def process(in_filename, out_filename):
in_file = open(in_filename)
out_file = open(out_filename, 'w')
header, sequence = utils.load(in_file)
in_file.close()
out_file.write(header + '\n')
out_file.write('\n'.join(utils.prepare_subsequences(utils.translate(sequence), 80)))
out_file.close()
def create_from_sequence(sequence):
states = list(set(sequence))
observations = list(set([utils.translate(g) for g in states]))
hmm = HMM(states, observations)
gene_count = {g: {h: 0.0 for h in states} for g in states}
for i, g in enumerate(sequence[:-1]):
gene_count[g][sequence[i + 1]] += 1.0
def normalize(value_dict):
n = float(sum(value_dict.values()))
if n == 0:
n = 1
return {k: v / n for k, v in value_dict.items()}
gene_count = {k: normalize(v) for k, v in gene_count.items()}
emissions = {g: {o: 1.0 if utils.translate(g) == o else 0.0 for o in observations} for g in states}
hmm.emissions = emissions
hmm.transitions = gene_count
return hmm
def check_translation(gene, aminoacid):
assert utils.translate(gene) == aminoacid
