(JW and MT discussed, JW commented on places in our code that should specify use_interchange=False)
MT – I need test molecules and FF for vsites
JW – This should be pretty straightforward, except that I need to come up with cases that exercise the match keyword values, wildcards in SMIRKS, and parameter collisions with or without matching names, and those are super confusing.
MT – This is super confusing. It would be good to loop in SB on the design of these tests.
Show res info defined/undefined to/from OMM/RDKit/OE spec
Planning to add somewhere in docs, along with other core props <https://github.com/openforcefield/openff-toolkit/issues/135>
Need eyes on anything?/PR clearance
PR clearance/issue discussion
MT – Found an issue with elements to masses
(Discussed in
, ultimately decided to have our own dict of atomic number to mass and symbol, which will live in openff-units)
MT – Finding charge (and other thing) differences in regression suite
JW – Possibilities are:
Changes in use of normalize_partial_charges kwarg - This doesn’t seem to be the case, and frequently the differences are too large to be explained by this
Changes in the topology refactor branch - We could test this by running create_openmm_system in the refactor branch with use_interchange=True/False
MT – I’ll start here tomorrow
Omega generating different conformers? Could test by running the same set twice with the same program version.
Some change in how aromaticity is perceived/assigned/preserved? Not sure how to test this, or what effect it’d have.
Add Comment