Add PDB test suite

[Yoshanuikabundi]

We need a collection of challenging PDB files to test loading.

Ideally, we would have a suite of PDB files covering:

Phase 1:

• Basic PDB file with a single peptide chain covering the canonical 20 AAs using standard names, standard terminii and side chains at pH 7, contiguous atom and residue serials starting at 1, TER entries, etc. #23
• Add test of a PDB file with Disulfide bonds (described by CONECT record) #24
• Add test of a PDB file with ACE/NME caps #25
• Add test of a PDB file with neutral N and C terminii #26
• Add test of a PDB file with All charged side chain variants #27
• Add test of a PDB file with Every canonical amino acid #28
• Add test of a PDB file with A non-canonical amino acid from the CCD #29
• Add test of a PDB file with Water, NaCl, other Sage-supported ions #34
• Add test of a PDB file with A ligand found in the CCD #35
• Add test of a PDB file with Ligand not found in the CCD (for loading with unknown_molecules) #36
• Add tests for files with too many atoms/residues #48
• Add test of a PDB file with Missing TER entries... #46
• Add test of a PDB file with ICode entries #45
• Add test of a PDB file with Discontiguous residue IDs #43

Phase 2:

• Add test of a PDB file with RNA #38
• Add test of a PDB file with DNA #39
• Add test of a PDB file with Nucleic acid termini #40
• Add test of a PDB file with (nice to have) oligosaccharides #41
• Add test of a PDB file with Named according to alternate atom names #42
• Add test of a PDB file with AltLoc entries (JW: It's OK if this behavior is "raise an error") #44
• Add test of a PDB file with A non-canonical amino acid not from the CCD (for loading with additional_substructures) #30
• Add test of a PDB file with With D-peptides #32
• Add test of a PDB file with With a cis peptide bond #33
• Add tests of PDB files from common software packages #47
• Add test of a PDB file with Completely custom polymer not found in CCD (for loading with residue_database) #37
• Add test of a PDB file with In strained poses #31

And some infrastructure around them:

• Developer documentation describing...
  • Which PDB files test which features
  • Where PDB files were sourced
• Test comparing to legacy PDB loader
• Direct tests for the above features

[j-wags]

Following today's ad board meeting, I had the following conversation with a partner company about their envisioned use case:

Partner
Hi Jeff - the modified residue workflow sounds very interesting and I’m looking forward to seeing the prototype! We’re interested in ligands with covalent warheads reacting with Cys residues. We’d like to be able to provide a “reacted” form of the ligand containing a S atom (usually with a single H attached in order to satisfy normal valence requirements) and get back a template for that ligand covalently attached to a Cys, where the S atom from the ligand now corresponds to the S in the Cys, and the two terminal H atoms from the ligand and Cys have been dropped. Obviously this could be generalized with SMIRKS patterns, etc. but we’d be fine with doing the atom identification ourselves and making the desired bond formations explicit. Does this seem reasonable?

Jeffrey Wagner
Yes, this makes a lot of sense. I think making a template that way is a great use case, and I’m pretty sure we can include it in our scope. Separately, will you have coordinates of the ligand in your desired binding pose as an input for this workflow as well?

Partner
Yes, we can provide poses in which the reacted form of the ligand is bonded to the Cys (so that the S atoms overlap).

[j-wags]

OpenFE's last push toward this provides a capped TPO residue and a prepared protein structure with TPO (CDK2).

[j-wags]

One of our ad board members recommended the COVPDB database from U Freiburg - At least some of the entries appear to be protonated too.