knucleotide.py

from sys import stdin
from collections import defaultdict

def gen_freq(seq, frame, frequences):
    ns = len(seq) + 1 - frame
    frequences.clear()
    for ii in xrange(ns):
        nucleo = seq[ii:ii + frame]
        frequences[nucleo] += 1
    return ns, frequences


def sort_seq(seq, length, frequences):
    n, frequences = gen_freq(seq, length, frequences)
    l = sorted(((freq, seq) for (seq, freq) in frequences.items()), reverse=True)

    print '\n'.join("%s %.3f" % (st, 100.0*fr/n) for (fr, st) in l)
    print


def find_seq(seq, s, frequences):
    n, t = gen_freq(seq, len(s), frequences)
    print "%d\t%s" % (t[s], s)


def main():
    frequences = defaultdict(int)
    for line in stdin:
        if line[:3] == ">TH":
            break

    seq = []
    for line in stdin:
        if line[0] in ">;":
            break
        seq.append( line[:-1] )
    sequence = "".join(seq).upper()

    for nl in 1,2:
        sort_seq(sequence, nl, frequences)

    for se in "GGT GGTA GGTATT GGTATTTTAATT GGTATTTTAATTTATAGT".split():
        find_seq(sequence, se, frequences)

main()