def test_pdb_path(self):
"""Test that our PDB path function works."""
self.assertEqual(
pdb.pdb_path("1ABC", ".pdb.gz"),
Path("ab/1abc.pdb.gz"))
self.assertEqual(
pdb.pdb_path("1AbC", ".cif"),
Path("ab/1abc.cif"))
self.assertEqual(
pdb.pdb_path("1abC", ".cif.gz", "A"),
Path("ab/1abc/1abc_A.cif.gz"))
def run(self, data, config=None, pipeline=None):
"""Run component."""
pdb_id = self.get_vals(data)
url = self.URL.format(PDB=pdb_id.lower(), type=self.struc_type.value)
path = pdb.pdb_path(pdb_id,
".{}.gz".format(self.struc_type.value),
base_dir=self.base_dir)
if (not self.overwrite) and path.exists():
return data
path.parent.mkdir(parents=True, exist_ok=True)
session = data["session"] if "session" in data else requests.Session()
with session.get(url, stream=True) as r:
with path.open("wb") as out_fh:
shutil.copyfileobj(r.raw, out_fh)
return data
def test_open_str(self):
"""Should be able to pass a string, not just a Path object."""
pdb_path = str(pdb.pdb_path("4n6v", ".cif.gz", base_dir=self.mmcif_dir))
with pdb.open_pdb(pdb_path) as pdb_in:
self.assertGreater(len(pdb_in.readlines()), 0)
def test_open_pdb_compressed(self):
"""Open a compressed mmcif file."""
pdb_path = pdb.pdb_path("4n6v", ".cif.gz", base_dir=self.mmcif_dir)
with pdb.open_pdb(pdb_path) as pdb_in:
self.assertGreater(len(pdb_in.readlines()), 0)
def run(self, data, config=None, pipeline=None):
"""Run the component."""
pdb_id = self.get_vals(data)
mmcif_parser = Bio.PDB.FastMMCIFParser()
pdb_parser = Bio.PDB.PDBParser()
source_file = pdb.find_pdb(pdb_id, base_dir=self.mmcif_dir)
if source_file is None:
self.logger.error("Could not find MMCIF file '%s' in '%s'", pdb_id,
self.mmcif_dir)
raise self.MissingSourceError(pdb_id)
results = []
self.logger.debug("Extracting chains from %s (%s)", pdb_id,
source_file)
with pdb.open_pdb(source_file) as pdb_in:
structure = mmcif_parser.get_structure(pdb_id, pdb_in)
self.logger.debug("Found %d chains in %s",
len(list(structure.get_chains())), pdb_id)
for chain in structure[0]:
pdb_file = pdb.pdb_path(pdb_id, ".pdb", chain.id,
self.chain_dir)
pdb_file.parent.mkdir(parents=True, exist_ok=True)
result = data.copy()
result["chain"] = chain.id
result["structure"] = str(pdb_file)
result["name"] = "{}_{}".format(pdb_id.lower(), chain.id)
if not pdb_file.exists() or self.overwrite:
self.logger.debug("Extracting chain %s from PDB %s",
chain.id, pdb_id)
# Store all captured log output in REMARK 999
general_logger = logging.getLogger("phyre_engine")
with phyre_engine.logutils.capture_log(
general_logger) as log_buf:
# Select conformations.
for selector in self.conf_sel:
chain = selector.select(chain)
template = Template.build(pdb_id, result["chain"],
chain)
log_buf.seek(0)
template.remarks[999].extend(log_buf.readlines())
with pdb_file.open("w") as pdb_out:
template.write(pdb_out)
else:
chain = pdb_parser.get_structure(
"", result["structure"])[0]["A"]
template = Template.build(pdb_id, result["chain"], chain)
self.logger.debug(
"Loaded existing chain %s of PDB %s from %s", chain.id,
pdb_id, pdb_file)
result["sequence"] = template.canonical_seq
result["original_residues"] = template.mapping
result["canonical_indices"] = template.canonical_indices
results.append(result)
return results