def __init__(self,
database: Database,
host='ftp.ncbi.nlm.nih.gov',
data_dir='gene/DATA/',
src_data_dir=APP_ROOT / 'data' / 'ncbi'):
"""Construct the NCBI ETL instance.
:param Database database: gene database for adding new data
:param str host: FTP host name
:param str data_dir: FTP data directory to use
:param Path src_data_dir: Data directory for NCBI
"""
super().__init__(database, host, data_dir, src_data_dir)
self._sequence_location = SequenceLocation()
self._chromosome_location = ChromosomeLocation()
self._data_url = f"ftp://{host}"
self._assembly = None
self._date_today = datetime.today().strftime('%Y%m%d')
def __init__(self,
database: Database,
host='ftp.ebi.ac.uk',
data_dir='pub/databases/genenames/hgnc/json/',
src_data_dir=APP_ROOT / 'data' / 'hgnc',
fn='hgnc_complete_set.json'):
"""Initialize HGNC ETL class.
:param Database database: DynamoDB database
:param str host: FTP host name
:param str data_dir: FTP data directory to use
:param Path src_data_dir: Data directory for HGNC
:param str fn: Data file to download
"""
super().__init__(database, host, data_dir, src_data_dir)
self._chromosome_location = ChromosomeLocation()
self._data_url = f"ftp://{host}/{data_dir}{fn}"
self._fn = fn
self._version = None