def get_NCBItaxid(accession, email=None, api_key=None, engine=None):
"""
Access NCBI to retrieve taxid information from the accession number
:param engine: connection to local database
:param api_key: API key for 10 requests per second
:param accession: NCBI accession number
:return: taxid
esearch -db nuccore -query LC315594.1|elink -target taxonomy |efetch
-format uid
"""
print('Get taxid with engine', engine)
if engine is not None:
query = "SELECT taxid FROM taxas WHERE accession='%s'" % accession
tax_id = engine.execute(query).first()[0]
else:
if not email:
raise BaseException('Please provide a valid email address')
try:
query_service = eutils.QueryService(email=email, api_key=api_key)
xml = query_service.elink({
'dbfrom': 'nuccore',
'db': 'taxonomy',
'id': accession
}).decode('utf')
tax_id = xml[xml.rfind('<Id>') + 4:xml.rfind('</Id>')]
except eutils.exceptions.EutilsRequestError:
print(accession)
raise
return tax_id
def get_eutils_client(cache_path, cache=None):
"""
:param cache_path: valid filesystem path to SQLite cache file
:param api_key: (optional) NCBI API Key obtainable from https://www.ncbi.nlm.nih.gov
:return: eutils Client object
"""
import eutils
return eutils.QueryService(cache_path, api_key=API_KEY, cache=cache)
"/', 'void.ttl', '" + inchikey_uri + "');\n")
# Elink
if todo_Elink:
run_as_test = config['ELINK'].getboolean('run_as_test')
apiKey = config['ELINK'].get('api_key')
pmid_cid_version = args.version
pack_size = config['ELINK'].getint('pack_size')
timeout = config['ELINK'].getint('timeout')
max_triples_by_files = config['ELINK'].getint('max_triples_by_files')
all_pmids = None
# Building requests
query_builder = eutils.QueryService(cache=False,
default_args={
'retmax': 10000000,
'retmode': 'xml',
'usehistory': 'n'
},
api_key=apiKey,
email="*****@*****.**")
# Build Elink ressource creator:
pmid_cid = Elink_ressource_creator(
ressource_name="PMID_CID",
version=pmid_cid_version,
dbfrom="pubmed",
db="pccompound",
ns_linking_id=("reference", "PMID"),
ns_linked_id=("compound", "CID"),
ns_endpoint=("endpoint", ""),
primary_predicate=("cito", "discusses"),
secondary_predicate=("cito", "isCitedAsDataSourceBy"),
namespaces=namespaces,
def qs():
return eutils.QueryService()