2023-06-13 alchemiscale Working Group meeting notes

alchemiscale.org user group

• user questions / issues / feature requests

  • JS – (JS+DD discuss openff-discussions #27)

• compute resources status

  • DD – I notice that we’re getting choked out on lilac now on the preemptible gpu queue and I’ll contact sweta.

    • JC – Looks like there’s plenty of GPUs free

  • JW – For my information, are we still using PRP?

    • DD – Yes, IA submitted some stuff last week and I ran them there. It’s working well.

    • IA – I didn’t submit anything last week.

    • DD – I’ll check on this

      • IP – This could be a good feature request - administrator tools to debug this sort of thing

  • JC – Also, we can later get onto F@H. Sukrit is my group’s representative on F@H.

    • DD – That’s still the plan.

    • JC – Also we can get on cloud resources like EC2, there are some free grants but after that it gets expensive

    • DD – Yeah, but let’s focus on F@H first. Free grants run out and then cost us money.

• call for new users

• current stack versions:

  • alchemiscale: 0.1.2

  • gufe: 0.7.3

  • openfe: 0.7.4

  • perses: protocol-neqcyc

    • IP – Working toward a new release, hopefully today or tomorrow. Unfortunately not with all the changes we need, but those will come next.

  • IA – Do we have a timeline for when we might be able to update gufe and openfe?

    • DD – Not in the next week - The bottleneck is me and my attention. Is two weeks enough time

    • IA – This is just something that DM asked for. So I assume it’s important for OpenFF

    • (discussion)

    • JW – As a representative for OPenFF, don’t rush the release.

JW : custom FFs for Protocol execution

• had confusion on: IA said we are able to test different charge methods, but not currently custom FFs

  • IA said this is because we can pre-charge the molecules before submitting as part of an AlchemicalNetwork, and then

• JC : the OFF FF specifies the charge model; are we overriding this then?

  • we’ve worked out a standard with OE and Greg Landrum on charges

• IA : will follow up if we are using these standards for charges

• JC : is the behavior currently that we are always overriding the charge model specified in the OFF FF? Or is it the other way around?

  • JW : IA, are you able to meet after this call to characterize current behavior, identify actions needed?

  • JC : openmmforcefields does have a hacky way to do this; OFF has a kwarg to do this as well

  • IA : do have a proposed solution in mind that would satisfy user needs, not create unpleasant surprises

• DD – How DO we plan to do custom FFs?

  • IA – That’ll be a longer discussion on the GUFE side.

  • JC – JHorton will be wanting to submit bespoke FF these. As a hack could implement URL support in openmmforcefields.

  • MH – Could have a data package that openmmforcefields can find.

  • JW – Slightly against URL approach since it can lose track of FF contents and break provenance.

  • MH – URL-loaded FFs could be recorded as full FF contents.

  • DD – This will be up to the protocol. Will depend on user

• JW – OpenFF wants this in the next ~3 months. What’s JC’s (JHorton’s) timeline?

  • JC – JH is starting imminently. We’ll put in URL support to openmmforcefields asap.

•  

•  

JC : follow-up on extends support in RelativeHybridTopologyProtocol and NonEquilibriumCyclingProtocol

• IA – JC, IA, and DS are meeting tomorrow

• IP – Can I come?

• (IA added IP)

JC + JS : follow-up on tools needed for AlchemicalNetwork generation?

• JS – IA+JC+DS, I’ll set up a meeting and invite you.

• JS – I’ll bring a wishlist to this and we can figure out whose to-do lists they are on/can go onto.

• DD – Please report back your conclusions here so we can prioritize effort.

•  

IP : Protein-ligand benchmarks working group update

• IP – From previous meeting, we wanted to make a release with minimal working version of edges from dataset, and then gather people to work with what we have.

• DD – I remember you were also recruiting, have you heard back from anyone?

  • IP – I haven’t heard back yet. Also we want to have things working before we have to listen to more people.

• DD – Feel free to reduce scope on these releases, kick issues to later milestones.

• Re RG’s proposal

  • IA – I like RG’s proposal

  • IP – Wouldnt this lead to redundant info?

    • JC – yes. But local could override global?

    • IP – I think we’d prefer something strict. What’s the downside?

    • JC – It’s bigger and more complicated

    • IA – The gain in file size… it would make sense to be able to look at an edge and know everything.

    • JC – I have no problem with the proposed approach.

  • IP – Does the remarks/annotations field reflect what we want to do with this?

    • JC – Yes

•  

• IA – Re my about PR that IA opened

  • JC – If the posesa re good thenw eshould use the geometric scorer since it’ll handle chirality and stuff.

• IA – questioin about radial graphs. I talked to MBOby about this, in the xtal series the ligand may ot always be in the series that was observed. So what should change when we don’t have a central ligand that’s xtal resolved?

  • JC – Could have quantitataive criteria. Could have minimum distance from other ligands, could have it be biased toward big ligands.

  • IA – Do we need a radial graph for 0.3 release?

  • IP – I think we need at least one tranformation network that’s good for benchmarking (gives good results). HOW it is created isn’t important to me right now. Minimum spanning tree would be fine.

  • JC – Agree

  •  

alchemiscale development : sprint in progress through 5/31 - 6/12

• architecture overview : https://drive.google.com/file/d/1ZA-zuqrhKSlYBEiAIqxwNaHXvgJdlOkT/view?usp=share_link

• coordination board : alchemiscale : Phase 2 - User Feedback and Documentation

• alchemiscale 0.1.3 milestone:

• alchemiscale 0.2.0 milestone:

• updates on In Review, In Progress, and Available cards

• (spotty notes See recording ~51 minutes)

• DD – Somewhat short on developer-time.

• JS – HMO is busy on some other stuff, but I’ll return him to work on tickets here when possible.

• LN – I’m busy for the coming weeks.

• JC – We may have funds for an additional developer if anyone knows someone good.

• JC – Also looking to hire someone who knows a lot about chemical/biological data. Would be interested in leads for this too.

new discussion items from ASAP roadmap: https://asapdiscovery.notion.site/Computational-Chemistry-Core-alchemiscale-related-roadmap-9052f215437246f9be4a11abd10f6d71

JC – If we have time next week, it’d be great to talk about components we’re building for alchemiscale.

DD – It’d be good to hear about topics from alchemical network generation.

IP – FYI - I’ll be offline for the next two weeks

DD – working group hiatus July 3 - 21

• JW + IA + JC approve