openff-recharge icon indicating copy to clipboard operation
openff-recharge copied to clipboard

Published 20 hours ago • verified publisher icon openforcefield

Re-jig BCCs to be more specific

Open lilyminium opened this issue 2 years ago • 2 comments
trafficstars

Description

Re-do OpenEye BCC collection

Status

  • [ ] Ready to go

lilyminium avatar Apr 17 '23 06:04 lilyminium

Codecov Report

Patch coverage: 100.00% and project coverage change: -0.02 :warning:

Comparison is base (a97eb10) 84.06% compared to head (f1c0957) 84.05%.

Additional details and impacted files 
@@            Coverage Diff             @@
##             main     #131      +/-   ##
==========================================
- Coverage   84.06%   84.05%   -0.02%     
==========================================
  Files          37       37              
  Lines        2021     2019       -2     
==========================================
- Hits         1699     1697       -2     
  Misses        322      322
Impacted Files Coverage Δ
openff/recharge/charges/bcc.py 94.26% <100.00%> (-0.10%) :arrow_down:

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Do you have feedback about the report comment? Let us know in this issue.

codecov[bot] avatar Apr 17 '23 06:04 codecov[bot]

The "correctness" is how well it corresponds to how OpenEye assigns things. There's a benchmark in scripts/validation, although it's too expensive to run in CI. The new parameters here do somewhat better than the original (and include phosphorus), but not perfectly. I'll see if I can a) improve the smirks a bit more and b) incorporate some of it into CI.

original-am1-bcc.json is what used to be there, it probably should be moved into a legacy file.

If I'm following the paper trail, scripts/convert-am1-bcc/generate_original_am1bcc.py is to be compared to the file now at scripts/convert-am1-bcc/legacy/convert.py? GitHub's UI doesn't pick up on this but I can just diff them locally.

That's right :)

There are some commented-out patterns in GENERAL_ATOM_CODES - intentional or temporary?

These were intentionally commented out to leave a bit of a trail in case the more general patterns should be reverted back to the stricter ones... thanks to your feedback I've realised this PR as a whole is probably still a bit premature.

lilyminium avatar Jul 04 '23 08:07 lilyminium