2023-10-31 alchemiscale Working Group meeting notes

alchemiscale.org user group

• user questions / issues / feature requests

• compute resources status

  • DD – Fully saturated with tasks, burned through about half the backlog yesterday. Running a lot of jobs on NRP. MH will spin up workers on OSG.

  • MH – I tried launching this yesterday and some technical issues had come up, it may be a bit of work for me to get this updated. Only getting a few workers on the OpenFE instance now. May need to tune resource requests.

• current stack versions:

  • alchemiscale: 0.2.1

  • gufe: 0.9.4

    • DD – Should I upgrade to 0.9.5?

    • RG + IA – Needed for OpenFE 0.14 energy stuff

    • IA – 0.14 adds HFEs, also adds ability to do back and forth analysis, MBAR overlap matrices. But it’d be great to finish the current queue.

    • DD – Yeah, can finish this up before I upgrade. Are there settings changes that will affect the data model on the server?

    • IA + MH – Unsure. This isn’t something we test for.

    • DD – I do smoke tests when I try to update, so we’ll know if there’s data model trouble. I’ll raise that in the channel

  • openfe: 0.13.0

    • (See above)

  • perses: protocol-neqcyc branch

    • DD – Will replace this with feflow once a release is made

  • openmmforcefields: 0.12.0

    • DD – Has anyone tried feeding a whole OFFXML yet? That might be JHorton’s thing.

      • JS – I don’t know if he’s tried it yet, I’ll try to get him in this meeting.

    • IA – There may be an issue with feeding OFFXMLs in (on the interface openfe and openmmforcefields)

    • JC – There’s a couple of issues that came up recently that I could use MHenry’s help on - I’ve tagged him on issues.

    • JW – I may have found an issue on some lowercasing behavior in the new OMMFFs, though this may be in QCSubmit/QCEngine so I’ll submit an MRE if I can isolate the issue.

    • IP – Testing new OMMFFs using PErses is going well, no issues yet.

JW : alchemiscale working group governance

• Goal of existing governance structure was to have major stakeholders able to propose/approve/block scope changes and vote on clarifications to Dotson’s work on Alchemiscale.

• We’ve largely completed the mutually agreed-upon goals laid out on that page.

• This meeting has been a useful touchpoint to sync up between users and developers, though it’s also been a source of confusion when messaging gets jumbled with different channels.

• Should we:

  • Keep stakeholders/governance structure?

  • Keep this meeting as a user group and informal developer sync up?

  • Define new goals?

• JC – Still hoping for a few major features that we’d laid out initially, like F@H. THe contract that we have with Datryllic is milestone-based, not sure how this fits into the shared effort structure. So one idea would be that we all have our own milestones set with Datryllic, and we use this meeting to sync/align efforts. So definitely keep this as a user group meeting, but I don’t know how to mesh the current governance structure with mnilestone-based contracts.

• RG – Re: confusion, there’s been some cross-contamination with conversations with Lily. I don’t see a need to hold onto approvership.

• JC – Re: coordination, we have funding for 1.5 years - are there shared needs we can contribute to?

  • RG – I think alchemiscale is mostly feature complete as far as our needs go.

  • JC – Maybe running on F@H would be useful for large-scale benchmarks?

  • RG – I think OFF is more planning on the F@H benchmarking. We’ll need to sync with OFF leadership to understand overlap there.

• JW –

  • JS – I think the current arrangement works well for ASAP. We have goals in JC’s contract wtih Datryllic. Current needs are mostly around living networks - should we discuss those in this meeting?

  • DD – I think we could still use this group for that purpose.

  • JC – Can I share the current contract+milestones?

    • DD – Sure, feel free to share milestones.

JW – So, we propose

• Dissolve governance structure

  • JW + JC + RG – Yes

• Continue this meeting as user group; (frequency?)

  • JS – Could split between dev/user meetings, maybe each one is semiweekly.

  • JW – That sounds fine. DD, could you take over this calendar meeting and make the topics alternate? Can still use my Zoom link.

    • DD – Sure, I’ll update this calendar event and run future meetings

  • JC – I think we’ll keep most of the value by continuing the user group meeting.

• Further goals+governance will be proposed in the future if anyone wants them, not assumed there will be any.

• JW – I’ll make sure project page gets updated with the change.

  • DD – Do consider adding F@H to the project goals as an “incomplete” item.

IP : Protein-ligand benchmarks working group update

• IP – No major update, everyone’s been traveling at different times

  • JC – Let’s sync up in person when you’re back in new york

• IA – MO, I think I’m waiting on something from you

  • MO – I don’t think I got that - let’s sync up.

• DD – Can we prune release milestone?

  • IP – Yes, most stuff can be moved to a subsequent release. #93 is the one critical thing.

• IA – OpenFE needs to review what we commit to on this front, we’ve spent a lot of time and need to discuss internally.

IP : feflow development:

• IP – Current state is that we have two protocols with the codebase. They use GUFE objects… Need to separate the stup object sothey can be run on F@H. There’s also a some stuff we need to port to openmmtools. There’s also some stuff in protein mutation protocols that could use the openmmtools setup code.

• JC – When we had gone through this before, we’d identified that there were components that need to be split intodifferent parts of the ecosystem, and we may need to review that. Solvating a PDB is different from setting up an alchemical system, and those components belong in different places. IT won’t be as simple as moving the setup step to OMMTools

• IP – What I mean is that having objects like hybridtopologyfactory in OMMTools.

• JC – Right, that needs to get into OMMTools.

• DD – Sounds like you have a roadmap, are there issues filed so we can track+triage these? We can use this group to coordinate efforts.

• IP – Current notes are here

• IA – A question on this - Should this be handled just in the Perses/OpenFE meeting instead of here? Would prevent confusion from different conversation happening in same meeting.

• IP – I prefer to discuss in detail in the other meeting and just summarize here.

• JW – Yeah, I’d be happy for those discussions/decisions to happen elsewhere, but it’d be great to get updates on what’s decided here.

• DD – Yeah, I’ll change this item to be more informational than about discussion.

alchemiscale development : current sprint spans 10/25 - 11/6

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

• coordination board : alchemiscale : Phase 3 - Folding@Home, new features, optimizations, targeted refactors

  • call for volunteers for available issues

• alchemiscale 0.3.0 milestone:

  • DD – FEFlow 3 – IP, I reviewed, anything else I can do?

    • IP – Nope, I’ll move it forward.

• alchemiscale-fah: 0.1.0 milestone

  • DD – I worked with Sukrit last week and got some feedback on proposed approach. JC, I know you’ve been traveling but I’d love feedback on this. Some Q’s for Joseph Coffland?, but your feedback could be great. https://github.com/openforcefield/alchemiscale-fah/issues/1#issuecomment-1776593087

    • JC – Sure, I’ll put this on my to-do list. Will try to respond today

    • DD – Great. I’ve heard back from Joseph and will meet with him and Sukrit to discuss. Some misalignment with adaptive sampling API, so I’m thinking about technical solutions to that.

  • DD – fah-alchemy 7 – noneqcyc protocol. This is in progress.

  • DD – I’ll start mocking out starts to the Good First Issue items on the milestone, but do let me know if you’re interested in helping.

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

  • DD – Goal continues to be interfacing with F@H by the end of the year.