tst.py

from frisk.kmerhash import *
import screed

r = screed.open('/home/kevin/ws/seqs/TAIR10_gen/TAIR10.fasta.gz')
for x in r:
    seq = x.sequence[:100000]
    break

winsz = 3000
ivs = []
for start in range(0, len(seq) - winsz + 1, winsz/2):
    winseq = seq[start:start+winsz]
    kv = build_kmer_vec(1, 8)
    hash_seq(str(winseq), kv)
    kf = kmer_freqs(kv)
    iv = ivom(kf)
    ivs.append(iv.sum() / len(iv))
    if start % 10 == 0:
        print start