Computational biology at CSU

"""
Functions for finding motifs and converting fasta
formatted files to csv format using regular expressions
 
Author: Tai Montgomery, 10/30/18
"""
 
import re
import sys
 
 
def motif_finder(input_file, motif):
	"""
	Count all non-overlapping occurences
	of a sequence motif
	"""
	try:
		file_handle = open(input_file)
	except:
		print('file could not be opened')
		return -1
 
	with file_handle:
		file = file_handle.read()
		file = re.sub('\n', '', file).upper()
		new_motif = motif.replace('N', '.')
		print('Motif matches: ' + str(len(re.findall(new_motif, file))))
 
 
def fasta_to_csv(input_file, output_file):
	"""
	Convert fasta file to csv
	"""
	try:
		infile = open(input_file)
		outfile = open(output_file, 'w')
	except:
		print('file could not be opened')
		return -1
	with infile, outfile:
		file = infile.read()
		tab_separated = re.sub('>(.*)\n', r'\1,', file)
		outfile.write(tab_separated)
 
 
if __name__ == '__main__':
	# For testing motif_finder, uncomment next two lines
	#input_file, motif = sys.argv[1:3]
	#input_file, motif, motif_finder(input_file, motif)
 
	# For testing fasta_to_csv, uncomment next two lines
	#input_file, output_file = sys.argv[1:3]
	#print(fasta_to_csv(input_file, output_file))