pdbx_poly_seq_scheme for beaded residues is incorrectly populated

[brindakv]

In coarse-grained models with coordinates in ihm_sphere_obj_site, when the full sequence of an entity is included in entity_poly and entity_poly_seq, but a segment in not modeled and therefore not included in ihm_sphere_obj_site, the pdbx_poly_seq_scheme is populated incorrectly and shows that the entire sequence is modeled. For example, _pdbx_poly_seq_scheme.auth_seq_num and _pdbx_poly_seq_scheme.auth_seq_num should have ?.

See https://files.rcsb.org/view/8QB4.cif for an example for pdbx_poly_seq_scheme when atom_site has missing residues.

[benmwebb]

I don't think this would be specific to coarse-grained models, since python-ihm tracks the mapping between seq_id and auth_seq_num in the asym, not in the atom or sphere objects. It wouldn't be possible to extract information on non-modeled residues at output time from the atom/sphere objects, since we don't enumerate those until after we've dumped all the entity and scheme tables, and we only enumerate them once by design (the Python interface is a generator, not a list). But what we could do is:

1. If any seq_id is included in the _ihm_residues_not_modeled table for all models, emit ? for _pdbx_poly_seq_scheme.auth_seq_num, .pdb_mon_id and .auth_mon_id. (I would expect .pdb_ins_code to also be ? but from your example it looks like . is used.)
2. On read, have make_mmcif check all read-in atoms/spheres and add any non-modeled ranges to the _ihm_residues_not_modeled table if not already there.

Would that work?

[brindakv]

@benmwebb Yes, that would work.