def download_assembly(dest_dir, accession, output_format, fetch_wgs, extract_wgs, expanded, quiet=False):
if output_format is None:
output_format = utils.EMBL_FORMAT
assembly_dir = os.path.join(dest_dir, accession)
utils.create_dir(assembly_dir)
# download xml
utils.download_record(assembly_dir, accession, utils.XML_FORMAT)
local_xml = utils.get_destination_file(assembly_dir, accession, utils.XML_FORMAT)
# get wgs and sequence report info
wgs_set, sequence_report = parse_assembly_xml(local_xml)
has_sequence_report = False
# download sequence report
if sequence_report is not None:
has_sequence_report = utils.get_ftp_file(sequence_report, assembly_dir)
# parse sequence report and download sequences
wgs_scaffolds = []
wgs_scaffold_cnt = 0
if has_sequence_report:
wgs_scaffolds = download_sequences(sequence_report.split('/')[-1], assembly_dir, output_format, expanded, quiet)
wgs_scaffold_cnt = len(wgs_scaffolds)
if wgs_scaffold_cnt > 0:
if not quiet:
print 'Assembly contains {} WGS scaffolds, will fetch WGS set'.format(wgs_scaffold_cnt)
fetch_wgs = True
else:
fetch_wgs = True
# download wgs set if needed
if wgs_set is not None and fetch_wgs:
if not quiet:
print 'fetching wgs set'
sequenceGet.download_wgs(assembly_dir, wgs_set, output_format)
# extract wgs scaffolds from WGS file
if wgs_scaffold_cnt > 0 and extract_wgs:
extract_wgs_scaffolds(assembly_dir, wgs_scaffolds, wgs_set, output_format, quiet)
def download_assembly(dest_dir,
accession,
output_format,
fetch_wgs,
quiet=False):
if output_format is None:
output_format = utils.EMBL_FORMAT
assembly_dir = os.path.join(dest_dir, accession)
utils.create_dir(assembly_dir)
# download xml
utils.download_record(assembly_dir, accession, utils.XML_FORMAT)
local_xml = utils.get_destination_file(assembly_dir, accession,
utils.XML_FORMAT)
# get wgs and sequence report info
wgs_set, sequence_report = parse_assembly_xml(local_xml)
has_sequence_report = False
# download sequence report
if sequence_report is not None:
has_sequence_report = utils.get_ftp_file(sequence_report, assembly_dir)
# download wgs set if needed
if wgs_set is not None and fetch_wgs:
if not quiet:
print 'fetching wgs set'
sequenceGet.download_wgs(assembly_dir, wgs_set, output_format)
# parse sequence report and download sequences
if has_sequence_report:
download_sequences(
sequence_report.split('/')[-1], assembly_dir, output_format, quiet)
def download_sequence(dest_dir, accession, output_format, expanded):
if output_format is None:
output_format = utils.EMBL_FORMAT
success = utils.download_record(dest_dir, accession, output_format, expanded)
if not success:
print 'Unable to fetch file for {0}, format {1}'.format(accession, output_format)
return success
def download_sequence(dest_dir, accession, format):
if format is None:
format = utils.EMBL_FORMAT
success = utils.download_record(dest_dir, accession, format)
if not success:
print 'Unable to fetch file for ' + accession + ', format ' + format
return success
def download_meta(accession, dest_dir):
utils.download_record(dest_dir, accession, utils.XML_FORMAT)
def download_sequence(dest_dir, accession, output_format, expanded):
success = utils.download_record(dest_dir, accession, output_format,
expanded)
if not success:
print('Unable to fetch file for {0}, format {1}'.format(
accession, output_format))
