2025-07-09 JAC/LW Check-In

2025-07-09 JAC/LW Check-In

Participants

  • @Jennifer Clark

  • @Lily Wang

Discussion topics

Notes

Notes

Zenodo Entries

The Industry Benchmark and MLPepper are the last ones to go

Industry Benchmark dataset: When a view was created with include_children=True, it overwrites exclude=[“trajectory”], resulting in a file size of 217GB. Setting include_children=False reduces to less that a GB.

On the openforcefield data page, is ready to go regardless , with JE’s approval I’ll merge once the other tickets are done

The Blog post is also updated and ready.

Genentech Project

Looked at the CIFs and it blatantly has the wrong charge for Fe hexacyanide. Since I’ve learned more I wanted to take another pass as TMOS to automatically detect the charge and oxidation state, this ended up being buggier than expected to still working on it.

I’m thinking I’ll set that aside and manually assess the other structures to finish my “report” and send to Richard this week.

Then I’d like to continue working on TMOS this week, because I don’t know if I should trust my CCD sorting if the charges are wrong…. I can reassess on Monday.

  • LW: plan sounds good (some discussion of actual issues and plan for tmos)

  • LW: wonder if these issues have affected the Psi4 calculations that have finished? Perhaps should look at them

  • JC: wonder if we should be focusing on Architector instead, where we can build our own complexes. I’ve been spending a lot of time getting our tools to work with existing databases, but it might be more productive to create a new dataset that follows the rules we set.

  • LW: where do the geometries come from in Architector?

  • JC: uses OpenBabel to put together ligand geometries, and contains a library of ligand-binding geometries. Then decides how to place them around the metal. Also is what the OMol25 dataset was generated from, which at least ran and finished. Architector is produced through LANL

  • LW: we should prioritise easy wins instead of rabbit trails. Your plan of looking at tmos for the rest of the sprint and then re-assessing sounds good.

  • JC: I reserve the right to hero merge a PR for TM Benchmark, if prudent

While Lily is Away

  • Set up FF Fitting Meeting

  • Check in with Julianne for the biopolymers meeting

Zenhub

Keep where it is:

  •  

Added Tickets:

  •  

Proposed Tickets:

  •  

Move to In Progress:

  •  

Close Tickets:

  •  

Action items

Decisions