2025-01-27 Science team meeting

Project updates

JCl

Previous Week:

• TM-FF

  • DS1-PDB CCD:

    • of the 154 structures in the primary dataset, 47 make it through the xyz2mol_tm script.

    • Tried again to adapt script to have a timeout so that a manual exclude list isn’t needed for larger datasets… that still isn’t working

  • DS2-COD

    • Finished draft of parsing scripts…but now not confident that the contents are relevant

      • LW: In the standup let’s move this back to the Science Backlog

  • DS3-CCD

    • Access update? Maybe skip it

      • LW: Still in progress, let’s focus on other for now.

    • Paper only directs readers to tmQMg repo with ~75k “structures“ (xyz files)

    • The xyz2mol_tm repo contained SMILES for tmQMg and ~220k structures of CSD.

    • We can:

      • Take xyz files and smiles for ~75k * 0.7 of tmQMg

        • LW: Leave this since for now

        • JCl: I’ll leave this for working with Chris' dataset since it’s the same dataset

      • Use cleaned CSD SMILES to generate dataset of ~220 k … recall that geometries are bad… can we use UFF to fix? (like Avogadro does)

        • LW: Let’s try it, do what Brent did and add that. Although UFF might make everything “the same” so use an RMSD comparison of conformers generated and minimized by RDKit.

        • https://www.rdkit.org/docs/source/rdkit.Chem.rdMolAlign.html#rdkit.Chem.rdMolAlign.GetAllConformerBestRMS

        • https://www.rdkit.org/docs/source/rdkit.Chem.rdMolAlign.html#rdkit.Chem.rdMolAlign.GetBestRMS

        • https://mattermodeling.stackexchange.com/questions/11835/how-to-pre-optimize-metal-organic-complexes-in-rdkit-using-uff

  • Sent message to EPA contact about ToxCast, haven't heard back.

  • Consider looking at other databases?

    • BCPC

    • HugeMDB

    • LW: Get the pipeline up and something through and then we can wrap back around to these.

• Dataset Longevity

  • Discussed in “iteration planning” I said it was a lower priority with the current issues. ok? I’ll continue chipping away if other tasks stall before I can request assistance

  • LW: Not an “MVP” but still a “Nice to Have”, we might bring this up as a topic at the onsite, so next sprint prepare a 10 min or so talk.

• NRP Issues

  • Still not resolved. Applied python API to isolate that it’s a MolSSI issue, messaged Ben today and he’s looking into it

  • LW: Let me know if you need anything from me. It sounds like this is very disruptive.

• Given QCSubmit datasets haven’t moved due to NRP issues, cancel meeting? Is Shirt student attending? Seems so.

  • LW: Post agenda items and let people know about our NRP issues. That will be a walk-through for Anika

Project updates

LW

• Past week:

• This/next week:

• Move/cancel meeting?