def test_label_parser(self):
"""label parser factory function cope with mixed structure labels"""
# the label parser factory function should correctly handle label lines
# with mixed separators
make = LabelParser(
"%(species)s:%(accession)s",
[[0, "accession", str], [2, "species", str]],
split_with=": ",
)
for label, expect in [
(">abcd:human:misc", "misc:abcd"),
("abcd:human:misc", "misc:abcd"),
(">abcd:Human misc", "misc:abcd"),
(">abcd Human:misc", "misc:abcd"),
(">abcd:Human misc", "misc:abcd"),
]:
self.assertEqual(make(label), expect)
# should raise an assertion error if template doesn't match at least
# one field name
self.assertRaises(
AssertionError,
LabelParser,
"%s:%s",
[[0, "accession", str], [2, "species", str]],
split_with=": ",
)
def test_groups(self):
"""correctly yield grouped sequences from fasta formatted data"""
data = [
">group1:seq1_id:species1",
"ACTG",
">group1:seq2_id:species2",
"ACTG",
">group2:seq3_id:species1",
"ACGT",
">group2:seq4_id:species2",
"ACGT",
]
expected = [
{
"species1": "ACTG",
"species2": "ACTG"
},
{
"species1": "ACGT",
"species2": "ACGT"
},
]
label_to_name = LabelParser(
"%(species)s",
[(0, "Group", str), (1, "seq_id", str), (2, "species", str)],
split_with=":",
)
parser = GroupFastaParser(data, label_to_name, aligned=True)
count = 0
for group in parser:
got = group.to_dict()
want = expected[count]
self.assertEqual(got, want)
self.assertEqual(group.info.Group, f"group{count + 1}")
count += 1
# check we don't return a done group
done_groups = ["group1"]
parser = GroupFastaParser(data,
label_to_name,
done_groups=done_groups,
aligned=True)
for group in parser:
got = group.to_dict()
want = expected[1]
self.assertEqual(got, want)
self.assertEqual(group.info.Group, "group2")