subpattern filter

Question

I'd like to generate a pattern for a python script, where any number of three words must exist in a specified pattern?

for example, given a sequence:

ATG GTC TGA CGA CGG CAG TAA AAA AAA GGG TGG GCA GCC TTT GAA GCC TTT

I'd like to find all occurrences of 19-21mers that contain at least one of any of the following words: TAG, TGA, or TAA

I tried to specify a pattern = '[A,G,C,T,\s]{21,26}^.*\b(TAT|TGA|CCC)\b.*$'

But it doesn't seem to work and I'm sure I'm doing something that shows what a noob I am.

Answer 1

I don't think regex is expressive enough to handle this with the length requirement.

However, you can break down this problem by using a window iterator to simulate an open read frame:

# From http://stackoverflow.com/questions/6822725/rolling-or-sliding-window-iterator-in-python:

from itertools import islice

def window(seq, n=2):
    "Returns a sliding window (of width n) over data from the iterable"
    "   s -> (s0,s1,...s[n-1]), (s1,s2,...,sn), ...                   "
    it = iter(seq)
    result = tuple(islice(it, n))
    if len(result) == n:
        yield result
    for elem in it:
        result = result[1:] + (elem,)
        yield result

sequence = "ATG GTC TGA CGA CGG CAG TAA AAA AAA GGG TGG GCA GCC TTT GAA GCC TTT"
codons = sequence.split()

orf = window(codons, 7)
matching_codons = ['TGA', 'TAA', 'TAG']

[sequence for sequence in orf if any(codon in matching_codons for codon in sequence)]

Dissecting the code

orf = window(codons, 7)

This defines a generator which will return each frame of length 7, moving the frame by 1 each iteration.

Then, the list comprehension iterates does two things.

1. It iterates over each sequence in our ORF:

   [sequence for sequence in orf] # returns all possible frames of length 7 in sequence

2. It filters the result, only returning sequences that contain any of the valid codons:

   [sequence for sequence in orf if any(codon in ['TGA', 'TAA', 'TAG'] for codon in sequence)] # Only matches sequences matching 'TGA', 'TAA', or 'TAG'

Finally, if you want the result to be the substrings themselves, use the following list comprension:

[' '.join(sequence) for sequence in window(codons, 7) if any(codon in ['TGA', 'TAA', 'TAG'] for codon in sequence)]

Result:

['ATG GTC TGA CGA CGG CAG TAA', 'GTC TGA CGA CGG CAG TAA AAA', 'TGA CGA CGG CAG TAA AAA AAA', 'CGA CGG CAG TAA AAA AAA GGG', 'CGG CAG TAA AAA AAA GGG TGG', 'CAG TAA AAA AAA GGG TGG GCA', 'TAA AAA AAA GGG TGG GCA GCC']

Answer 2

import re
string_to_read="ATG GTC TGA CGA CGG CAG TAA AAA AAA GGG TGG GCA GCC TTT GAA GCC TTT"
res=re.search('(TAT|TGA|TAA)', string_to_read)
if res:
    print('matched %s'%res.groups())

This regex will tell you if any of those 3 sequences exist in the string you are testing against.

If you need to match that all 3 exist, you could test for all 3 independently

if re.match('TAT', string_to_read) and re.match('TGA', string_to_read) and re.match('TAA', string_to_read):
    print('has all 3')

instead of some clever regex with all the combinatorics of those 3 sequences. If you don't wanna run 3 separate regexes, you could do something with a regex like (TAT)|(TGA)|(AAA) and then keeping a tally of all the groups you matched, see if you've hit all of them.