def build_translation_table(filename):
"""Given a file object of a CIF file this will produce a translation table
usable by translate
"""
translation_table = {}
data = []
with open(filename, 'r') as raw:
parser = PdbxReader(raw)
parser.read(data)
pdb = data[0]
pdb_id = pdb.getName()
# If there is no assembly gen then the whole AU is in one file. Here there
# are no biological assemblies but I don't know how many models there are,
# so we return a defaultdict which will always return '1_555'.
if not pdb.getObj('pdbx_struct_assembly_gen'):
return {pdb_id: defaultdict(lambda: defaultdict(lambda: '1_555'))}
operator_table = table(pdb, 'pdbx_struct_oper_list')
if not operator_table:
raise MissingOperatorTableError(filename)
translation_table[pdb_id] = {}
for gen_row in rows(pdb, 'pdbx_struct_assembly_gen'):
assembly_id = gen_row['assembly_id']
if not translation_table[pdb_id].get(assembly_id):
translation_table[pdb_id][assembly_id] = {}
# Here I am assumming that the AU is always 1_555.
if '(' in gen_row['oper_expression']:
model_builder = DefaultUsingKey(lambda k: 'P_%s' % k)
return {pdb_id: defaultdict(lambda: model_builder)}
models = gen_row['oper_expression'].split(',')
for model in models:
name = operator_table[model]['name']
translation_table[pdb_id][assembly_id][model] = name
return translation_table
