def test_trim_seqs():
'It tests the trim seq function'
seqs = []
seqs.append([SeqWrapper(SEQRECORD, SeqRecord(Seq('aaCTTTC')), None)])
seqs.append([SeqWrapper(SEQRECORD, SeqRecord(Seq('CTTCaa')), None)])
seqs.append([SeqWrapper(SEQRECORD, SeqRecord(Seq('aaCTCaa')), None)])
seqs.append([SeqWrapper(SEQRECORD, SeqRecord(Seq('actg')), None)])
seqs.append([SeqWrapper(SEQRECORD, SeqRecord(Seq('AC')), None)])
trim_packet = {SEQS_PASSED: seqs, ORPHAN_SEQS: []}
trim_lowercased_seqs = TrimLowercasedLetters()
trim = TrimOrMask()
# pylint: disable=W0141
trim_packet = trim(trim_lowercased_seqs(trim_packet))
res = [get_str_seq(s) for l in trim_packet[SEQS_PASSED] for s in l]
assert res == ['CTTTC', 'CTTC', 'CTC', 'AC']
seqs = []
seq = SeqItem('s', ['>s\n', 'aaCTTTC\n'])
seqs.append([SeqWrapper(SEQITEM, seq, 'fasta')])
trim_packet = {SEQS_PASSED: seqs, ORPHAN_SEQS: []}
trim_packet = trim(trim_lowercased_seqs(trim_packet))
res = [get_str_seq(s) for l in trim_packet[SEQS_PASSED] for s in l]
assert res == ['CTTTC']
# with pairs
seq = SeqItem('s.f', ['>s.f\n', 'aaCTTTC\n'])
seq1 = SeqItem('s.r', ['>s.r\n', 'aaCTTTC\n'])
seq2 = SeqItem('s1.f', ['>s1.f\n', 'aa\n'])
seq3 = SeqItem('s1.r', ['>s1.r\n', 'aaCTTTC\n'])
seqs = []
seqs.append([SeqWrapper(SEQITEM, seq, 'fasta'),
SeqWrapper(SEQITEM, seq1, 'fasta')])
seqs.append([SeqWrapper(SEQITEM, seq2, 'fasta'),
SeqWrapper(SEQITEM, seq3, 'fasta')])
trim_packet = {SEQS_PASSED: seqs, ORPHAN_SEQS: []}
trim_lowercased_seqs = TrimLowercasedLetters()
trim = TrimOrMask()
# pylint: disable=W0141
trim_packet = trim(trim_lowercased_seqs(trim_packet))
res = [get_str_seq(s) for l in trim_packet[SEQS_PASSED] for s in l]
orphan_res = [get_str_seq(s) for s in trim_packet[ORPHAN_SEQS]]
assert orphan_res == ['CTTTC']
assert ['CTTTC', 'CTTTC'] == res
# no drag
trim_packet = {SEQS_PASSED: seqs, ORPHAN_SEQS: []}
trim_lowercased_seqs = TrimLowercasedLetters()
trim = TrimOrMask()
# pylint: disable=W0141
trim_packet = trim(trim_lowercased_seqs(trim_packet))
res = [get_str_seq(s) for l in trim_packet[SEQS_PASSED] for s in l]
orphan_res = [get_name(s) for s in trim_packet[ORPHAN_SEQS]]
assert orphan_res == ['s1.r']
assert ['CTTTC', 'CTTTC'] == res
def test_trimming(self):
'The sequences are trimmed according to the recommendations.'
