#! /usr/bin/env python

"""
collection of regular expression patterns used by sequence
handling programs
$Id: Patterns.py,v 1.1 2004/10/10 02:53:09 nghoffma Exp $
"""

import re

#to find all single gaps

singleGaps = r"[-.~]"
singleGapsReo = re.compile( singleGaps, re.I)

#fasta recognizing pattern

#fastaPattern = r">(?P<name>[ \t]*\S+)[ \t]*(?P<hea>.*)\n(?P<seq>[-*a-z~.\s]+)\n"
fastaPattern = r">(?P<name>[ \t]*\S+)[ \t]*(?P<hea>.*)\n(?P<seq>[-*a-z~.\s?]+)\n"
fastaPatternReo = re.compile( fastaPattern, re.I )

#selex pattern

#selexPattern = r"^[ \t]*(?P<name>\S+)[ \t]+(?P<seq>[-_*a-z~. ]+)\n"
selexPattern = r"^\s*(?P<name>\S+)\s+(?P<seq>[-_*a-z~. \t]+)\n"

selexPatternReo = re.compile( selexPattern, re.I | re.M )

# recognizes whitespace, digits, and non-alphanumeric characters
allButAlphaRexp = re.compile(r'[^a-zA-Z]')

def describe():
	"""
Patterns.py

Collection of regular expression patterns used by sequence handling programs.
"""
	
	print describe.__doc__

if __name__ == '__main__':
	describe()