2023-09-13 All Hands Meeting

Updates

• JW – Software ecosystem

  • Some messiness with Pydantic v2 release recently, Matt is leading the charge to update our ecosystem.

  • Big breaking QCFractal release happened yesterday:

    • JW is working on fixes for QCSubmit to continue working with the breaking release

  • QCSubmit is no longer broken for working with legacy QCFractal

  • BespokeFit had an upstream dependency incompatibility, that’s been fixed so conda installs should work again.

  • Nagl will soon be available publicly in the openFF toolkit.

  • OpenFF toolkit has prototype support for custom substructure loading - It’s basically a subset of SMARTS but with explicit formal charges and bond orders.

• Fundraising updates (DLM):

  • (New partner not public)

  • Ongoing discussions - brief list - Talking to US Merck, SandboxAQ, Novartis, Biogen, a few others.
    
    

Recognitions

• David Hahn’s paper benchmarking binding FE calcs has gone to preprint and been submitted to JCIM. Glad to see that going out. Current state of open source force fields in protein-ligand binding affinity predictions

• JE – Open mic for shout-outs - Would love to have folks bring in cool tidbits of progress. Could turn these into social media posts.

• JW – Paul Labute’s work on force field fitting is pretty impressive.

Q&A

• DC: On BespokeFit (relevant for our upcoming training workshops) should we use an all-in-one environment?

  • JW: Yes.

• DM – Really cool xtalpi FF paper that just came out. We’ll probably run something like a journal club when LW is back - Would love to discuss which of their ideas we can apply to our work. Is that this paper? Development and Comprehensive Benchmark of a High Quality AMBER-Consistent Small Molecule Force Field with Broad Chemical Space Coverage for Molecular Modeling and Free Energy Calculation

• JW – The scale of alchemiscale is truly impressive.

  • DD – up to 500 workers

• LM – What is alchemiscale?

  • DD – It’s a distributed way to run free energy calculations. Allows for stuff to run on a wide variety of compute backends/clusters.

  • Lets us make use of the “table scraps” of academic clusters.

  • Weekly “Alchemiscale” meeting is infrastructure-focused.

  • Currently we can use HPC / Kubernetes clusters.

    • Major push for the fall is to run on F@H

  • Strategic plan is to make evaluating candidates force fields massively parallel

• JE – From OMSF level, we have Julia from Rosetta on OMSF staff now. She led a big benchmarking push for Rosetta to make sure the releases were held to standard measurements. It may be good to get her input on our benchmarking strategy.

  • JC – Is she still at Flatiron?

  • JE – I think she announced she was leaving.

  •  

• JC – Landscape analysis for NIH resubmission - Would love to have eyes on this and make sure we’re not missing any big efforts.

  • https://docs.google.com/spreadsheets/d/1e1nGFvSwu-DL04pzQxf_Ehke4TQ6XAa3yuK-ovo_IkA/edit#gid=0

  • .

  •  

Question about QCSubmit

JW, TG, JC, BW

to_basic_result_collection() bug

BW – kwarg driver=hessian

JW is not the only person who finds this confusing.

Ask Chris Iacovella