2022-06-06 Interchange adoption Meeting notes

Owned by Jeffrey Wagner, created
Last updated: Jun 06, 2022
1 min read

 Date

Jun 6, 2022

 Participants

  • @Jeffrey Wagner

  • @Lily Wang

  • @Diego Nolasco (Deactivated)

  • @Matt Thompson

  • @Irfan Alibay

  • @Jeffry Setiadi

  • @Pavan Behara

  • @Michael Shirts

  • Repo:

  • Notes: https://docs.google.com/presentation/u/2/d/1s_sm_yjuoOPUnEjJlpvH52RICho0GgEEAI7F5YNry4M/edit?usp=sharing

    • Note – Proposed changes (p. 4) are already in motion / already done. JW approves them all

  • Slides from last update: https://docs.google.com/presentation/d/1L-awvi8TyM-oUepq187x4GbsjmiAxAsOqcFVlMGJG3Q/edit#slide=id.g120fb3f0578_0_534

  • Project plan: https://openforcefield.atlassian.net/l/c/XQciS95t

 Discussion topics

Item

Notes

Item

Notes

Scope: user adoption scope and regression testing

  • MT: regression testing environment ( ) uses OpenEye, not RDKit or AmberTools because the OE envs have been found to be more reproducible

  • JW: reminder that Interchange still does not support GBSA or custom ParameterHandlers, which are necessary for some fitting and science efforts

    • People who need these should continue staying on 0.10.x, which will continue being maintained until Interchange supports them

    • No more backdoor to old code paths in 0.11.x, to save on maintenance costs

  • MT: virtual sites not supported on OpenFF topologies now

  • MT: more standard implementation of FF settings, where PME is used for periodic, but not for non-periodic (see OFF-EP-0005)

  • VSite tests: JW approves Simon’s replacement tests, except:

    • Changing version from 0.10.2 to 0.10.6

  • JW: proposes energy tests to check corner cases in exclusions

    • MT: agrees

Interchange approver

  • Replacing Simon with Jeff, with no objections / unanimous agreement from all present

Status

  • MT: some single-molecule tests are failing, due to charge differences. Still investigating why

    • JW: how much energy should we spend on this?

    • LW: we should figure out why they’re different – if it’s runtime differences, should know if that’s the case

    • MT: planning to spend some time in the next couple days on this

  •  

Next steps

  • MT will work on charge comparison failures

  • JW will update project plan to say vsite tests are the 0.10.6 test suite + a single molecule energy comparison from that test suite

  •  

 Action items

 Decisions