def test_genbank_bad_loc_wrap_parsing(self):
with warnings.catch_warnings():
warnings.simplefilter("ignore", BiopythonParserWarning)
with open(path.join("GenBank", "bad_loc_wrap.gb")) as handle:
record = GenBank.read(handle)
self.assertEqual(1, len(record.features))
loc = record.features[0].location
self.assertEqual(loc, "join(3462..3615,3698..3978,4077..4307,4408..4797,4876..5028,5141..5332)")
def test_000_genbank_bad_loc_wrap_warning(self):
with warnings.catch_warnings():
warnings.simplefilter("error", BiopythonParserWarning)
with open(path.join("GenBank", "bad_loc_wrap.gb")) as handle:
# self.assertRaises(BiopythonParserWarning, GenBank.read, handle)
try:
record = GenBank.read(handle)
except BiopythonParserWarning as e:
self.assertEqual(str(e), "Non-standard feature line wrapping (didn't break on comma)?")
else:
self.assertTrue(False, "Expected specified BiopythonParserWarning here.")
def parse(path='./flat_files/'):
path = Path(path)
print "parsing records at {}".format(path.absolute())
records = []
for p in path.listdir():
try:
gbr = GenBank.read(open(p))
records.append(gbr)
except:
print 'error with file', p
print "parsed %s records.." % len(records)
return records
def fetch_gb(id, taxon=None):
r = eutils.fetch(db='nuccore', id=id, rettype='gb')
if r.ok:
gb = GenBank.read(StringIO(r.text))
filename = f"{gb.locus}{'.' + taxon if taxon else ''}.gb"
file_out = os.path.join(config.genbank_dir, filename)
if os.path.exists(file_out):
print(f"[WARN] {file_out} already exists")
else:
with open(file_out, 'w') as fh:
fh.write(r.text)
print(f"[INFO] Fetched file: {file_out}")
else:
print(f"[WARN] Could not download '{id}'")
def parse(path='./flat_files/'):
path = Path(path)
print "parsing records at {}".format(path.absolute())
records = []
for p in path.listdir():
try:
gbr = GenBank.read(open(p))
records.append(gbr)
except:
print 'error with file', p
print "parsed %s records.." % len(records)
return records
def from_genbank(cls, filepath: str):
try:
gb = GenBank.read(file=filepath)
source = GenBank.get_source_data(gb)
return cls.insert(
accession=gb.accession[0],
organism=gb.organism,
date_released=GenBank.format_date(gb.date),
host=source.get('host'),
date_collected=source.get('collection_date'),
country=source.get('country'),
)
except Exception as e:
logging.warning(
f"Error inserting {filepath} to {cls.__name__}: {e}")
def add_file(cls, filepath: str):
"""TODO(seanbeagle): Create scraping tool for genbank data similar to get_source_data()"""
try:
gb = GenBank.read(file=filepath)
accession = gb.accession[0]
logging.debug(f"Adding {accession} to {cls}...")
record = cls.insert(
accession=accession,
version=gb.version,
filepath=filepath, # TODO: Ensure this is absolute filepath
date_downloaded=now(),
downloaded_by=getpass.getuser(),
num_features=len(gb.features),
length=len(gb))
return record
except Exception as e:
logging.debug(f"Could not insert GenBank record: {e}")
def fetch(id: str):
logging.info(f"Fetching GenBank id={id}")
r = eutils.fetch(db='nuccore', id=id, rettype='gb')
if r.ok:
gb = GenBank.read(string=r.text)
accession = gb.accession[0]
filename = f"{accession}{'.' + config.taxon if config.taxon else ''}.gb"
file_out = os.path.join(FileSystem.dir['genbank'], filename)
record = GenBank.query.filter_by(version=gb.version).first()
if record and os.path.exists(record.filepath):
logging.info(
f"{accession} already exists as file='{record.filepath}' and GenBank.id={record.id}"
)
else:
with open(file_out, 'w') as fh:
fh.write(r.text)
GenBank.add_file(file_out)
print(f"[INFO] Fetched file: {file_out}")
return file_out
else:
print(f"[WARN] Could not download '{id}'")
def test_genbank_read(self):
"""GenBank.read(...) simple test."""
with open(path.join("GenBank", "NC_000932.gb")) as handle:
record = GenBank.read(handle)
self.assertEqual(['NC_000932'], record.accession)
def test_genbank_read(self):
with open(path.join("GenBank", "NC_000932.gb")) as handle:
record = GenBank.read(handle)
self.assertEqual(['NC_000932'], record.accession)
