2021_10_13 Thompson/Wagner check in

High-level updates

• JW – Will be working with IP on PDB-loading in the Topology refactor branch.

  • JW – I could instead be working on “friendly topology metadata assignment” behaviors, like “what if a topology has one molecule with residue names+numbers, and another molecule without it, and we need to export to a format where it’s bad if made-up-info collides?”. Do you need anything else done first?

  • MT – I see it being a long time until the topology branch hits production, and all of my work is based on the topology branch. So I can’t make an interchange release with this new functionality until the toolkit has the release. So I’m in favor of avoiding the complex stuff now, and just putting out a minimal toolkit package with the topology refactor asap.

  • (General) MT is concerned about the lack of solid goalposts to getting the topology refactor in a stable release. JW is concerned that timelines/effort requirements for implementation are still unknown and so we shouldn’t commit to a release date. We’ve agreed to start making nightly builds of the topology refactor branches of OpenFF toolkit and Interchange. omnia/label/openff-topology-refactor

• MT – Working on AMBER export. The formats are a mess. This will be painful. I have some terms working but expect that it will take a while to get everything out.

• MT – Plans for productionizing ff14SB?

  • JW – Will need to spend another week on this. I need to verify parameter deduplication PR and make sure there aren’t possible manglings of cheminformatics representations of proteins.

  • MT – I’ll start using it to test things like alanine dipeptide energies using toolkit vs. interchange.

  • JW – Could also use OpenMM’s ff14SB implementation as a reference

•  

PRs and quick blockers

• Additional needs from topology refactor?

  • MT – None yet. It’s kinda annoying to iterate through atoms that don’t know their own topology index.

  • JW – The object structure would get a bit more complex if atoms knew which topology they were in (the current implementation of the topology doesn’t have molecules know their topology like in slide 4 of this), but it’s possible that it would be worth it to avoid the API break

• MT – I also got multiple molecules to be written to a GROMACS file, but haven’t tried running them yet. There’s some deep peculiarities about the assumptions regarding residues and molecules in these output formats.

  • JW – If you could keep a list of these peculiarities it would be really helpful - This would let me implement “friendly” export behavior to different ecosystems.

• Progress on docs PR?

  • MT – JM got back to me, I’ve been working on implementation and haven’t gotten back to him yet, I’ll do so this afternoon.