Page Comparison

Updates from MolSSI

Ben

• One PR for lowercased keywords on QCFractal

• Need to prepare yet PR for index on keywords in storage socket for faster queries

• Added into branch fixes made to live public QCArchive

Torsdiondrive dataset for dipeptides

Chapin

• having a hard time getting proline conformers out of OpenEye; giving myself one more week to get these, then will proceed without prolinegg

• there’s a bug where OE fails to recognize proline as an amino acid group, so its tools for residues don’t just work

  • had to work around to try and generate these manually

  • still trying to get reasonable conformers

Dipeptide set from Eastman, Chodera

Jeff

• Github link macro
  link https://github.com/openmm/qmdataset/pull/5

• 400 * 50 = 20,000 single point energies

• JW: think it makes sense to make this a disposable dataset that we move to submission quickly, then allow Eastman, Chodera, to look closely at issues that result

QCArchive advancements

Jeff

• JW: got a grant that’s split between OpenFF, MolSSI; next need to determine detailed scope for what we do

• DD: Jeff and I meeting next week to hash out detailed scope for OpenFF; thinking STANDARDSv3 implementation, in-server error cycling, etc.

• BP: enough to hire two people, postdocs or software scientists

  • been working on determining what we want to do on MolSSI side, capacity planning

  • problem is, by the time we find someone, I may already have many of these things solved

    • many of the things in the grant already on docket, working on in the next branch

  • there is a job posting on OpenFF’s side; will have to follow up with John on this

  • been considering the value of hiring an ML/AI person to diversify QCArchive’s value proposition

• JW: think we have distinct enough needs on each side, so having one person in each org may make the most sense

• BP: think it says both people would be working under me?

• CC: could it make sense for this person to make more tutorials, cookbooks?

  • BP: since this is going to be billed as a postdoc, needs to be a science person, and it needs to advance their own career; this presents a challenge

• BP: kind of person I have in mind fits into Aim 1 of the grant

  • JW: I think people who have worked on community challenges or have written reviews may be good candidates?

  • BP: many of those people will probably be professors already…

  • JW: true, and AI/ML folks of this level are probably sought after by industry, which we can’t compete with typically

  • BP: may seek out previous fellows as candidates

• DD: Jeff and I will meet next week to detail what OpenFF wants to do on our end; sounds like it will be a distinct person from the MolSSI hire