create.py

import random
from StringIO import StringIO
import string
import zlib
import base64 as b64


def get_kmer(read_str, overlap_size, kmer_size):
  if read_str.pos > overlap_size:
    read_str.seek(overlap_size * -1, 1)
    front_overlap = read_str.read(overlap_size)
  else:
    front_overlap = ''
  read = read_str.read(kmer_size)
  try:
    rear_overlap = read_str.read(overlap_size)
    read_str.seek(overlap_size * -1, 1)
  except:
    rear_overlap = ''
  return front_overlap, read, rear_overlap


read_str = StringIO('abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789')
read_str = StringIO(''.join(random.choice(string.ascii_lowercase + string.ascii_uppercase + string.digits) for _ in range(9*10240)))
read_str = StringIO(b64.b64encode(zlib.compress(open("./test", 'r').read())))
kmers = []

flag = 'some_placeholder_string'
while flag:
  kmer_size = random.choice(range(100, 200))
  overlap_size = random.choice(range(10, 15))
  front_overlap, kmer, rear_overlap = get_kmer(read_str, overlap_size, kmer_size)
  kmers.append("%s%s%s" % (front_overlap, kmer, rear_overlap))
  flag = rear_overlap

random.shuffle(kmers)
for kmer in kmers:
  print kmer