def __call__(self, seqrecords):
'It filters out the seqrecords not found in the list.'
stats = self._stats
threshold = self.threshold
reverse = self.reverse
ignore_masked = self.ignore_masked
stats[PROCESSED_PACKETS] += 1
processed_seqs = []
for seqrecord in seqrecords:
stats[PROCESSED_SEQS] += 1
try:
quals = seqrecord.letter_annotations['phred_quality']
except KeyError:
msg = 'Some of the input sequences do not have qualities: {}'
msg = msg.format(seqrecord.id)
raise WrongFormatError(msg)
if ignore_masked:
seq = str(seqrecord.seq)
seg_quals = [quals[segment[0]: segment[1] + 1]
for segment in get_uppercase_segments(seq)]
qual = sum(sum(q) * len(q) for q in seg_quals) / len(quals)
else:
qual = sum(quals) / len(quals)
passed = True if qual >= threshold else False
if reverse:
passed = not(passed)
if passed:
processed_seqs.append(seqrecord)
stats[YIELDED_SEQS] += 1
return processed_seqs
def _do_check(self, seq):
seq_object = seq.object
try:
quals = seq_object.letter_annotations['phred_quality']
except KeyError:
msg = 'Some of the input sequences do not have qualities: {}'
msg = msg.format(get_name(seq))
raise WrongFormatError(msg)
if self.ignore_masked:
str_seq = str(seq_object.seq)
seg_quals = [quals[segment[0]: segment[1] + 1]
for segment in get_uppercase_segments(str_seq)]
qual = sum(sum(q) * len(q) for q in seg_quals) / len(quals)
else:
qual = sum(quals) / len(quals)
return True if qual >= self.threshold else False
def _do_check(self, seq):
seq_object = seq.object
try:
quals = seq_object.letter_annotations['phred_quality']
except KeyError:
msg = 'Some of the input sequences do not have qualities: {}'
msg = msg.format(get_name(seq))
raise WrongFormatError(msg)
if self.ignore_masked:
str_seq = str(seq_object.seq)
seg_quals = [quals[segment[0]: segment[1] + 1]
for segment in get_uppercase_segments(str_seq)]
qual = sum(sum(q) * len(q) for q in seg_quals) / len(quals)
else:
qual = sum(quals) / len(quals)
return True if qual >= self.threshold else False
def test_masked_locations():
"It test the masked locations function"
assert list(get_uppercase_segments("aaATTTTTTaa")) == [(2, 8)]
assert list(get_uppercase_segments("aaATTTaTTaa")) == [(2, 5), (7, 8)]
assert list(get_uppercase_segments("AAATaaa")) == [(0, 3)]
assert list(get_uppercase_segments("aaaaAAAA")) == [(4, 7)]
seq = "AATTaaTTaaTTT"
assert list(get_uppercase_segments(seq)) == [(0, 3), (6, 7), (10, 12)]
assert list(get_uppercase_segments("AATT")) == [(0, 3)]
assert not list(get_uppercase_segments("aatt"))
def test_masked_locations():
'It test the masked locations function'
assert list(get_uppercase_segments('aaATTTTTTaa')) == [(2, 8)]
assert list(get_uppercase_segments('aaATTTaTTaa')) == [(2, 5), (7, 8)]
assert list(get_uppercase_segments('AAATaaa')) == [(0, 3)]
assert list(get_uppercase_segments('aaaaAAAA')) == [(4, 7)]
seq = 'AATTaaTTaaTTT'
assert list(get_uppercase_segments(seq)) == [(0, 3), (6, 7), (10, 12)]
assert list(get_uppercase_segments('AATT')) == [(0, 3)]
assert not list(get_uppercase_segments('aatt'))
def test_masked_locations():
'It test the masked locations function'
assert list(get_uppercase_segments('aaATTTTTTaa')) == [(2, 8)]
assert list(get_uppercase_segments('aaATTTaTTaa')) == [(2, 5), (7, 8)]
assert list(get_uppercase_segments('AAATaaa')) == [(0, 3)]
assert list(get_uppercase_segments('aaaaAAAA')) == [(4, 7)]
seq = 'AATTaaTTaaTTT'
assert list(get_uppercase_segments(seq)) == [(0, 3), (6, 7), (10, 12)]
assert list(get_uppercase_segments('AATT')) == [(0, 3)]
assert not list(get_uppercase_segments('aatt'))
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
