records = krbioio.read_sequence_file(input_file, 'gb', ret_type='list')
# print('Found', len(records), 'in', input_file)
excluded_qualifiers = ['translation', 'db_xref', 'exception',
'rpt_unit_seq', 'gene_synonym', 'rpt_type', 'satellite', 'transl_table',
'replace', 'rpt_unit_range', 'protein_id', 'codon_recognized',
'EC_number', 'function', 'estimated_length', 'mobile_element_type',
'codon_start', 'transl_except', 'number', 'standard_name', 'allele',
'inference']
feature_dict = dict()
taxa_dict = dict()
for record in records:
txid = krncbi.get_ncbi_tax_id_for_record(record)
for feature in record.features:
if feature.type != 'source':
for qualifier in feature.qualifiers:
if qualifier not in excluded_qualifiers:
key = feature.type + '.' + qualifier
qualifier_label = feature.qualifiers[qualifier][0]
qualifier_label_key = key + '.' + qualifier_label
if key not in feature_dict.keys():
feature_dict[key] = list()
if qualifier_label_key not in taxa_dict.keys():
taxa_dict[qualifier_label_key] = list()
feature_dict[key].append(qualifier_label)
taxa_dict[qualifier_label_key].append(txid)
# print(key)
# print(feature_dict[key])
