def test_parse_vdj_paths_with_ref(self):
"""parse_vdj_paths should work with builtin filename fragments"""
shared = {
"fasta": True,
"input": str(self.path / "input"),
"type": "dir"
}
attrs_list_exp = [{
"path": self.path / "input/D.fasta",
"segment": "D"
}, {
"path": self.path / "input/J.fasta",
"segment": "J"
}, {
"path": self.path / "input/V.fasta",
"segment": "V"
}]
for attrs in attrs_list_exp:
attrs.update(shared)
for locus_segment in sorted(
["IGHV", "IGHD", "IGHJ", "IGKV", "IGKJ", "IGLV", "IGLJ"]):
attrs_list_exp.append({
'fasta': True,
'type': 'internal',
'input': 'rhesus/imgt',
'locus': locus_segment[:3],
'path': DATA /
f"germ/rhesus/imgt/{locus_segment[:3]}/{locus_segment}.fasta",
'ref': 'imgt',
'segment': locus_segment[3],
'species': 'rhesus'
})
attrs_list = vdj.parse_vdj_paths([self.path / "input", "rhesus/imgt"])
self.assertEqual(attrs_list, attrs_list_exp)
def test_parse_vdj_paths_duplicates(self):
"""parse_vdj_paths shouldn't repeat paths that come from multiple input names"""
# Instead it'll just squish both input names into the same entries.
shared = {
"fasta": True,
"input": str(self.path / "input"),
"type": "dir"
}
attrs_list_exp = [{
"path": self.path / "input/D.fasta",
"segment": "D"
}, {
"path": self.path / "input/J.fasta",
"segment": "J"
}, {
"path": self.path / "input/V.fasta",
"segment": "V"
}]
for attrs in attrs_list_exp:
attrs.update(shared)
ref_paths = [self.path / "input", self.path / "input/D.fasta"]
attrs_list_exp[0]["type"] = "file"
attrs_list_exp[0]["input"] = "; ".join([str(p) for p in ref_paths])
attrs_list = vdj.parse_vdj_paths(ref_paths)
self.assertEqual(attrs_list, attrs_list_exp)
def test_parse_vdj_paths(self):
"""parse_vdj_paths should give a list of dicts with info parsed from the given paths"""
shared = {
"fasta": True,
"input": str(self.path / "input"),
"type": "dir"
}
attrs_list_exp = [{
"path": self.path / "input/D.fasta",
"segment": "D"
}, {
"path": self.path / "input/J.fasta",
"segment": "J"
}, {
"path": self.path / "input/V.fasta",
"segment": "V"
}]
for attrs in attrs_list_exp:
attrs.update(shared)
# It should be able to take Path objects, strings, or lists
ref_path_inputs = [
self.path / "input",
str(self.path / "input"), [self.path / "input"]
]
for ref_paths in ref_path_inputs:
attrs_list = vdj.parse_vdj_paths(ref_paths)
self.assertEqual(attrs_list, attrs_list_exp)
def test_parse_vdj_paths_with_files(self):
"""parse_vdj_paths should work with filenames as inputs"""
# In this case they're sorted as they're given, since the sorting is
# by-ref and then by-file.
mkdict = lambda s: {
"path": self.path / f"input/{s}.fasta",
"input": str(self.path / f"input/{s}.fasta"),
"segment": s,
"fasta": True,
"type": "file"
}
attrs_list_exp = [mkdict(s) for s in ["V", "D", "J"]]
paths = [
self.path / f"input/{segment}.fasta"
for segment in ["V", "D", "J"]
]
attrs_list = vdj.parse_vdj_paths(paths)
self.assertEqual(attrs_list, attrs_list_exp)
def test_parse_vdj_paths(self):
with self.assertRaises(IgSeqError):
vdj.parse_vdj_paths(self.path / "missing")