seq1 = 'gggtctcatcatcaggg'.upper()
seq = SeqRecord(Seq(seq1), annotations={TRIMMING_RECOMMENDATIONS: {}})
seq = SeqWrapper(SEQRECORD, seq, None)
seqs = [seq]
trim_packet = {SEQS_PASSED: [seqs], ORPHAN_SEQS: []}
trim_rec = get_annotations(seq)[TRIMMING_RECOMMENDATIONS]
seq_trimmer = TrimOrMask()
trim_rec['vector'] = [(0, 3), (8, 13)]
get_annotations(seq)[TRIMMING_RECOMMENDATIONS] = trim_rec
trim_packet2 = seq_trimmer(trim_packet)
res = [get_str_seq(s) for l in trim_packet2[SEQS_PASSED] for s in l]
assert res == ['CTCA']
trim_rec['vector'] = [(0, 0), (8, 13)]
get_annotations(seq)[TRIMMING_RECOMMENDATIONS] = trim_rec
trim_packet2 = seq_trimmer(trim_packet)
res = [get_str_seq(s) for l in trim_packet2[SEQS_PASSED] for s in l]
assert res == ['GGTCTCA']
trim_rec['vector'] = [(0, 1), (8, 12)]
trim_rec['quality'] = [(1, 8), (13, 17)]
get_annotations(seq)[TRIMMING_RECOMMENDATIONS] = trim_rec
trim_packet2 = seq_trimmer(trim_packet)
assert not trim_packet2[SEQS_PASSED]
trim_rec['vector'] = [(0, 0), (8, 13)]
trim_rec['quality'] = []
get_annotations(seq)[TRIMMING_RECOMMENDATIONS] = trim_rec
trim_packet2 = seq_trimmer(trim_packet)
res = [get_str_seq(s) for l in trim_packet2[SEQS_PASSED] for s in l]
assert res == ['GGTCTCA']
trim_packet2[SEQS_PASSED][0][0]
assert TRIMMING_RECOMMENDATIONS not in get_annotations(
trim_packet2[SEQS_PASSED][0][0])
def test_quality_trimming(self):
'It trims the edges'
trim = TrimOrMask()
trim_quality = TrimByQuality(window=5, threshold=30)
seq = SeqRecord(Seq('ACTGCTGCATAAAA'))
quals = [10, 10, 20, 30, 30, 30, 40, 40, 30, 30, 20, 20, 10, 10]
seq.letter_annotations['phred_quality'] = quals
seq = SeqWrapper(SEQRECORD, seq, None)
trim_packet = {SEQS_PASSED: [[seq]], ORPHAN_SEQS: []}
trim_packet2 = trim(trim_quality(trim_packet))
seq2 = trim_packet2[SEQS_PASSED][0][0]
assert get_int_qualities(seq2) == [20, 30, 30, 30, 40, 40, 30, 30, 20]
# all bad
trim_quality = TrimByQuality(window=5, threshold=60)
trim_packet = {SEQS_PASSED: [[seq]], ORPHAN_SEQS: []}
trim_packet2 = trim(trim_quality(trim_packet))
assert not trim_packet2[SEQS_PASSED]
# all OK
trim_quality = TrimByQuality(window=5, threshold=5)
trim_packet = {SEQS_PASSED: [[seq]], ORPHAN_SEQS: []}
trim_packet2 = trim(trim_quality(trim_packet))
seq2 = trim_packet2[SEQS_PASSED][0][0]
assert get_int_qualities(seq2) == quals
seq = SeqRecord(Seq('ACTGCTGCATAA'))
quals = [20, 20, 20, 60, 60, 60, 60, 60, 20, 20, 20, 20]
trim_quality = TrimByQuality(window=5, threshold=50)
seq.letter_annotations['phred_quality'] = quals
seq = SeqWrapper(SEQRECORD, seq, None)
trim_packet = {SEQS_PASSED: [[seq]], ORPHAN_SEQS: []}
trim_packet2 = trim(trim_quality(trim_packet))
seq2 = trim_packet2[SEQS_PASSED][0][0]
assert get_int_qualities(seq2) == [20, 60, 60, 60, 60, 60, 20]
quals = [
40, 18, 10, 40, 40, 5, 8, 30, 14, 3, 40, 40, 40, 11, 6, 5, 3, 20,
10, 12, 8, 5, 4, 7, 1
]
seq = SeqRecord(Seq('atatatatagatagatagatagatg'))
