Modify structure and update mmcif block without losing existing mmCIF information?

[m-crown]

Hi, I am working with a pair of structures for which I have rotation and translation vectors already. I am able to apply the rotation correctly to the structure, but when i do rotated_structure.update_mmcif_block(rotated_block) I lose some of the preexisting information in the cif file that is necessary for viewing secondary structure of the polymer chains in for example molstar.

Is there a recommend way to update the coordinates which only updates the structure coordinates? Similarly, if i add a chain to the structure, and want to update the mmcif block, I would lose this information.

Thanks in advance for any help with this!

[wojdyr]

A few months ago I wanted to do something similar (for gemmi set), to change only coordinates or B-factors and leave everything else intact, with formatting unchanged. I ended up writing custom code that changes only selected lines in mmCIF or PDB files. But if you consider adding a chain, that's probably not the way to go.

The problem with updating structure is that things may get out of sync. For example, if instead of changing coordinates you change chain names, you also need to update secondary structure description. There are plenty of such dependencies and tracking what was changed and what needs to be updated would be too convoluted, it's much simpler to rewrite everything.

Do you know why molstar can't show the secondary structure? Categories with SS (_struct_conf, _struct_sheet etc) should be read and written back.

[m-crown]

It seems to be due to loss of information in the _chem_comp category. When I update the structure, resulting _chem_comp has only id and type set to '.' , where in the original there is id, type, mon_nstd_flag, name, pdbx_synonyms, formula and formula weight. If _chem_comp.type is set to '.', it doesn't allow secondary structure rendering in molstar. My current approach is to update the _chem_comp category after updating the structure, which brings the possibility back, but I just wondered if there was a way to prevent this information being lost in the process itself.

[wojdyr]

would it work if _chem_comp was missing entirely? This would be the easiest. If not, I'll try to add _chem_comp.type next week.

[m-crown]

Can confirm deleting _chem_comp allows the secondary structure to be visualised. Nonetheless, it would be nice to keep the info when transforming, but I'll mark your suggestion as the answer. Thanks!

[wojdyr]

_chem_comp.type can't be inferred from a pdb file and is also not available in mmCIF files that are not coming from the PDB, so while preserving it would help, it'd be only in some scenarios. There is an option to skip it when writing:

structure.make_mmcif_document(gemmi.MmcifOutputGroups(True, chem_comp=False))

[m-crown]

Perfect, thank you