def test_custom_phred_offset(self):
ascii_chars = "*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\"
obs = _encode_phred_to_qual(list(range(51)), phred_offset=42)
self.assertEqual(obs, ascii_chars)
with self.assertRaises(ValueError) as cm:
_encode_phred_to_qual([42, -1, 33], phred_offset=42)
self.assertIn("-1", str(cm.exception))
self.assertIn("[0, 84]", str(cm.exception))
obs = npt.assert_warns(UserWarning, _encode_phred_to_qual, [42, 255, 33], phred_offset=42)
self.assertEqual(obs, "T~K")
def test_illumina13_variant(self):
# test entire range of possible ascii chars for illumina1.3
all_illumina13_ascii = "@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_`abcdefghijk" "lmnopqrstuvwxyz{|}~"
obs = _encode_phred_to_qual(list(range(63)), variant="illumina1.3")
self.assertEqual(obs, all_illumina13_ascii)
with self.assertRaises(ValueError) as cm:
_encode_phred_to_qual([42, -1, 33], variant="illumina1.3")
self.assertIn("-1", str(cm.exception))
self.assertIn("[0, 62]", str(cm.exception))
obs = npt.assert_warns(UserWarning, _encode_phred_to_qual, [42, 63, 33], variant="illumina1.3")
self.assertEqual(obs, "j~a")
def test_illumina18_variant(self):
# test entire range of possible ascii chars for illumina1.8
all_illumina18_ascii = "!\"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKL" "MNOPQRSTUVWXYZ[\\]^_"
obs = _encode_phred_to_qual(list(range(63)), variant="illumina1.8")
self.assertEqual(obs, all_illumina18_ascii)
with self.assertRaises(ValueError) as cm:
_encode_phred_to_qual([42, -1, 33], variant="illumina1.8")
self.assertIn("-1", str(cm.exception))
self.assertIn("[0, 62]", str(cm.exception))
obs = npt.assert_warns(UserWarning, _encode_phred_to_qual, [42, 63, 33], variant="illumina1.8")
self.assertEqual(obs, "K_B")
def test_custom_phred_offset(self):
ascii_chars = '*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\'
obs = _encode_phred_to_qual(list(range(51)), phred_offset=42)
self.assertEqual(obs, ascii_chars)
with self.assertRaises(ValueError) as cm:
_encode_phred_to_qual([42, -1, 33], phred_offset=42)
self.assertIn('-1', str(cm.exception))
self.assertIn('[0, 84]', str(cm.exception))
obs = npt.assert_warns(UserWarning,
_encode_phred_to_qual, [42, 255, 33],
phred_offset=42)
self.assertEqual(obs, 'T~K')
def test_sanger_variant(self):
# test entire range of possible ascii chars for sanger
all_sanger_ascii = (
"!\"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOP" "QRSTUVWXYZ[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~"
)
obs = _encode_phred_to_qual(list(range(94)), variant="sanger")
self.assertEqual(obs, all_sanger_ascii)
with self.assertRaises(ValueError) as cm:
_encode_phred_to_qual([42, -1, 33], variant="sanger")
self.assertIn("-1", str(cm.exception))
self.assertIn("[0, 93]", str(cm.exception))
obs = npt.assert_warns(UserWarning, _encode_phred_to_qual, [42, 94, 33], variant="sanger")
self.assertEqual(obs, "K~B")
def test_illumina13_variant(self):
# test entire range of possible ascii chars for illumina1.3
all_illumina13_ascii = ('@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_`abcdefghijk'
'lmnopqrstuvwxyz{|}~')
obs = _encode_phred_to_qual(list(range(63)), variant='illumina1.3')
self.assertEqual(obs, all_illumina13_ascii)
with self.assertRaises(ValueError) as cm:
_encode_phred_to_qual([42, -1, 33], variant='illumina1.3')
self.assertIn('-1', str(cm.exception))
self.assertIn('[0, 62]', str(cm.exception))
obs = npt.assert_warns(UserWarning,
_encode_phred_to_qual, [42, 63, 33],
variant='illumina1.3')
self.assertEqual(obs, 'j~a')
def test_sanger_variant(self):
# test entire range of possible ascii chars for sanger
all_sanger_ascii = ('!"#$%&\'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOP'
'QRSTUVWXYZ[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~')
obs = _encode_phred_to_qual(list(range(94)), variant='sanger')
self.assertEqual(obs, all_sanger_ascii)
with self.assertRaises(ValueError) as cm:
_encode_phred_to_qual([42, -1, 33], variant='sanger')
self.assertIn('-1', str(cm.exception))
self.assertIn('[0, 93]', str(cm.exception))
obs = npt.assert_warns(UserWarning,
