def test_write_sequence_object(self):
with FastaWriter(self.path) as fw:
fw.write(SequenceRecord("name", "CCATA"))
fw.write(SequenceRecord("name2", "HELLO"))
assert fw._file.closed
with open(self.path) as t:
assert t.read() == '>name\nCCATA\n>name2\nHELLO\n'
def test_reverse_complement(self):
assert SequenceRecord("name1",
"ACGTUMRWSYKVHDBNacgtumrwsykvhdbn",
"/AAAA/6E/EEEEEEEEEEEE/EEEEA///E/"
).reverse_complement() == \
SequenceRecord("name1",
"nvhdbmrswykaacgtNVHDBMRSWYKAACGT",
"/E///AEEEE/EEEEEEEEEEEE/E6/AAAA/")
def test_reverse_complement_none_qualities(self):
assert SequenceRecord("name1",
"GATTACA",
None
).reverse_complement() == \
SequenceRecord("name1",
"TGTAATC",
None)
def test_twoheaders(self):
with FastqWriter(self.path, two_headers=True) as fq:
fq.write(SequenceRecord("name", "CCATA", "!#!#!"))
fq.write(SequenceRecord("name2", "HELLO", "&&&!&"))
assert fq._file.closed
with open(self.path) as t:
assert t.read(
) == '@name\nCCATA\n+name\n!#!#!\n@name2\nHELLO\n+name2\n&&&!&\n'
def test_read(self):
s1 = BytesIO(b'@r1\nACG\n+\nHHH\n')
s2 = BytesIO(b'@r2\nGTT\n+\n858\n')
with TwoFilePairedEndReader(s1, s2) as psr:
assert [
(SequenceRecord("r1", "ACG",
"HHH"), SequenceRecord("r2", "GTT", "858")),
] == list(psr)
def test_write_to_file_like_object(self):
bio = BytesIO()
with FastaWriter(bio) as fw:
fw.write(SequenceRecord("name", "CCATA"))
fw.write(SequenceRecord("name2", "HELLO"))
assert bio.getvalue() == b'>name\nCCATA\n>name2\nHELLO\n'
assert not bio.closed
assert not fw._file.closed
def test(self):
reads = [
(SequenceRecord('A/1 comment', 'TTA',
'##H'), SequenceRecord('A/2 comment', 'GCT',
'HH#')),
(SequenceRecord('B/1', 'CC',
'HH'), SequenceRecord('B/2', 'TG', '#H')),
]
bio = BytesIO()
with InterleavedPairedEndWriter(bio) as writer:
for read1, read2 in reads:
writer.write(read1, read2)
assert bio.getvalue() == (b'@A/1 comment\nTTA\n+\n##H\n'
b'@A/2 comment\nGCT\n+\nHH#\n'
b'@B/1\nCC\n+\nHH\n'
b'@B/2\nTG\n+\n#H\n')
def test(self):
expected = [
(SequenceRecord('read1/1 some text', 'TTATTTGTCTCCAGC',
'##HHHHHHHHHHHHH'),
SequenceRecord('read1/2 other text', 'GCTGGAGACAAATAA',
'HHHHHHHHHHHHHHH')),
(SequenceRecord('read3/1', 'CCAACTTGATATTAATAACA',
'HHHHHHHHHHHHHHHHHHHH'),
SequenceRecord('read3/2', 'TGTTATTAATATCAAGTTGG',
'#HHHHHHHHHHHHHHHHHHH')),
]
with InterleavedPairedEndReader("tests/data/interleaved.fastq") as isr:
reads = list(isr)
assert reads == expected
with dnaio.open("tests/data/interleaved.fastq", interleaved=True) as f:
reads = list(f)
assert reads == expected
def test_set_name_bad(self):
seq = SequenceRecord("name1", "A", "=")
with pytest.raises(ValueError) as error:
seq.name = "näme1"
error.match("ASCII")
def test_write_zero_length_sequence_record(self):
bio = BytesIO()
with FastaWriter(bio) as fw:
fw.write(SequenceRecord("name", ""))
assert bio.getvalue() == b'>name\n\n', '{!r}'.format(bio.getvalue())
def test_is_mate_succes(self):
assert SequenceRecord("name1", "A",
"=").is_mate(SequenceRecord("name2", "GC", "FF"))
def test_missing_final_newline(self):
# Files with a missing final newline are currently allowed
fastq = BytesIO(b'@r1\nA\n+\nH')
with dnaio.open(fastq) as f:
records = list(f)
assert records == [SequenceRecord('r1', 'A', 'H')]
FastqReader,
InterleavedPairedEndReader,
FastaWriter,
FastqWriter,
InterleavedPairedEndWriter,
TwoFilePairedEndReader,
)
from dnaio import record_names_match, SequenceRecord
from dnaio.writers import FileWriter
from dnaio.readers import BinaryFileReader
TEST_DATA = Path(__file__).parent / "data"
SIMPLE_FASTQ = str(TEST_DATA / "simple.fastq")
# files tests/data/simple.fast{q,a}
simple_fastq = [
SequenceRecord("first_sequence", "SEQUENCE1", ":6;;8<=:<"),
SequenceRecord("second_sequence", "SEQUENCE2", "83<??:(61")
]
simple_fasta = [SequenceRecord(x.name, x.sequence, None) for x in simple_fastq]
tiny_fastq = b'@r1\nACG\n+\nHHH\n@r2\nT\n+\n#\n'
class TestFastaReader:
def test_file(self):
with FastaReader("tests/data/simple.fasta") as f:
reads = list(f)
assert reads == simple_fasta
def test_bytesio(self):
def test_set_qualities_none(self):
seq = SequenceRecord("name1", "A", "=")
seq.qualities = None
assert seq.qualities is None
def test_set_qualities_bad(self):
seq = SequenceRecord("name1", "A", "=")
with pytest.raises(ValueError) as error:
seq.qualities = "Ä"
error.match("ASCII")
def test_fastq_bytes_two_headers(self):
assert SequenceRecord("name", "ACGT", "====").fastq_bytes(two_headers=True) == \
b"@name\nACGT\n+name\n====\n"
def test_init_qualities_none(self):
seq = SequenceRecord("name1", "A", None)
assert seq.qualities is None
def test_init_qualities_bad(self):
with pytest.raises(ValueError) as error:
SequenceRecord("name1", "A", "ä")
error.match("ASCII")
def test_too_many_qualities(self):
with pytest.raises(ValueError):
SequenceRecord(name="name", sequence="ACGT", qualities="#####")
def test_init_sequence_none(self):
with pytest.raises(TypeError) as error:
SequenceRecord("name1", None, "=")
error.match("str")
def test_init_sequence_bad(self):
with pytest.raises(ValueError) as error:
SequenceRecord("name1", "Ä", "=")
error.match("ASCII")
def test_set_sequence_none(self):
seq = SequenceRecord("name1", "A", "=")
with pytest.raises(TypeError) as error:
seq.sequence = None
error.match("str")
def test_fastq_bytes(self):
assert SequenceRecord("name", "ACGT", "====").fastq_bytes() == \
b"@name\nACGT\n+\n====\n"