JW – I need to leave for governing board meeting at the hour. Alternate host?
MT – Will do
Individual updates
JW –
Toolkit RC releases (tags) and worked with MT on building conda packages. See #general and #developers for RC instructions. Several small polishes/bugfixes queued up for coming days.
Working on annual report
Annual meeting
DNA QC dataset coming in from chodera lab - Looks like we’re finishing up SPICE and we have the bandwidth for it. Thanks PB for responding promptly!
Need to make a call about whether to try and replicate DHahn’s pmx using F@H infrastructure.
DD – Wouldn’t it be better to get results based on non-expert inputs?
JW – Could you join lead team meeting today? LW, DN, and I will tackle that first in “hard decisions”
DN –
Annual report - Working on slide deck. KCJ has tasked me with writing a 4-5 page text report in addition to the slides. This is nice because it lets me structure my knowledge of OpenFF.
Working on improving communication with stakeholders, made a mistake on Friday about announcing release candidate.
JW – No worries, it’s a subtle engineering-language thing. The best we can do is keep learning.
KMeier has been pushing for more info on finances, which we hadn’t traditionally had great control over (Virginia Tech)/been clear about communicating. This is an important part of my job.
JW – I’m very excited to have some more attention on our current+future finances. Things were really chaotic before and it led to low-quality decisionmaking at high levels.
DD – Yeah, the PI/grant money mindset is different from what we need.
We need to communicate progress better - We make more sense as a program, with projects under it - Toolkit Rosemary, Interchange, F@H interface, etc.
MT
Made release candidates for toolkits and interchange. Went pretty smoothly. In hindsight it could have been done a bit quicker, but we avoided the bad “breaking things in production” outcomes. We can now make new RCs really easily and push them to conda-forge. Excited to receive more user feedback - especially external.
Refactored vsite code. I’ve been able to write to OpenMM and GROMACS in ways that get accurate single point energies, but will crash after a few timesteps of simulation. This needs to be fixed - I’d like to recruit someone with OpenMM experience to work with me on this next week. But right now this is something that ParmEd can’t really handle so Interchange will be a first in this field.
Working on SciPy talk, going to be working on this for the next two days. I only have 30 minutes so I won’t be giving a complete overview of the project - I’ll talk about a bit of the initiative and then give links for the rest.
JW – Let me know if you’d like advice/review on the slides.
DD
Protein-ligand benchmark
Reworked DAG system to be more dask-like; putting PR up today in gufe
this will make protocol developers' lives easier
performed research cycles on storage questions, serialization, hashing
aiming to put together coherent approach from working notes this week
relevant but somewhat distinct from OpenFE's needs
QCArchive
SPICE sets complete
need to update OpenFF profile for PRP, including publications that PRP compute went to
PB
Mostly fitting related work. While working on DEXP fits found some API changes slowing down the fits.
ELF10 charges being the default in toolkit since 0.10.4 optgeo target batch sizes should be small.
Also, an issue with angle parameter (a40) going out of (0, pi), which is systemic and similar to sulfonamide issue before and openmm runs are failing during optimization as it doesn't accept values out of this range anymore (since 7.7). I tried using the MSM starting point refit I had, but some targets with MD energy convergence issues on the first step itself.
Creating FB inputs for Jessica’s torsion multiplicity work with the new datasets.
A user submitted a DNA dataset to QCA-d-s with some validation issues, will work with DD on resolving