mmcif.find_template_residue() fails on Mac for template names containing '_'

[Tristan Croll]

Reported by Yang Lee, just confirmed by me in ChimeraX 1.8. In Windows:

open TRP_LL from ccd

... works correctly, but the identical command in the Mac version fails with "Unknown CCD ligand name". Digging into the API, I can confirm the failure happens in mmcif.find_template_residue() - apparently for any name containing an underscore, but only if the relevant template hasn't been downloaded. For templates that already exist in ~/Downloads/ChimeraX/CCD there is no error (I have a series of HIS_...cif templates in there, indicating that fetching of these *used* to work). Some sort of obscure Unicode issue, perhaps?

[Greg Couch]

In the daily build, it's failing on Windows with:

open TRP_LL from ccd ignoreCache true

And this presumably has been failing since December 2022 when the PDB broke ligand-expo.rcsb.org and told us to use ​https://wwpdb.org/pub/pdb/refdata/chem_comp/ instead. ligand-expo appears to be working again.

I've sent email to the RCSB asking for clarification about what we should do. In the interim, I've added code to fetch the CCD file from ligand-expo if there is an underscore in it.

FYI, TRP_LL is from the "Protonation Variants Companion Dictionary" given at ​https://www.wwpdb.org/data/ccd. ChimeraX only includes the C- and N-terminal amino acid variants (_LEO2H and _LSN3).

[Greg Couch]

The PDB is working on retiring the ligand-expo, so we will need to change the code again someday.