seq.letter_annotations['phred_quality'] = quals
seq = SeqWrapper(SEQRECORD, seq, None)
trim_packet = {SEQS_PASSED: [[seq]], ORPHAN_SEQS: []}
trim_quality = TrimByQuality(window=5, threshold=25)
trim_packet2 = trim(trim_quality(trim_packet))
seq2 = trim_packet2[SEQS_PASSED][0][0]
assert get_int_qualities(seq2) == [40, 18, 10, 40, 40]
quals = [
40, 40, 13, 11, 40, 9, 40, 4, 27, 38, 40, 4, 11, 40, 40, 10, 10,
21, 3, 40, 9, 9, 12, 10, 9
]
seq = SeqRecord(Seq('atatatatatatatatatatatata'))
seq.letter_annotations['phred_quality'] = quals
seq = SeqWrapper(SEQRECORD, seq, None)
trim_quality = TrimByQuality(window=5, threshold=25)
trim_packet = {SEQS_PASSED: [[seq]], ORPHAN_SEQS: []}
trim_packet2 = trim(trim_quality(trim_packet))
seq2 = trim_packet2[SEQS_PASSED][0][0]
expected = [40, 4, 27, 38, 40]
assert get_int_qualities(seq2) == expected
quals = [
40, 40, 13, 11, 40, 9, 40, 4, 27, 38, 40, 4, 11, 40, 40, 10, 10,
21, 3, 40, 9, 9, 12, 10, 9
]
seq = SeqRecord(Seq('atatatatatatatatatatatata'))
seq.letter_annotations['phred_quality'] = quals
seq = SeqWrapper(SEQRECORD, seq, None)
trim_packet = {SEQS_PASSED: [[seq]], ORPHAN_SEQS: []}
trim_quality = TrimByQuality(window=5, threshold=25, trim_left=False)
trim_packet2 = trim(trim_quality(trim_packet))
seq2 = trim_packet2[SEQS_PASSED][0][0]
assert get_int_qualities(seq2) == [
40, 40, 13, 11, 40, 9, 40, 4, 27, 38, 40
]
quals = [
40, 40, 13, 11, 40, 9, 40, 4, 27, 38, 40, 4, 11, 40, 40, 10, 10,
21, 3, 40, 9, 9, 12, 10, 9
]
seq = SeqRecord(Seq('atatatatatatatatatatatata'))
seq.letter_annotations['phred_quality'] = quals
seq = SeqWrapper(SEQRECORD, seq, None)
trim_packet = {SEQS_PASSED: [[seq]], ORPHAN_SEQS: []}
trim_quality = TrimByQuality(window=5, threshold=25, trim_right=False)
trim_packet2 = trim(trim_quality(trim_packet))
seq2 = trim_packet2[SEQS_PASSED][0][0]
assert get_int_qualities(seq2) == [
40, 4, 27, 38, 40, 4, 11, 40, 40, 10, 10, 21, 3, 40, 9, 9, 12, 10,
9
]
quals = [
40, 40, 13, 11, 40, 9, 40, 4, 27, 38, 40, 4, 11, 40, 40, 10, 10,
21, 3, 40, 9, 9, 12, 10, 9
]
seq = SeqRecord(Seq('atatatatatatatatatatatata'))
seq.letter_annotations['phred_quality'] = quals
seq = SeqWrapper(SEQRECORD, seq, None)
trim_quality = TrimByQuality(window=5,
threshold=25,
trim_right=False,
trim_left=False)
trim_packet = {SEQS_PASSED: [[seq]], ORPHAN_SEQS: []}
trim_packet2 = trim(trim_quality(trim_packet))
seq2 = trim_packet2[SEQS_PASSED][0][0]
assert get_int_qualities(seq2) == quals
# With SeqItems
seq = SeqItem('s', [
'@s\n', 'atatatatatatatatatatatata\n', '\n',
'II.,I*I%<GI%,II++6$I**-+*\n'
])
seq = SeqWrapper(SEQITEM, seq, 'fastq')
trim_quality = TrimByQuality(window=5,
threshold=25,
trim_right=True,
trim_left=False)
trim_packet = {SEQS_PASSED: [[seq]], ORPHAN_SEQS: []}
trim_packet2 = trim(trim_quality(trim_packet))
seq2 = trim_packet2[SEQS_PASSED][0][0]
assert seq2.object.lines[3] == 'II.,I*I%<GI\n'
