def test_string_with_many_gaps_returns_original_without_gap(self):
seq = "A-CGT--"
actual = remove_gaps(seq)
expected = "ACGT"
self.assertEqual(actual, expected)
def test_empty_string_returns_empty(self):
seq = ""
actual = remove_gaps(seq)
expected = ""
self.assertEqual(actual, expected)
def test_string_with_no_gaps_returns_original(self):
seq = "ACGT"
actual = remove_gaps(seq)
expected = seq
self.assertEqual(actual, expected)
def test_all_input_space_doesnt_break(self, seq):
actual = remove_gaps(seq)
self.assertFalse("-" in actual)
def interval_partition(self):
"""Return a list of intervals in which we have
consensus sequence longer than min_match_length, and
a list of the non-match intervals left."""
match_intervals = []
non_match_intervals = []
match_count, match_start, non_match_start = 0, 0, 0
logging.debug("consensus: %s" % self.consensus)
for i in range(self.length):
letter = self.consensus[i]
if letter != "*":
# In a match region.
if match_count == 0:
match_start = i
match_count += 1
elif match_count > 0:
# Have reached a non-match. Check if previous match string is long enough to add to match_regions
match_string = remove_gaps(
self.consensus[match_start:match_start + match_count])
match_len = len(match_string)
logging.debug("have match string %s" % match_string)
if match_len >= self.min_match_length:
if non_match_start < match_start:
non_match_intervals.append(
[non_match_start, match_start - 1])
logging.debug(
f"add non-match interval [{non_match_start},{match_start - 1}]"
)
end = match_start + match_count - 1
match_intervals.append([match_start, end])
logging.debug(f"add match interval [{match_start},{end}]")
non_match_start = i
match_count = 0
match_start = non_match_start
end = self.length - 1
if self.length < self.min_match_length:
# Special case: a short sequence can still get classified as a match interval
added_interval = "match" if "*" in self.consensus else "non_match"
if added_interval == "match":
match_intervals.append([0, end])
else:
non_match_intervals.append([0, end])
logging.debug(
f"add whole short {added_interval} interval [0,{end}]")
match_count = 0
non_match_start = end + 1
# At end add last intervals
if match_count > 0:
if match_count >= self.min_match_length:
match_intervals.append([match_start, end])
logging.debug(
f"add final match interval [{match_start},{end}]")
if non_match_start < match_start:
end = match_start - 1
if match_count != self.length and non_match_start <= end:
non_match_intervals.append([non_match_start, end])
logging.debug(f"add non-match interval [{non_match_start},{end}]")
# check all stretches of consensus are in an interval, and intervals don't overlap
for i in range(self.length):
count_match = 0
for interval in match_intervals:
if interval[0] <= i <= interval[1]:
count_match += 1
count_non_match = 0
for interval in non_match_intervals:
if interval[0] <= i <= interval[1]:
count_non_match += 1
assert count_match | count_non_match, (
"Failed to correctly identify match intervals: position %d "
"appeared in both/neither match and non-match intervals" % i)
assert count_match + count_non_match == 1, (
"Failed to correctly identify match intervals: position "
"%d appeared in %d intervals" %
(i, count_match + count_non_match))
return match_intervals, non_match_intervals