k-nucleotide

#!/usr/bin/python3
import collections
import itertools
import sys

def count(counter, dna_seq, length):
    return counter.update(
        dna_seq[i:i + length] for i in range(len(dna_seq) - length + 1))

def percentage(counter, keys):
    total = sum(counter[key] for key in keys)
    for occurences, key in sorted((-counter[key], key) for key in keys):
        print("%s %.3f" % (key, -occurences * 100.0 / total))
    print()

it = itertools.dropwhile(lambda l: not l.startswith(">THREE"), sys.stdin)
next(it)
it = itertools.takewhile(lambda l: not l.startswith(">"), it)
dna_seq = ''.join(l.strip().upper() for l in it)
counter = collections.Counter()
for length in (1, 2, 3, 4, 6, 12, 18):
    count(counter, dna_seq, length)
bases = "ATCG"
percentage(counter, bases)
percentage(counter, tuple(i + j for i in bases for j in bases))
for seq in ("GGT", "GGTA", "GGTATT", "GGTATTTTAATT", "GGTATTTTAATTTATAGT"):
    print("%-7d %s" % (counter[seq], seq))