Random sequence generator and basic regexp

seq.py

import random
import re

def generate_random_sequence(seq_length=20, n=10):
    """
    Random sequence generator.

    Args:
     seq_length: Int #sequence length
     n: Int #total number of sequences

    Returns:
     barcode_seqs: List<Str>
    """

    n_code = {0: "A", 1: "T", 2: "C", 3: "G"}
    barcode_seqs = []
    for c in range(0, n):
        barcode = ""
        for i in range(0, seq_length):
            base_index = random.randint(0, 3)
            barcode += n_code[base_index]
        barcode_seqs.append(barcode)
    return barcode_seqs


if __name__ == "__main__":
    # generate 500 random sequences (20-mer)
    seqs = generate_random_sequence(n=500)
    
    # define regular expression pattern
    head_ttt_p = re.compile(r"^TTT[ATGC]{17}$")
    tail_ttt_p = re.compile(r"^[ATGC]{17}TTT$")
    A_at_10_p  = re.compile(r"^[ATGC]{9}[A][ATGC]{10}$")
    GCGC_p     = re.compile(r"(?:GC){2}")
    GCGC_or_ATAT_p = re.compile(r"(?:AT){2}|(?:GC){2}")

    for seq in seqs:
        # matching
        if head_ttt_p.match(seq):
            print "TTT in the head -> %s" % (seq)

        if tail_ttt_p.match(seq):
            print "TTT in the tail -> %s" % (seq)

        if A_at_10_p.match(seq):
            print "A that is located at position 10 -> %s" % (seq)

        if GCGC_p.match(seq):
            print "GCGC in any position in the sequence -> %s" % (seq)

        if GCGC_or_ATAT_p.match(seq):
            print "GCGC or ATAT in any position in the sequence -> %s" % (seq)