def test_get_longest_section():
'It gets the longest section from a list of sections'
segments = [(0, 3), (10, 34)]
assert (10, 34) == get_longest_segment(segments)
segments = [(0, 3), (10, 13)]
segment = get_longest_segment(segments)
assert segment == (0, 3) or segment == (10, 13)
def test_get_longest_section():
'It gets the longest section from a list of sections'
segments = [(0, 3), (10, 34)]
assert (10, 34) == get_longest_segment(segments)
segments = [(0, 3), (10, 13)]
segment = get_longest_segment(segments)
assert segment == (0, 3) or segment == (10, 13)
def _do_trim(self, seq):
str_seq = get_str_seq(seq)
unmasked_segments = get_uppercase_segments(str_seq)
segment = get_longest_segment(unmasked_segments)
if segment is not None:
segments = []
if segment[0] != 0:
segments.append((0, segment[0] - 1))
len_seq = len(str_seq)
if segment[1] != len_seq - 1:
segments.append((segment[1] + 1, len_seq - 1))
_add_trim_segments(segments, seq, kind=OTHER)
else:
segments = [(0, len(seq))]
_add_trim_segments(segments, seq, kind=OTHER)
return seq
def _do_trim(self, seq):
str_seq = get_str_seq(seq)
unmasked_segments = get_uppercase_segments(str_seq)
segment = get_longest_segment(unmasked_segments)
if segment is not None:
segments = []
if segment[0] != 0:
segments.append((0, segment[0] - 1))
len_seq = len(str_seq)
if segment[1] != len_seq - 1:
segments.append((segment[1] + 1, len_seq - 1))
_add_trim_segments(segments, seq, kind=OTHER)
else:
segments = [(0, len(seq))]
_add_trim_segments(segments, seq, kind=OTHER)
return seq
def __call__(self, seqs):
'It trims the masked segments of the seqrecords.'
trimmed_seqs = []
for seq in seqs:
str_seq = get_str_seq(seq)
unmasked_segments = get_uppercase_segments(str_seq)
segment = get_longest_segment(unmasked_segments)
if segment is not None:
segments = []
if segment[0] != 0:
segments.append((0, segment[0] - 1))
len_seq = len(str_seq)
if segment[1] != len_seq - 1:
segments.append((segment[1] + 1, len_seq - 1))
_add_trim_segments(segments, seq, kind=OTHER)
trimmed_seqs.append(seq)
return trimmed_seqs
def __call__(self, seqrecords):
'It trims the masked segments of the seqrecords.'
stats = self._stats
stats[PROCESSED_PACKETS] += 1
trimmed_seqs = []
for seqrecord in seqrecords:
stats[PROCESSED_SEQS] += 1
seq = str(seqrecord.seq)
unmasked_segments = get_uppercase_segments(seq)
segment = get_longest_segment(unmasked_segments)
if segment is not None:
stats[YIELDED_SEQS] += 1
segments = []
if segment[0] != 0:
segments.append((0, segment[0] - 1))
len_seq = len(seqrecord)
if segment[1] != len_seq - 1:
segments.append((segment[1] + 1, len_seq - 1))
_add_trim_segments(segments, seqrecord, kind=OTHER)
trimmed_seqs.append(seqrecord)
#trimmed_seqs.append(seqrecord[segment[0]:segment[1] + 1])
return trimmed_seqs
