def frequent_words_with_mismatches_v2(text, k, hamming_dist):
"""(Failed) attempt to improve the efficiency of v1"""
#initialize frequency array
frequency_array = list()
for index in range(4 ** k):
frequency_array.append(0)
#list all patterns in text
patterns_in_text = list()
for index in range(len(text)-k+1):
pattern = text[index:index+k]
patterns_in_text.append(pattern)
#create list of all possible kmers
possible_kmers = all_possible_kmers(k)
#go through patterns and compare to kmers
for pattern in patterns_in_text:
for kmer in possible_kmers:
if patterncount.hamming_distance(pattern, kmer) <= hamming_dist:
frequency_array[pattern_to_number(kmer)] = frequency_array[pattern_to_number(kmer)] + 1
frequent_patterns = set([])
max_count = max(frequency_array)
for index in range(4 ** k):
if frequency_array[index] == max_count:
pattern = number_to_pattern(index, k)
frequent_patterns = frequent_patterns.union(set([pattern]))
return (frequent_patterns, max_count)
def frequent_words_with_mismatches_v1(text, k, hamming_dist):
"""Find the most frequent word(s) of length within a text, allowing up to
hamming_dist mismatches. N.B. that the most frequent word need not actually
occur in the text -- for example, input ("ATATA", 3, 2) returns 10 distinct
patterns that occur twice, despite the fact that there are only two distinct
3-mers that actually occur in the text, and one of those patterns that
actually does occur in the text ('TAT') is not one of the most frequent
3-mers. Runs very inefficiently; use v3 instead!"""
#enumerate all possible kmers
possible_kmers = set(all_possible_kmers(k))
#enumerate kmers that actually occur in text
patterns_in_text = list()
for index in range(len(text)-k+1):
pattern = text[index:index+k]
patterns_in_text.append(pattern)
print patterns_in_text
#pare down to those that are within hamming_dist of each pattern in text
relevant_kmers = set([])
for kmer in possible_kmers:
for pattern in patterns_in_text:
if patterncount.hamming_distance(pattern, kmer) <= hamming_dist:
relevant_kmers = relevant_kmers.union(set([kmer]))
break
else:
pass
relevant_kmers = relevant_kmers.union(patterns_in_text) #add all patterns that actually occur in text!
#initialize frequency array
frequency_array = list()
for index in range(4 ** k):
frequency_array.append(0)
#loop through text and count frequencies
for pattern in patterns_in_text:
pattern_list = [kmer for kmer in relevant_kmers if patterncount.hamming_distance(kmer, pattern) <= hamming_dist]
pattern_list = set(pattern_list)
pattern_list = pattern_list.union(set([pattern]))
code_list = [pattern_to_number(pat) for pat in pattern_list]
for code in code_list:
frequency_array[code] = frequency_array[code] + 1
#count most frequent patterns
frequent_patterns = set([])
max_count = max(frequency_array)
for index in range(4 ** k):
if frequency_array[index] == max_count:
pattern = number_to_pattern(index, k)
frequent_patterns = frequent_patterns.union(set([pattern]))
return (frequent_patterns, max_count)
