0):
		exit('Overlap can\'t be negative')
	if (args['overlap'] > args['read_length']):
		exit('Overlap can\'t be larger than read-length')
	if (args['read_length'] < 0):
		exit('Read length can\'t be negative')


	if (args['output'] is ''):
		out = sys.stdout
	else:
		out = open(args['output'], 'w')


	
	# Open files, find loop max
	infile = open(args['input'], 'r')
	lines = infile.readlines()
	del lines[0]

	max = getSize(lines)

	# Begin loop
	position = 0
	while ((position + args['read_length']) <= (max - 1)):
		read = getRead(lines, position, args['read_length'])
		
		if (read is ''):
			break
		
		if args['headers']:
			header = ">Split_reads-" + args['input'] + "-pos" + str(position)
			out.write(header + '\n')
		out.write(read + '\n')

		position += (args['read_length'] - args['overlap'])



	if (not args['output'] is ''):
		out.close()

	# End of program