Versions Compared

Old Version 3

changes.mady.by.user Jeffrey Wagner

Saved on

compared with

New Version Current

changes.mady.by.user Hannah Turney

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

Item

Notes

HT slides

  • Post slides here:

    View file
    name11Jan24_JW_KCL_update.pdf

  • Discussion of workflow:

    • Making kekule structure of copolymer

    • Making 3d coords and assigning physics parameters

    • Making higher-level morphology of whole system (nanoparticle vs. gel etc)

  • MM+HT – Several options here. Could use Shirts lab tools for polymer building but some limitations around branching/arbitrary consistencies. Could use ambertools/tleap with custom monomers.

    • JW – May be good to do “quick and dirty” implementation initially to get an understanding of problem domain, and then plan to refactor each year as the parameters become more known.

  • Data sharing - If working with Shirts lab, can’t share structures since there’s IP protection.

    • JW – Maybe come up with similar molecules that can be publicly reported for debugging? This will be generally difficult if you can’t provide reproducing cases for bugs.

  • JW – Proposal - Use RDKit reactions to construct polymer, generate conformer using RDKit, load into openFF using from_rdkit, (possibly use packmol wrapper in Interchange to construct simulation box of several molecules) use nagl to charge, simulate.

  • MM – Interested in participating in validation? Want to compare different charge models etc.

    • JW – Not interested in validation but Lily Wang would be. Charge models I can think of would be Nagl and Espaloma for whole-molecule, or for constructing librarycharges Antechamber AM1BCC, RESP (OpenFF has an ugly solution for this, I can point you to this if you run out of other options), and maybe an XTB charge model (if that exists?). Our FFs are trained against AM1BCC charges so there’s a chance that performance would get worse with other models.

    • JW – OpenFF’s current FFs are trained against TIP3P water.

  • JW – Happy to help put together a prototype to go from polymer definitions → simulation at any point. Let me know if I can help with a screenshare, or work on a prototype.



...