def main():
parser = argparse.ArgumentParser(description="Extract tRNA and rRNA features from a sequence specified by NCBI GenBank ID")
parser.add_argument("ncbi_id", help="NCBI GenBank ID to extract RNA features from.")
parser.add_argument("--reuse", action="store_true", default=False,
help="Reuse exising file if available")
args = parser.parse_args()
# Download file in GenBank format
# Could download into a StringIO buffer by using the
# ncbi_acc_download.download functions directly, but then we can't --reuse
filename = "{}.gbk".format(args.ncbi_id)
if not os.path.isfile(filename):
cfg = Config(recursive=True, format="genbank", out=filename)
download_to_file(args.ncbi_id, cfg, filename)
# Extract RNA features
records = list(SeqIO.parse(filename, "genbank"))
for record in records:
for feature in record.features:
if feature.type not in ("tRNA", "rRNA"):
continue
# this crashes if there is no locus tag, but NCBI genomes should have these
locus_tag = feature.qualifiers['locus_tag'][0]
header = ">{type}|{locus_tag}".format(type=feature.type, locus_tag=locus_tag)
seq = feature.extract(record.seq)
print(header, seq, sep='\n')
def test_write_stream(mocker):
req = mocker.Mock()
req.iter_content = mocker.Mock(side_effect=requests.exceptions.ChunkedEncodingError)
handle = StringIO()
cfg = Config()
with pytest.raises(DownloadError):
download.write_stream(req, handle, "FAKE", cfg)
def test_download_wgs_parts_wgs_scafld(req):
cfg = Config(format="genbank")
wgs_contig = open(full_path('wgs_scafld.gbk'), 'rt')
with open(full_path('wgs_full.gbk'), 'rt') as handle:
full_file = handle.read()
req.get(ENTREZ_URL, text=full_file)
outhandle = wgs.download_wgs_parts(wgs_contig, cfg)
assert outhandle.getvalue() == full_file
wgs_contig.close()
def test_download_wgs_parts_wgs(req):
cfg = Config(format="genbank")
wgs_contig = open(full_path('wgs.gbk'), 'rt')
req.get(ENTREZ_URL, body=open(full_path('wgs_full.gbk'), 'rt'))
outhandle = wgs.download_wgs_parts(wgs_contig, cfg)
wgs_full = open(full_path('wgs_full.gbk'), 'rt')
assert outhandle.getvalue() == wgs_full.read()
wgs_full.close()
wgs_contig.close()
def test_download_wgs_no_parts(req):
cfg = Config(format="genbank")
supercontig = open(full_path('supercontig_full.gbk'), 'rt')
req.get(ENTREZ_URL, status_code=404)
outhandle = wgs.download_wgs_parts(supercontig, cfg)
supercontig_full = open(full_path('supercontig_full.gbk'), 'rt')
assert outhandle.getvalue() == supercontig_full.read()
supercontig_full.close()
supercontig.close()
def test_download_wgs_parts_no_biopython():
old_have_biopython = wgs.HAVE_BIOPYTHON
wgs.HAVE_BIOPYTHON = False
cfg = Config(format="genbank")
handle = StringIO()
new_handle = wgs.download_wgs_parts(handle, cfg)
wgs.HAVE_BIOPYTHON = old_have_biopython
assert handle == new_handle
def test_download_wgs_parts_wgs_retry(req):
cfg = Config(format="genbank")
wgs_contig = open(full_path('wgs.gbk'), 'rt')
req.get(ENTREZ_URL, response_list=[
{"text": u'Whoa, slow down', "status_code": 429, "headers": {"Retry-After": "0"}},
{"body": open(full_path('wgs_full.gbk'), 'rt')},
])
outhandle = wgs.download_wgs_parts(wgs_contig, cfg)
wgs_full = open(full_path('wgs_full.gbk'), 'rt')
assert outhandle.getvalue() == wgs_full.read()
wgs_full.close()
wgs_contig.close()