_encode_phred_to_qual, [42, 94, 33],
variant='sanger')
self.assertEqual(obs, 'K~B')
def test_illumina18_variant(self):
# test entire range of possible ascii chars for illumina1.8
all_illumina18_ascii = ('!"#$%&\'()*+,-./0123456789:;<=>?@ABCDEFGHIJKL'
'MNOPQRSTUVWXYZ[\\]^_')
obs = _encode_phred_to_qual(list(range(63)), variant='illumina1.8')
self.assertEqual(obs, all_illumina18_ascii)
with self.assertRaises(ValueError) as cm:
_encode_phred_to_qual([42, -1, 33], variant='illumina1.8')
self.assertIn('-1', str(cm.exception))
self.assertIn('[0, 62]', str(cm.exception))
obs = npt.assert_warns(UserWarning,
_encode_phred_to_qual, [42, 63, 33],
variant='illumina1.8')
self.assertEqual(obs, 'K_B')
def _generator_to_fastq(obj, fh, variant=None, phred_offset=None,
id_whitespace_replacement='_',
description_newline_replacement=' ', lowercase=None):
formatted_records = _format_fasta_like_records(
obj, id_whitespace_replacement, description_newline_replacement, True,
lowercase=lowercase)
for header, seq_str, qual_scores in formatted_records:
qual_str = _encode_phred_to_qual(qual_scores, variant=variant,
phred_offset=phred_offset)
fh.write('@')
fh.write(header)
fh.write('\n')
fh.write(seq_str)
fh.write('\n+\n')
fh.write(qual_str)
fh.write('\n')
def _generator_to_fastq(obj, fh, variant=None, phred_offset=None,
id_whitespace_replacement='_',
description_newline_replacement=' ', lowercase=None):
formatted_records = _format_fasta_like_records(
obj, id_whitespace_replacement, description_newline_replacement, True,
lowercase=lowercase)
for header, seq_str, qual_scores in formatted_records:
qual_str = _encode_phred_to_qual(qual_scores, variant=variant,
phred_offset=phred_offset)
fh.write('@')
fh.write(header)
fh.write('\n')
fh.write(seq_str)
fh.write('\n+\n')
fh.write(qual_str)
fh.write('\n')
def test_no_phred_scores(self):
self.assertEqual(_encode_phred_to_qual([], variant='sanger'), '')
def test_unrecognized_variant(self):
with self.assertRaises(ValueError) as cm:
_encode_phred_to_qual([1, 2, 3], variant='illumina')
self.assertIn('variant', str(cm.exception))
self.assertIn("'illumina'", str(cm.exception))
def test_solexa_variant(self):
with self.assertRaises(ValueError) as cm:
_encode_phred_to_qual([1, 2, 3], variant='solexa')
self.assertIn('719', str(cm.exception))
def test_variant_and_phred_offset_provided(self):
with self.assertRaises(ValueError) as cm:
_encode_phred_to_qual([1, 2, 3], variant='sanger', phred_offset=64)
self.assertIn('both', str(cm.exception))
self.assertIn('`variant`', str(cm.exception))
self.assertIn('`phred_offset`', str(cm.exception))
def test_solexa_variant(self):
with self.assertRaises(ValueError) as cm:
_encode_phred_to_qual([1, 2, 3], variant="solexa")
self.assertIn("719", str(cm.exception))
def test_missing_variant_and_phred_offset(self):
with self.assertRaises(ValueError) as cm:
_encode_phred_to_qual([1, 2, 3])
self.assertIn("`variant`", str(cm.exception))
self.assertIn("`phred_offset`", str(cm.exception))
self.assertIn("encode", str(cm.exception))
def test_no_phred_scores(self):
self.assertEqual(_encode_phred_to_qual([], variant="sanger"), "")
def test_unrecognized_variant(self):
with self.assertRaises(ValueError) as cm:
_encode_phred_to_qual([1, 2, 3], variant="illumina")
self.assertIn("variant", str(cm.exception))
self.assertIn("'illumina'", str(cm.exception))
def test_missing_variant_and_phred_offset(self):
with self.assertRaises(ValueError) as cm:
_encode_phred_to_qual([1, 2, 3])
self.assertIn('`variant`', str(cm.exception))
self.assertIn('`phred_offset`', str(cm.exception))
self.assertIn('encode', str(cm.exception))
def test_solexa_variant(self):
with self.assertRaises(NotImplementedError) as cm:
_encode_phred_to_qual([1, 2, 3], variant='solexa')
self.assertIn('719', str(cm.exception))
def test_variant_and_phred_offset_provided(self):
with self.assertRaises(ValueError) as cm:
_encode_phred_to_qual([1, 2, 3], variant="sanger", phred_offset=64)
self.assertIn("both", str(cm.exception))
self.assertIn("`variant`", str(cm.exception))
self.assertIn("`phred_offset`", str(cm.exception))