def test_edge_trimming(self):
'It trims the edges'
trim = TrimOrMask()
trim_edges = TrimEdges(left=1)
trim_packet = trim(trim_edges(self._some_seqs()))
res = [get_str_seq(s) for l in trim_packet[SEQS_PASSED] for s in l]
assert res == ['CCG', 'AACCCGGG']
trim_edges = TrimEdges(right=1)
trim_packet = trim(trim_edges(self._some_seqs()))
res = [get_str_seq(s) for l in trim_packet[SEQS_PASSED] for s in l]
assert res == ['ACC', 'AAACCCGG']
trim_edges = TrimEdges(left=1, right=1)
trim_packet = trim(trim_edges(self._some_seqs()))
res = [get_str_seq(s) for l in trim_packet[SEQS_PASSED] for s in l]
assert res == ['CC', 'AACCCGG']
trim_edges = TrimEdges(left=2, right=2)
trim_packet = trim(trim_edges(self._some_seqs()))
res = [get_str_seq(s) for l in trim_packet[SEQS_PASSED] for s in l]
assert res == ['ACCCG']
trim_edges = TrimEdges(left=3, right=3)
trim_packet = trim(trim_edges(self._some_seqs()))
res = [get_str_seq(s) for l in trim_packet[SEQS_PASSED] for s in l]
assert res == ['CCC']
trim = TrimOrMask(mask=True)
trim_edges = TrimEdges(left=1)
trim_packet = trim(trim_edges(self._some_seqs()))
res = [get_str_seq(s) for l in trim_packet[SEQS_PASSED] for s in l]
assert res == ['aCCG', 'aAACCCGGG']
trim_edges = TrimEdges(right=1)
trim_packet = trim(trim_edges(self._some_seqs()))
res = [get_str_seq(s) for l in trim_packet[SEQS_PASSED] for s in l]
assert res == ['ACCg', 'AAACCCGGg']
trim_edges = TrimEdges(left=1, right=1)
trim_packet = trim(trim_edges(self._some_seqs()))
res = [get_str_seq(s) for l in trim_packet[SEQS_PASSED] for s in l]
assert res == ['aCCg', 'aAACCCGGg']
trim_edges = TrimEdges(left=2, right=2)
trim_packet = trim(trim_edges(self._some_seqs()))
res = [get_str_seq(s) for l in trim_packet[SEQS_PASSED] for s in l]
assert res == ['accg', 'aaACCCGgg']
trim_edges = TrimEdges(left=3, right=3)
trim_packet = trim(trim_edges(self._some_seqs()))
res = [get_str_seq(s) for l in trim_packet[SEQS_PASSED] for s in l]
assert res == ['accg', 'aaaCCCggg']
# test overlapping mask
trim1 = TrimEdges(left=3, right=3)
trim2 = TrimEdges(left=4, right=4)
trim_packet = trim(trim2(trim1(self._some_seqs())))
res = [get_str_seq(s) for l in trim_packet[SEQS_PASSED] for s in l]
assert res == ['accg', 'aaacCcggg']
# With a SeqItem
trim = TrimOrMask(mask=False)
trim_edges = TrimEdges(left=1, right=1)
seq = SeqItem('s', ['>s\n', 'ACTTTC\n'])
seqs = [[SeqWrapper(SEQITEM, seq, 'fasta')]]
trim_packet = {SEQS_PASSED: seqs, ORPHAN_SEQS: []}
trim_packet = trim(trim_edges(trim_packet))
res = [get_str_seq(s) for l in trim_packet[SEQS_PASSED] for s in l]
assert res == ['CTTT']
trim = TrimOrMask(mask=True)
seq = SeqItem('s', ['>s\n', 'ACTTTC\n'])
seqs = [[SeqWrapper(SEQITEM, seq, 'fasta')]]
trim_packet = {SEQS_PASSED: seqs, ORPHAN_SEQS: []}
trim_packet = trim(trim_edges(trim_packet))
res = [get_str_seq(s) for l in trim_packet[SEQS_PASSED] for s in l]
assert res == ['aCTTTc']