2025-05-28 JAC/LW Check-In

PR: 435: TM CCD Errors

'geomeTRIC run_json e' 11 cases

• Existing issue with xtb discussed in previous QCSubmit meeting. Will not pursue

'Error getting task r' 97 cases

• We contributed a PR to QCFractal to provide more information on this exception.

• I can copy the string and then read write the json fine…could be mixing the xtb output with json? xtb is deprecated so we won’t bother fixing.

• I’m thinking I should open an issue in QCADS or QCF (I’d ask Ben) for posterity and then close it.

  • LW: Ok sounds good

Hessian dataset, PR453:

• Running 3 replicas per MW, 2k more have finished, can ramp it up or still fine to let it cruise?

  • LW: Sounds fine to me

Genentech Atomic Spin

You may have read the email to Richard about the need for clarification. It turns out that there are two types of spin, so we can offer one but not the other:

Atomic Spin Population:

• Already in current output under “Lowdin charges” heading, we recently had a PR accepted to allow us to pull that quantity directly in PSI4.

Atomic Spin Magnetic Moment:

• A wavefunction product that is output in ORCA but not PSI4. We could in theory get it using the post-processing tool MultiWfn but only if absolutely necessary.

  • LW: QCF tends to be rigid so post-processing on the node seems like a long shot so you can ask Ben but I doubt it. We need to hear back from Richard how important this is and whether it requires that level of effort.

Meta OMol25 Dataset: ASEDatabase

• Sent ASE Databases to Chris:

  • 1,293,895 non-bio TMCs from the training dataset

  • 40,120 bio-TMC from the validation dataset

I don’t like how they handled spin, and for this reason I don’t think it’s a threat to our Genentech effort. They chose the highest spin and optimized (for 5 steps) on that, without regard for whether another spin state had a more favorable energy.

Sage Full Dataset Archival:

Everything is fine

Should I edit the Sage 2.0.0 dataset to adhere to standards?

• LW: You can wipe the previous Sage 2.0.0 datasets from QCA and QDS and re-do consistently with Sage 2.1.0 and 2.2.0

What do you think of the README metadata section?

• LW: That looks fine

Working on Archival SOP, Zenodo Metadata:

Is the location of Archive.md ok?

• LW: That's fine

• Title: Same as force field

• Authors (Contributors): All submitters, curators, and generators? Need names and ORCIDs, how to split? With authorship of paper as Authors (and any generators that aren’t in that list) and others as Contributors?

• Description: Copy from release notes? README? Paper abstract?
  - Sage 2.0.0; Sage 2.1.0 (No release); Sage 2.2.0
  Make a Sage 2.1.0 release? Does that make sense since each version has its own repo?

  • LW: You can just use what you used for the QCA dataset.

• License: Creative Commons by 4.0

• Funding: From Publications (cumulative?)

• Related Identifiers: GitHub link

• Journal: Paper DOI

Correct that only lead team has zenodo access?

• Someone in lead team should give JC Zenodo keys

Maybe I won’t do Zenodo until you’re back? Will finish prepping.

• LW: You can make draft uploads and then me/Jeff will have a look

It looks like we can easily access the data in a view without a client.

• LW: I’ll look into it and send you something.

Thoughts / priorities for your absence?

1. Genentech datasets

2. Do what you can with Archival

3. Next week set the agenda for FF-Fitting and Biopolymers and cancel meetings if no agenda. Both meetings should be record.

   1. Record to computer > upload to Drive > link to meeting notes

      1. https://drive.google.com/drive/u/3/folders/1rXTSR6EdjYPIm-64eWUlxrJtCipZYTcY

Zenhub:

Added Tickets:

• Look at MolAssembler - for pi-bonded ligands

Proposed Tickets:

Move to In Progress:

• #795: Combine pipeline elements into LTS-SOP

• Docker image

Close Tickets:

• #793: Determine file conversion strategy from output of qcportal to a future proof file format

• #791: Determine qcportal capability to download datasets locally