open force field

An open and collaborative approach to better force fields

March 20, 2019 Advisory Board Meeting

Posted on 20 Mar 2019 by Arjun Narayanan and John Chodera

The Open Force Field Consortium Advisory Board met on 20 March, 2019. The minutes are summarized as follows:

Status Updates

  • OFFI NIH proposal submitted and available publicly
  • Hiring update:
    • Karmen Condic-Jurkic starts April 1 as half-time scientific coordinator/project manager
    • Interviews ongoing for bespoke torsion fitting and potentially electrostatics role

Open Force Field Toolkit update

  • Alpha release slated for April 2
  • Will contain RDKit backend support in abstracted cheminformatics structure (allowing future support for alternative cheminformatics packages to be easily added)
    • Supposed to give same results regardless of RDKit or OpenEye (tested with ~2k molecules from eMolecules to have same energies, forces, and force terms)
    • Differences in aromaticity models could result in some differences in corner cases
    • We are adding a “paranoid mode” so we can detect differences between toolkits that arise when building force fields
  • SDF with standardized tags will be the supported method for importing molecules with custom charges
    • OpenEye and RDKit on-board with standardized tags for charges
  • There may have been minor changes to XML forcefield format since the January preview release
    • Changes are summarized here
  • OFFI hope is to have weekly/biweekly bugfix or feature releases and conda package updates
  • Partners and users are directed to use the Github issue tracker to report issues or problems

Technical questions

  • How is Desmond currently used by Partners? What kind of Desmond support is needed?
    • Need for FEP+ compatibility will likely have to involve discussions with Schrodinger
    • Industry would ideally like to compare forcefields (including OPLS3/3e) and FEP protocols (FEP+, YANK, etc.)
    • Kaushik to work with PIs to provide some examples for testing
  • Desmond’s msys release allows dms files to be written directly, so could be used to construct system with SMIRNOFF parameters
    • Advantage of being independent of Maestro/Schrodinger

Reports on usage by partners since January meeting

  • Limited use so far - many people awaiting RDKit release to begin internal use
  • Training:
    • Plan for YouTube video for basics, followed by webex
    • Mobley to check with Jeff Wagner on training plan
    • Request for tutorial on using SMIRNOFF w/ YANK
    • Chodera to organize, probably need at least 1 month after April release to integrate SMIRNOFF support into YANK

Other stuff

  • Poster at ALCHEMICAL FREE ENERGY WORKSHOP 2019 in Göttingen. Thoughts are to highlight OFF Consortium from industry (user) perspective (D. Kuhn, Katharina)
  • Daniel and Ross to connect to work on this to coordinate with GRC poster to ensure complimentarity and common usage of some materials. Mobley to help by providing figures, etc.
  • GRC informal session/get-together
    • People still waiting to see who will be attending
    • 1:1 time with Jeff Wagner might be better use of time at the GRC than an organized session
  • Next Open Force Field Consortium face-to-face meeting scheduled for August 30-31, 2019 (Friday and Saturday) in San Diego
  • Group for evaluating protein-ligand binding free energies being assembled by John Chodera (see Slack) to curate additional benchmark sets