Begin setting up torsion automation #13
Conversation
|
/run-torsions torsion-dev/tmp.yaml |
1 similar comment
|
/run-torsions torsion-dev/tmp.yaml |
|
/run-torsions torsion-dev/tmp.yaml |
|
/run-torsions torsion-dev/tmp.yaml |
|
/run-torsions torsion-dev/tmp.yaml |
|
/run-torsions torsion-dev/larger/tmp.yaml devtools/torsions.yaml |
|
A workflow dispatched to run torsion benchmarks for this PR has just finished.
|
|
A workflow has been dispatched to run the benchmarks for this PR.
|
|
A workflow dispatched to run torsion benchmarks for this PR has just finished.
|
|
/run-torsions torsion-dev/larger/tmp.yaml devtools/torsions.yaml |
|
/run-torsions torsion-dev/larger/tmp.yaml devtools/torsions.yaml |
|
A workflow has been dispatched to run the benchmarks for this PR.
|
|
A workflow dispatched to run torsion benchmarks for this PR has just finished.
|
|
/run-torsions torsion-dev/larger/tmp.yaml devtools/torsions.yaml |
1 similar comment
|
/run-torsions torsion-dev/larger/tmp.yaml devtools/torsions.yaml |
|
A workflow has been dispatched to run the benchmarks for this PR.
|
|
A workflow dispatched to run torsion benchmarks for this PR has just finished.
|
|
/run-torsions torsion-dev/larger/tmp.yaml devtools/torsions.yaml |
|
I don't think we took notes on this, so - last week, @ntBre and I discovered that some data points (maybe 5-10%) were reporting both "no qm points" and "no mm points" which was a surprise because the molecules themselves were well-formed and processed into inputs. The problem was that YAMMBS looked up IDs by (mapped) SMILES alone, which works fine for molecules but doesn't work for torsion drives. Because a torsion drive can drive a given mapped SMILES at different atom quartets, the mapped SMILES alone cannot uniquely identify an ID and the lookup needs to also pass the dihedral indices through. See openforcefield/yammbs@b7df922 |
|
A workflow has been dispatched to run the benchmarks for this PR.
|
|
A workflow dispatched to run torsion benchmarks for this PR has just finished.
|
|
@ntBre could you give this a once-over (or a thorough review if you wish) and merge? The only thing I'm not sure about is what the current |
|
Sure, I'll take a look first thing tomorrow. I actually deleted my |
| if (comment.body.startsWith('/run-optimization-benchmarks')) { | ||
| benchmark_workflow = "opt.yaml" | ||
| } else if (comment.body.startsWith('/run-torsion-benchmarks')) { |
There was a problem hiding this comment.
The switch from run-benchmark to run-optimization-benchmarks and the new run-torsion-benchmarks could be added to the main README, but it's fine if that's not included in this PR.
| git commit -m "Add benchmark results" | ||
| git push |
There was a problem hiding this comment.
This isn't strictly related to the PR, but I realized last night that kicking off an optimization benchmark and a torsion benchmark at the same time in the same PR would probably cause one of them to fail at git push. One (or both) of them probably need a git pull --rebase before the push. Or that workflow could just not be supported.
There was a problem hiding this comment.
Good call, I've opened a separate issue because I don't want to fix that right now
| @@ -22,4 +22,4 @@ dependencies: | |||
| - tqdm | |||
|
|
|||
| - pip: | |||
| - git+https://github.com/openforcefield/yammbs.git@0.1.1 | |||
| - git+https://github.com/openforcefield/yammbs.git@add-torsion-models | |||
There was a problem hiding this comment.
Do you want to update this to the latest yammbs release before merging?
There was a problem hiding this comment.
And is this env still necessary?
|
Thanks! |
Development checklist
Submission Checklist
submissionsdirectory containing the YDS input file and optionally a force field.offxmlfile/run-benchmark path/to/submissionJob status: successto be posted