2025-03-20 JAC/LW check-in

A way to use UCI login for MS Office suite?

• LW: asking about Overleaf anyway, will add MS Office to it. Also add Jen to Slack

Thanks for reviewing Chodera TM Dataset so quickly:

• Chris wants to update it so need to incorporate changes

• “return_result“ (See below)

  • LW: Lexie used “orbitals and eigenvalues” but according to Ben we could “easily” store 250GB-1TB. ~4 conformers x 130k mols 4 conformers x 50k mols -> 1 TB

  • JAC: Ok, we will do wavefunction=”none” then

• JAC: Is there a way to perform DFT for the metal at one basis set and organics at another?

  • LW: Oooh good question, start with working with psi4 to run that, then ask Ben how to implement in psi4

Levels of Theory

• Went through the ~35 papers I downloaded that used DFT of similar systems. Most used B3LYP with low Pople basis set, a sig. fraction of which used LANL2DZ for the metal. This is a fine but outdated model Chemistry. Learned about new-to-me functionals and basis sets (very small number). Want to go through them.

• Peter Eastman voiced some opinions on our existing level of theory for our charged model and alluded to faster/better alternatives. What do you think about if I reach our to him to compare notes?

  • LW: Sounds good, if you set that up I’d like to attend.

For PDB CCD PR

• Can I update seme how for unit conversion on molecule geometries? Didn’t realize QCElemental requires Bohr

  • LW: Yes, you can delete old entries, update with new entries and submit PR to overwrite the scaffold.json

• After genentech meeting, wondering if xtb is the best thing to do with Ferrocenes? Expected to fall apart?

  • LW: Instead of doing that: add constraints

• Jeff got xtb image working, yay!

Recommend code for characterizing chemical space?

• ADMET is working on something similar and have an active learning component (will work well with QCA projects?) I don’t see much on their GitHub

  • checkmol (Checkmol/Matchmol Homepage ), exmol (Getting Started — exmol documentation ) may not have much support for metals

    • The profiling of chemical groups is somewhat new in exmol and may not have docs

  • JAC: worst comes to worst I’ll hand pick

SBIR STTR Update:

• Attended webinar and I think a partnership with a company would require an STTR

Data archival

• Do you think what Ben shared about field name changes will alter our archival plan?

  • Yes, we should wait and see what happens before archiving

  • Sage releases are not a high priority, we can move those to backlog and focus on scaffold validation etc.

• I was thinking that instead of packaging the docker image with the zenodo instance, we will reference a docker image from a different zenodo instance?

  • Are you open to it? We can wait until I try to make it to see how big it is

    • LW: If less than a couple GB, then it can be packaged together

  • JAC: BTW I’m 95% sure QCSubmit added those pesky keywords as defaults

  • LW: Put in an hour or so to see if you can turn that off, if not do it locally
    Let’s get the scaffold.py into QCPortal so that we can have proper validation without QCSubmit. All related tasks can go to backlog

Conferences?

• Probably leaning toward ACS – August

  • “Updates at OpenForcefield”, maybe focusing data generation? NAGL, PTM, forcefield improvements, improvements and philosophy of dataset strategy.

Zenhub:

Tickets:

• Add constraints to xtb calculations for PDB CCD

• Looking for chemical space assessment tools (checkmol (Checkmol/Matchmol Homepage ), exmol (Getting Started — exmol documentation )

• tests for scaffold PR

  • Need to write tests if we plan to rely on it and want Ben to fix any issues that arise.

  • Should be easy with my recent experience using FractalSnowflake

• Assess ability to use basis set for metal and organics separately.

  • start with working with psi4 to run that, then ask Ben how to implement in QCFractal

"orbitals_a"

"orbitals_a"

"orbitals_b"

"orbitals_b"

"eigenvalues_a"

"eigenvalues_a"

"eigenvalues_a"

"eigenvalues_b"

"eigenvalues_b"

"eigenvalues_b"

"occupations_a"

"occupations_a"

"occupations_b"

"occupations_b"

"fock_a"

"fock_b"

"density_a"

"density_b"