Individual updates | DN Governing board meeting today is cancelled Still lots of discussion around vsites in rosemary. SB sent a message this morning that implied that he thinks it could work. It also seems like SB would be willing to help LW proceed with vsites implementation. But there hasn’t been any update on that situation other than CBy’s feedback. We have an all-hands meeting next week. Will LW be participating next week? JW – I don’t think so. DN – I’m wondering about responsibilities for LW’s onboarding. So if you could help get get up to speed on what’s happening that would be a huge help. JW – I can lead LW’s onboarding, but it would be great if DN was there for at least half of it since we’ll be communicating our high-level/strategic understanding of the project.
DN – It’d be great if folks could help LW get up to speed with a short presentation on each person’s work. Like a 5 minute talk and a slide or two.
CC Continuing to try and set up benchmark simulations. One of the big hurled is automation - I want to be able to hit “go” and have some automation submit all the jobs to the cluster queue. Sidechain dihedral torsiondrives with capped 1-mers. We’d run into a problem where 3 of the 1-mers were getting stuck due to failing jobs. So I resubmitted with different initial conformers, and those 3 molecules are also getting stuck, possibly around the same geometries. I’m inclined to think this is a problem with the optimization, not the conformer. So I’ll prepare more about this for tomorrow’s QCF meeting. DD – I remember we had discussed what a systematic failure would mean (which it seems like we’ve encountered, since different starting conformers are hitting the same problem). Do we think that what we’re trying to do is fundamentally impossible? CC – That’s what I’m trying to prove/disprove now. I’ll give a more detailed update at the QCF meeting tomorrow.
Met with MGilson to go voer the draft of the LiveCOMS review, agreed on some big organizational changes. So I’ll work on that this week and push it back to the authors. One of the big things is standardizing the discussion of various observables, so that each section reads in a similar way in terms of justification, tools, nuances of the measurement/simulation. So we’re still making progress on that doc. LD’Amore reached out to me asking about FF support for RNA oligonucleotides (following up from a slack back-and-forth in late 2021/early 2022). I’m comfortable leading this update and will run my drafts past the team in the ff-release meeting this week. JW – I’d e happy to give feedback on this at the ff-release call MT – It would be great if the endpoint of this discussion were public-facing - External people are constantly asking me about when we’ll support various things (both infrastructure and science), and so it’d be nice to have something to point people to so I don’t need to spend your time answering the same question over and over gain. JW – Also important to note the difference between proof of concept work/feasibility studies and “official FF releases”, where we may have a RNA/DNA FF soon, but it won’t necessarily “be an OpenFF release” for another year. CC – Optimistic that starting to talk about this now can motivate people to ask how they can contribute.
MT Pretty quiet - Lots of merges last week. Nothing too flashy but important progress toward biopolymer release. Support for 0.4 electrostatics tag - This means automatic upconversion of read FFs (so, reading and writing Sage to/from a file will convert electrostatics section from 0.3 to 0.4). Interchange will also restrict support to only handle the 0.4 electrostatics section, which should be fine since the OFF Toolkit will always upconvert to that. Dealing with some tech debt in Interchange, trying to bring in pydantic correctly (which is challenging). One problem, for example, is that molecules couldn’t roundtrip to JSON if they had conformers. This came up in Interchange testing even though it was a toolkit issue. I’m working with JMitchell on the openff-units docs. I’d previously made a quick README but this will convert them into proper docs. SB had rewritten the vsite code a few weeks ago, this PR went into the 0.10.X toolkit line. It took a while longer for these changes to get incorporated into the 0.11.X toolkit branch. This clears the way for yet another implementation of vsites in Interchange, which I’m optimistic about since there are defined tests and existing previous behavior to match to. So that’s my top priority this week. (Discussion about responsibility for informing adjacent developers about SMIRNOFF spec changes) DD – Probably best to start versioning the standards repo and have date-based releases, then tell people to subscribe to repo release notifications. MT – I’d be OK with the above, but I want to make sure that we don’t go through the cognitive stress of deciding on whether/how to talk about each spec release. So a single place or list of places where we broadcast releases without any decisionmaking would be helpful. JW – I’ll make a “release process” for standards in the repo wiki, which will contain instructions for how we update consumers on major changes.
DD JW – Made the pre-alpha biopolymer toolkit release Catching up on PR backlog (vsite port and 0.4 electrostatics, along with many others) Oracle cloud compute meeting This week, will be focusing on continuing to fight down technical debt, prepare for 0.11 release
|