2022-05-31 QC meeting notes

Updates from MolSSI

• BP: DD and I’ve been doing working sessions on next QCF and making changes to qcsubmit

  • DD: Thanks for helping with the changes.

• BP – Looking at hiring one of two postdocs. Waiting on letters of rec. If both are interested we may recommend that OpenFF hire the other one (could be a good backup if planned OpenFF hire falls through)

Throughput status

• New OpenFF sets from Jessica:

  • OpenFF multiplicity correction optimization set v1.0 - 400 opts, complete.

  • OpenFF multiplicity correction torsion drive data v1.0 - 99 TDs, seems to be stuck at 230 opts, need to dig in.

• New OpenFF sets from Chapin:

  • OpenFF Protein Capped 1-mers 3-mers Optimization Dataset v1.0 - 423/759 opts complete.

  • OpenFF Protein Capped 3-mer Backbones v1.0 - 0/54 TDs complete.

    • CC – These are big molecules so they’ll take a while

• SPICE sets: around 63K calcs last week

  • SPICE PubChem Set 4 Single Points Dataset v1.2: 122716 from 114606, 34 more remaining.

  • SPICE PubChem Set 5 Single Points Dataset v1.2: 55874 from 0, around 67K more remaining.

• PB – Should we reprioritize?

  • DD – I can update priorities here. Multiplicity sets have openff-high priority. I’ll switch Lilac over to prioritizing the openff tag, followed by openff-tscc, and finally openff-spice.

  • PB – Should we do 50/50 openff vs. spice?

  • DD – I’ve switched over Lilac priority to the internal sets, and spun up a lot of PRP containers. Could also throw in MPI compute since JM’s torsiondrives are smaller so we may be able to burn through them.

  • PB – CC’s 3-mers set is larger, will we need to keep those off MPI?

    • DD – Not sure, it may fit OK

    • PB – Could be good to run a test manager and see whether we hit resource limits

    • DD – Will do.

User questions/issues

• PB: Can we pass custom basis sets to QCF?

  • BP: Not yet, I think qcschema can handle that but there are no engines that may process it as of now.

  • …

  • PB – If I don’t use density fitting but I just use SCF PK, would that work?

  • BP – I’m more concerned about how you’d specify the basis set string. If you called it a “custom_basis”, then that could be overridden/lead to a collision in the future due to the deduplication system. Which basis set do you want to use?

  • PB – Looking at adding additional polarization functions on top of DZVP for specific atoms like Sulfur. Instead of making everything triple-zeta, adding a few functions for problematic atoms

  • BP – That’s not really possible, at least not in a relatively straightforward way.

  •