bfast_index.py

def _index_bfast(ref_file):
    """Indexes bfast in color and nucleotide space for longer reads.

    This preps for 40+bp sized reads, which is bfast's strength.
    """
    dir_name = "bfast"
    window_size = 14
    bfast_nt_masks = [
   "1111111111111111111111",
   "1111101110111010100101011011111",
   "1011110101101001011000011010001111111",
   "10111001101001100100111101010001011111",
   "11111011011101111011111111",
   "111111100101001000101111101110111",
   "11110101110010100010101101010111111",
   "111101101011011001100000101101001011101",
   "1111011010001000110101100101100110100111",
   "1111010010110110101110010110111011",
    ]
    bfast_color_masks = [
    "1111111111111111111111",
    "111110100111110011111111111",
    "10111111011001100011111000111111",
    "1111111100101111000001100011111011",
    "111111110001111110011111111",
    "11111011010011000011000110011111111",
    "1111111111110011101111111",
    "111011000011111111001111011111",
    "1110110001011010011100101111101111",
    "111111001000110001011100110001100011111",
    ]
    local_ref = os.path.split(ref_file)[-1]
    if not exists(dir_name):
        run("mkdir %s" % dir_name)
        with cd(dir_name):
            run("ln -s %s" % os.path.join(os.pardir, ref_file))
            # nucleotide space
            run("bfast fasta2brg -f %s -A 0" % local_ref)
            for i, mask in enumerate(bfast_nt_masks):
                run("bfast index -d 1 -n 4 -f %s -A 0 -m %s -w %s -i %s" %
                        (local_ref, mask, window_size, i + 1))
            # colorspace
            run("bfast fasta2brg -f %s -A 1" % local_ref)
            for i, mask in enumerate(bfast_color_masks):
                run("bfast index -d 1 -n 4 -f %s -A 1 -m %s -w %s -i %s" %
                        (local_ref, mask, window_size, i + 1))