def fetch(params, seqid=None, db=None, update=False):
"""
Obtains data from NCBI
"""
for p in params:
# Skip if exists (or not update).
if p.json and not update:
continue
# The JSON representation of the data.
json_name = resolve_fname(name=p.acc, format="json")
# GenBank representation of the data.
gbk_name = resolve_fname(name=p.acc, format="gb")
# Fetch and store genbank from remote site.
ncbi_efetch(p.acc, db=db, gbk_name=gbk_name)
# Convert genbank to JSON.
data = jsonrec.parse_file(fname=gbk_name, seqid=seqid)
# Save JSON file.
save_json_file(fname=json_name, data=data)
def fetch_data(data, param):
"""
Obtains data from NCBI. Fills each parameter with a json field.
"""
db = "protein" if param.protein else "nuccore"
for name in data:
# Pretend no data if it is an update.
json = None if param.update else get_json(name)
# The data exists, nothing needs to be done.
if json:
continue
# The JSON representation of the data.
json_name = resolve_fname(name=name, format="json")
# GenBank representation of the data.
gbk_name = resolve_fname(name=name, format="gb")
# Genome assembly data.
if name.startswith("GCA") or name.startswith("GCF"):
ncbi.genome(name=name, fname=gbk_name, update=param.update)
else:
# Genbank data.
ncbi.genbank_save(name, db=db, fname=gbk_name)
# Convert Genbank to JSON.
data = jsonrec.parse_file(fname=gbk_name, seqid=param.seqid)
# Save JSON file.
save_json_file(fname=json_name, data=data)
def get_json(name, seqid=None, update=False, inter=False, strict=False):
"""
Attempts to return a JSON formatted data based on a name.
"""
# Data is an existing path to a file.
if os.path.isfile(name):
data = jsonrec.parse_file(name, seqid=seqid)
return data
# Not a local file, attempt to resolve to storage.
# Report as not found if update is requested.
if update:
return None
# The JSON representation of the data.
json_name = resolve_fname(name=name, format="json")
# GenBank representation of the data.
gbk_name = resolve_fname(name=name, format="gb")
# Found the JSON representation of the file.
if os.path.isfile(json_name):
logger.info(f"found {json_name}")
data = read_json_file(json_name)
return data
# No JSON file but there is a genbank file.
if os.path.isfile(gbk_name):
logger.info(f"found {gbk_name}")
data = jsonrec.parse_file(fname=gbk_name, seqid=seqid)
data = save_json_file(fname=json_name, data=data)
return data
# If not found and interactive mode create a JSON from the name itself.
if inter:
data = jsonrec.make_jsonrec(seq=name, seqid=seqid)
return data
# At this point the data was not found
if strict:
utils.error(f"data not found: {name}")
return None
def get_json(name, seqid=None, inter=False, strict=False):
"""
Attempts to return a JSON formatted data based on a name.
"""
# Data is an existing path to a JSON file.
if os.path.isfile(name):
try:
data = jsonrec.parse_file(name, seqid=seqid)
except Exception as exc:
logger.error(f"JSON parsing error for file {name}: {exc}")
sys.exit(-1)
return data
# The JSON representation of the data.
json_name = resolve_fname(name=name, format="json")
# GenBank representation of the data.
gbk_name = resolve_fname(name=name, format="gb")
# Found the JSON representation of the file.
if os.path.isfile(json_name):
logger.info(f"found {json_name}")
data = read_json_file(json_name)
return data
# There is no JSON file but there is a GenBank file.
if os.path.isfile(gbk_name):
logger.info(f"found {gbk_name}")
data = jsonrec.parse_file(fname=gbk_name, seqid=seqid)
data = save_json_file(fname=json_name, data=data)
return data
# Interactive input, make JSON from name
if inter:
data = jsonrec.make_jsonrec(name, seqid=seqid)
return data
# Raise error if in strict mode
if strict:
utils.error(f"data not found: {name}")
return None