def mainKmerFinder(_seq, _k, _d):
kmers = kmer_finder_mismatches(_seq, _k, _d)
seq_kmers = list()
for i in xrange(len(_seq) - _k + 1):
seq_kmers.append(_seq[i : i + _k])
seq_kmers.append(reverse(_seq[i : i + _k]))
output = list()
for kmer in kmers:
if kmer in seq_kmers:
output.append(kmer)
return output
def kmer_finder_mismatches(seq, k, d):
kmer_freq = defaultdict(int)
for i in range(len(seq)-k+1):
kmer_freq[seq[i:i+k]] += 1
mismatch_count = defaultdict(int)
for kmer, freq in kmer_freq.iteritems():
temp_mutations = set()
for mismatch in mutations(kmer, d):
temp_mutations.add(mismatch)
for mismatch in temp_mutations:
mismatch_count[mismatch] += freq
temp_mutations = set()
for mismatch in mutations(reverse(kmer), d):
temp_mutations.add(mismatch)
for mismatch in temp_mutations:
mismatch_count[mismatch] += freq
max_count = max(mismatch_count.values())
return sorted([kmer for kmer, count in mismatch_count.iteritems() if count == max_count])