2024-09-30 Mitchell/Wagner Check-in meeting notes

General updates

• JW

  • Added warning for multiple molecules to toolkit

    • JM – I thought our attitude was “we don’t support multiple molecules”

    • JW – Yeah, but we need multi molecule molecules to load SPICE and for some possible representations of organometallics (which people will begin working on shortly)

  • Expect Interchange 0.4 to drop in the next few weeks, requiring pydantic 2.

  • If you’ve got any ideas, feel free to start a list of possiblemini-workshop topics. We’ll want to run those early next year and it may be easiest to come up with ideas far in advance.

  •  

• JM

  • Update on Rosemary?

    • JW – Given how delayed it’s been (relative to our expectations) we asked the governing board when we should make a formally-supported PTM pathway using ff14SB and maybe some additional infrastructure. CC is working on a new major effort and should know by late October whether his current FFs are on the right path, so at the Nov gov board meeting we’ll have a go/no-go vote on making a formal PTM simulation tool.

    • JM – Would rosemary use NAGL?

    • JW – Yes (or very likely), CC’s fitting experiments have started using NAGL as the charge method for the protein FF.

    • JM – Is it possible that Rosemary will just be the NAGL release?

    • JW – (I think I’m remembering this right) NAGL will be a technical detail, not corresponding to a FF release. It’ll just become a toolkit that provides AM1BCC charges

    • JM – Will we fit BCCs or do other charge optimization after the first-gen NAGL AM1BCC model?

    • JW – No, NAGL second gen will use RESP or something else really fancy.

    • JM – I found a FF (GROMOS) that directly trained charges to simulation.

    • JW – Not a bad idea, but we prefer to keep the charges physics based so we can debug when needed.

    •  

    •  

Trello

https://trello.com/b/dzvFZnv4/infrastructure

PDB census (if applicable)