2020-02-20 Force Field Release meeting notes
Date
Feb 20, 2020
Participants
@Hyesu Jang
@Lee-Ping Wang
@Christopher Bayly
@David Mobley
@John Chodera
@Michael Gilson
@Jessica Maat (Deactivated)
@Simon Boothroyd
@Owen Madin
Goals
Update for the point release plan
Discussion topics
Time | Item | Presenter | Notes |
|---|---|---|---|
10 min | Point release plan | Hyesu Jang |
Action items
Meeting Note
1. Coverage set has not a good coverage on torsion SMIRKS
most of the uncovered SMIRKS which covers non-ring torsions don’t have exocyclic rotations;
Need to include new molecules outside ‘Roche set’ and ‘Coverage set' -> which shows the need of redesigning the dataset we are using for the fitting;
NCI 250k dataset (and other datasets from qca-dataset-submission repo) was suggested to explore (@John Chodera )
2. Point release plan
The point release will take care of (1) tetrazole-like chemistry by including new torsions and new improper torsions (2) N-N rotations ;
The point release will be ready in a week;
@Hyesu Jang @Jeffrey Wagner will work together for the point release
3. Improper torsions and torsions for conjugated system and the impact of chemical environment
are we fitting improper in the re-fitting? - since we don’t have improper torsion targets in the fitting, the change in parameter during the re-fitting will have a negligible impact;
Since the consideration of the chemical environment is required to accurately describe conjugated systems, simply using the existing force field terms defined by local environment can’t be successful. The purpose of the point release is more for a specific fix not for a global fix. More thorough consideration on conjugated system will be able to be achieved once the WBO implementation is done.
4. Discussion about the inclusion of In-ring torsions in the fitting (description of ring puckeing)
There was a discussion about inclusion of in-ring torsions. For describing chair flips, need to think about the proper way to drive the torsions to induce a chair flip. One possible approach will be multidimensional torsion scan (@Lee-Ping Wang )