def compare(a, b):
"""Return cost of comparing a to b"""
l = min(len(a.sequence), len(b.sequence))
length_diff = max(len(a.sequence), len(b.sequence)) - l
dist_prefixes = hamming_distance(a.sequence[:l], b.sequence[:l])
dist_suffixes = hamming_distance(a.sequence[-l:], b.sequence[-l:])
return 5 * min(dist_prefixes, dist_suffixes) + length_diff
def naive_has_similar(self, s, distance):
for t in self.strings:
if len(t) != len(s):
continue
if hamming_distance(t, s) <= distance:
return True
return False
def is_similar_with_junction(s, t, mismatches, cdr3_core):
"""
Return whether strings s and t have at most the given number of mismatches
*and* have at least one identical junction.
"""
# TODO see issue #81
if len(s) != len(t):
return False
if 0 < mismatches < 1:
delta = cdr3_core.start if cdr3_core is not None else 0
distance_ok = hamming_distance(s, t) <= (len(s) - delta) * mismatches
else:
distance_ok = hamming_distance(s, t) <= mismatches
if cdr3_core is None:
return distance_ok
return distance_ok and ((s[:cdr3_core.start] == t[:cdr3_core.start]) or
(s[cdr3_core.stop:] == t[cdr3_core.stop:]))
def is_similar(s, t):
# m = max_hamming
if '-' in s or '-' in t:
# Remove suffix and/or prefix where sequences do not overlap
s = s.lstrip('-')
t = t[-len(s):]
s = s.rstrip('-')
if len(s) < min_overlap:
return False
t = t[:len(s)]
# TODO allowed Hamming distance should be reduced relative to the overlap length
# m = max_hamming * len(s) / len(original_length_of_s)
return hamming_distance(s, t) <= max_hamming
def print_similar(a, b, colored: bool):
l = min(len(a.sequence), len(b.sequence))
dist_prefixes = hamming_distance(a.sequence[:l], b.sequence[:l])
dist_suffixes = hamming_distance(a.sequence[-l:], b.sequence[-l:])
if dist_prefixes <= dist_suffixes:
a_prefix = ''
b_prefix = ''
a_common = a.sequence[:l]
b_common = b.sequence[:l]
a_suffix = a.sequence[l:]
b_suffix = b.sequence[l:]
else:
a_prefix = a.sequence[:-l]
b_prefix = b.sequence[:-l]
a_common = a.sequence[-l:]
b_common = b.sequence[-l:]
a_suffix = ''
b_suffix = ''
s = format_indel(a_prefix, b_prefix, colored)
edits = []
for i, (ac, bc) in enumerate(zip(a_common, b_common)):
if ac != bc:
if colored:
s = '{' + red(ac) + ' → ' + green(bc) + '}'
else:
s = '{' + ac + ' → ' + bc + '}'
edits.append(s)
else:
edits.append(ac)
s += ''.join(edits)
s += format_indel(a_suffix, b_suffix, colored)
print('~', a.name, '--', b.name)
print(s)
print()
def _aa_mutations(self):
# Earlier versions of this code used edit distance to compute the number of mutations,
# but some FR1 alignments are reported with a frameshift by IgBLAST. By requiring that
# reference and query lengths are identical, we can filter out these cases (and use
# Hamming distance to get some speedup)
if (not self.aa_reference or not self.aa_sequence
or len(self.nt_sequence) != len(self.nt_reference)):
return None
dist = hamming_distance(self.aa_reference, self.aa_sequence)
# If the mutation rate is still obviously too high, assume something went
# wrong and ignore the computed value
if dist / len(self.aa_reference) >= 0.8:
return None
return dist
def set_aa_mut_columns(record, database):
"""
Compute amino acid mutation rate for all regions on V and also for V
itself as the sum of the regions (that is, excluding the CDR3)
"""
total_length = 0
total_dist = 0
n_regions = 0
for airr_col, region in (
("fwr1", "FR1"),
("cdr1", "CDR1"),
("fwr2", "FR2"),
("cdr2", "CDR2"),
("fwr3", "FR3"),
):
record[region + "_aa_mut"] = None
start = record[airr_col + "_start"]
end = record[airr_col + "_end"]
if start is None or end is None:
continue
sequence_aa = nt_to_aa(record["sequence"][start - 1 : end])
germline_aa = database.v_regions_aa[record["v_call"]].get(region)
if germline_aa is None:
continue
# Some FR1 alignments are reported with a frameshift by IgBLAST. By requiring that
# reference and query lengths are identical, we can filter out these cases (and use
# Hamming distance to get some speedup)
if len(germline_aa) != len(sequence_aa):
continue
dist = hamming_distance(germline_aa, sequence_aa)
mut_aa = dist / len(germline_aa)
if mut_aa >= 0.8:
# assume something went wrong
continue
total_dist += dist
n_regions += 1
total_length += len(germline_aa)
record[region + "_aa_mut"] = 100.0 * mut_aa
if n_regions == 5:
record["V_aa_mut"] = 100.0 * total_dist / total_length
else:
record["V_aa_mut"] = None
def naive_find_all_similar(self, s, distance):
for t in self.strings:
if len(t) != len(s):
continue
if hamming_distance(t, s) <= distance:
yield t
def naive_has_similar(t):
for s in strings:
if hamming_distance(s, t) <= dist:
return True
return False
def linked(s, t):
return hamming_distance(s, t) <= mismatches
def test_hamming_distance_incorrect_length():
with pytest.raises(IndexError):
hamming_distance('A', 'BC')
def test_hamming_distance():
assert hamming_distance('', '') == 0
assert hamming_distance('A', 'A') == 0
assert hamming_distance('HELLO', 'HELLO') == 0
assert hamming_distance('ABC', 'DEF') == 3
assert hamming_distance('ABCXDEF', 'ABCYDEF') == 1
