2022-08-15 Core Developers meeting notes

General updates

• JW – Offline friday

• JW – Employees, remember to resign from Remote if Karmen/Darlamae told you to.

Individual updates

• MT

  • Chipping away at same problems as previous weeks. Reported vsite issue to PEastman. Seems like we’re doing stuff right, interested to hear what he says.

  • Cleaned up OpenMM export functionality - Everything had ended up in one gigantic file. Everything seems to be working except vsites

  • Working on updating OpenMMForceFields. Initially just went in to update for 0.11, but there are a few other things that stand out as problematic. Also their CI isn’t working so I’m having to fix that in the process.

  • Hoping to get to big releases this week.

    • JW – I don’t think the vsite thing is blocking to release - We’re matching the 0.10.6 behavior.

    • MT – Let’s discuss this during a working session.

• DD

  • Protein-ligand benchmarks

    1. merged gufe#39

    2. focused on making gufe DAG system use the same tokenization scheme as the rest of gufe in gufe#36

    3. need to get in gufe#42 after this

    4. performed some research cycles on distributed internals (serialization, communication protocol). Needed to ensure gufe DAGs could be cast into a form executable by a distributed scheduler, including tasks that simply wait for files from Folding@Home (without blocking)

  • This week, want to wrap up the above two PRs. IP and MH are going to use Perses to put together a noneq cycling protocol. Will probably iterate a bit on this. After that I’ll focus on storage, compute, and the execution interface.

  •  

• JW –

  • Something strange with TIP4P vsites. Worked with MT and spoke to PEastman. Maybe related to velocity initialization?

  • Met with AHogan to open , exceptions have to be assigned manually which can become time intensive so working on speedups in

  • Transitioned to JustWorks

  • Working on icode handling - RGowers identified this as necessary

  • Getting close to 0.11 release, just working with MT and JMitchell on final runtime issues and examples cleanup.

• CC

  • Finalized ELF10 library charges for amino acids

    • Compared ELF10 to Amber ff99 charges (2022-08-11 Protein FF meeting note )

    • Main conclusions

      • Backbone amide and carbonyl have larger dipole
        moment

      • Aliphatic carbons have smaller magnitudes

      • Groups with formal charges have larger magnitudes
        except arginine

  • Working with PB to define the “starting point” for the protein FF

  • Good progress on LiveCoMS review - still slogging through comments this week

• DN

  • Meeting with Katharina and Thomas on Friday

    • They clarified that BI's departure has nothing to do with the annual report. Additionally, they requested information regarding the time invested in each project.

  • There is quite a heated discussion about some of Espaloma's results. Katharina Meier appears to have taken in a student of John's, who has made some advances in Espaloma, and this has made her curious about the possibility of using Espaloma as the technology for the next generation of force fields. This discussion will be taken to the advisory board meeting.

  • And this week starts very lively with an important discussion about roadmaps.