search_seq.py

import random;

#Author: Michael Thomas/ Aram Comjean
#Date: Nov/ 2015
#Hungtington's Disease CAG count


# generates random sequence with at least the number of "cag" repeats specified

def generate_random_seq (number_cag_repeats):

    # the length of the total string will be length_k + number_cag_repeats * 3
    length_k = 10000
    # the cag repeats can't start before "off_ends" or within "off_ends" of the end
    # eg if off_ends = 100 the cag repeat will show up between 100 and 900
    off_ends = 200

    # get random distance from start for cag repeats.  Make sure in-frame
    min_cag_start = off_ends
    max_cag_end = length_k - off_ends - (3 * number_cag_repeats)
    distance_from_start = random.randint(min_cag_start, max_cag_end)
    distance_from_start = distance_from_start - distance_from_start%3

    # woh
    # https://answers.yahoo.com/question/index;_ylt=A0LEViwjg0ZWI0gAtJUnnIlQ;_ylu=X3oDMTBydWNmY2MwBGNvbG8DYmYxBHBvcwM0BHZ0aWQDBHNlYwNzcg--?qid=20091007014822AAfsdzT&p=create%20string%20from%20repeat%20sequence%20python
    cag_repeat_string =  ('CAG' * number_cag_repeats)

    #create sequence

    cag_sequence = ''

    for x in range(length_k):
        # build string
        cag_sequence +=  random.choice('ACTG');

    #string insert the cag repeats

    #http://stackoverflow.com/questions/4022827/how-to-insert-some-string-in-the-given-string-at-given-index-in-python
    cag_sequence = cag_sequence[:distance_from_start] + cag_repeat_string + cag_sequence[distance_from_start:]

    return cag_sequence

#get random sequence with CAGs here
rand_seq = generate_random_seq(49)

#list compreension to split seq into codons
codons = [rand_seq[i:i+3] for i in range(0,len(rand_seq),3)]

#initialize count/pos
count = 0
pos = 0

#loop to check for CAGs and count
#if they are concurrant

max_in_row = 0;
max_pos = 0;

for i in codons:
    
    #print pos ,' - ', codons[pos]
    
    if codons[pos] == 'CAG':
        if (count == 0):
            current_repeat_location = pos
        count = count + 1
        if (count>max_in_row):
            max_in_row = count
            max_pos = current_repeat_location

    else:
        count =0;
    pos = pos + 1


print "# of 'CAG' found in a row: ",max_in_row
print "# of 'CAG' reparts start at codon: ",max_pos