def get_approximate_frequent_words(text, k, d):
# Flag indicies of candidate kmers from neighborhoods
is_neighbor_flags = [0] * pow(4, k)
for position, kmer in ApproximatePatternMatching_1F.yield_kmers_by_position(text, k):
neighbors = calculate_neighbors(kmer, d)
for neighbor in neighbors:
number = frequency_array_lib.pattern_to_number(neighbor)
is_neighbor_flags[number] = 1
# Count approximate occurences of candidates
chunk_size = 5
translation_table = build_translation_table(chunk_size)
frequency_array = [0] * pow(4, k)
for i in range(len(is_neighbor_flags)):
if 1 == is_neighbor_flags[i]:
kmer = frequency_array_lib.number_to_pattern(i, k)
reverse_complement = generate_strand_complement(kmer, translation_table, chunk_size)
frequency = ApproximatePatternMatching_1F.count_approximate_matches(kmer, text, d) + ApproximatePatternMatching_1F.count_approximate_matches(reverse_complement, text, d)
frequency_array[i] = frequency
# Find max count
max_count = -1
for count in frequency_array:
if count > max_count:
max_count = count
# Find most frequent kmers
frequent = []
for i in range(len(frequency_array)):
if frequency_array[i] == max_count:
kmer = frequency_array_lib.number_to_pattern(i, k)
frequent.append(kmer)
return frequent
def calculate_neighbors(pattern, d):
nucleotides = ['A', 'C', 'G', 'T']
if 0 == d:
return [pattern]
if 1 == len(pattern):
return nucleotides
neighbors = []
first_symbol, suffix = pattern[0:1], pattern[1:]
suffix_neighbors = calculate_neighbors(suffix, d)
for suffix_neighbor in suffix_neighbors:
if ApproximatePatternMatching_1F.calc_hamming_distance(suffix_neighbor, suffix) < d:
for nucleotide in nucleotides:
neighbors.append(nucleotide + suffix_neighbor)
else:
neighbors.append(first_symbol + suffix_neighbor)
return neighbors
