2021-04-27 Perses/Topology API Meeting notes

Proposed API points

• Topology.replace_molecule(top_mol_1, new_mol, atom_map, core_atoms)

  • JC – Access this as a method to TopologyMolecule instead?

  • JC – Should first argument be a integer index? Would that be simpler than accepting a TopMol?

Return types for substructures?

• (General) – Should be able to provide a capped fragment, with capping atoms labeled

• Not sure about data type for just substructures – JC will provide more feedback after

Hybrid topology API?

• (General) – Could the hybrid topology let users slice something and get the old molecule or new molecule? Could it slice and let us get the before/after coordinates?

  • JW – I’m not sure that we want the hybrid topology to contain a copy of the original/new topology, or have links to them.

• (General) – Perses will make the first “Hybrid topology” API, and once it’s tested and stable, consider merging it into OpenFF toolkit.

• IP – It’s probably confusing to call the thing returned by the OpenFF toolkit a “hybrid topology” – Maybe call it an “atom mapping“ instead?

Coordinate generation? Coordinate storage in Topology(/TopologyMolecule)?

• JW – We’ll want to add a way to store coordinates in a Topology – probably at the TopologyMolecule level.

• (General) – We’re not sure about where coordinates in molecule transformations/replacements will come from, and whether it’ll be easiest to (generate the old and new molecules in 3D outside of the OpenFF toolkit, and then load them as separate molecule) or (take the original 3D molecule that was loaded, modify its graph in the OpenFF toolkit, and then use some tool to generate coordinates for the new/changed atoms).

Capping reactions?

• Which specific reaction for peptides?

  • DR – I’d like minimal caps, because I’ll need to ignore/remove them later – Maybe just cap with “H”?

  • JW – people doing partial charge stuff will want more substantial caps, like methyls – Maybe offer SMIRKS/reaction SMARTS input for this

• Is there a reaction database we should support?

  • We should have one (or a few) default built-in reactions for capping peptides and nucleic acids.

• In which formats should users be able to input reactions?

  • SMIRKS