The glycoprotein builder / preprocessor changes the chain ID of some residues.

[gitoliver]

Dar6I192F_align_cutnoTer.txt

I generated this file in UCSF Chimera from pdb ID 9bdf.
To reproduce the issue: Rename it to .pdb instead of .txt and upload to glycam.org/gp. Add Man5 to each N-linked site. Increase search exhaustiveness to 10. Neither of those things likely matter, I suspect a bug in the preprocessing code that adds ACE/NME caps. Download the default structure and note that towards the end of each chain the chain IDs get swapped for some amino acids.

Here is the resulting file (renamed to .txt so that github will upload it):
https://glycam.org/gp/download/9a7a28fd-d888-4cf5-a289-210b586f73d4
Dar6I192F_align_cutnoTer_GP_default_structure_f46d7d2e-1411-4efa-bebf-8c2ab1e09a32.txt
Dar6I192F_align_cutnoTer_GP_project_f46d7d2e-1411-4efa-bebf-8c2ab1e09a32.zip

Go to the first OXT entry and you see that the last few residues have the wrong chain character.

[gitoliver]

Note default.pdb and 9bdfToDar6_align_cutnoTer.pdb are renamed to .txt to allow upload to github.
the_input.txt
default.txt
9bdfToDar6_align_cutnoTer.txt
logBoy.log

Everything looks fine in the logs, see logBoy.log. I guess the mistake happens later during residue renumbering?
@vraid this is code that was changed to use your new data structures. Do you see where the chain IDs might be getting messed up?