0.10.x merge into main #1235
0.10.x merge into main #1235
Conversation
* run CI on new 0.10.x branch * Update .github/workflows/CI.yaml Co-authored-by: Iván Pulido <2949729+ijpulidos@users.noreply.github.com> --------- Co-authored-by: Iván Pulido <2949729+ijpulidos@users.noreply.github.com>
* update comments * bump ci --------- Co-authored-by: Iván Pulido <2949729+ijpulidos@users.noreply.github.com> Co-authored-by: Mike Henry <11765982+mikemhenry@users.noreply.github.com>
* use python executable from env * run tests using the shell * parmed 4 seems to be giving us some issues * actually I think we want the newer parmed * pin pymbar for now * ooof, parmed != pymbar * add some more debugging to figure out how we are getting old parmed * Update devtools/conda-envs/test_env.yaml * see if shell=True fixes it * removed parmed by mistake
* switch to using pymbar 3 & 4 support from openmmtools * fix typo on import * fix yaml * switch to using pymbar 4 * missed a pymbar import * missed another one * bump ci * missed a import * go back to how it was * Update devtools/conda-envs/test_env.yaml --------- Co-authored-by: Iván Pulido <2949729+ijpulidos@users.noreply.github.com>
…ipeline (#1210) * Initial and final topologies serialized per phase. * Using properties instead of private attrs. * Fix test * Remove uneeded code/attributes * bump ci * Store phase topologies separately * Fix tests. vacuum topologies expected. * Better docstring. --------- Co-authored-by: Mike Henry <11765982+mikemhenry@users.noreply.github.com>
* Added note about example for adding oe license * make docker file much simpler * build images in CI * forgot to add conda-forge * fix permissions on a step * get oe_license file mounted in docker container * mount path must be absolute * setup singulairty to test + fix testing on docker image * add some testing deps to the image * add -v for tests, fix envar * tests are failing, want to test rest of pipeline * use latest openmm version * hardcode perses version for now * bump ci * make sure we can make openeye dir * support a dev build as well * don't hardcode value * fix name clash * forgot to add conda-forge * bump ci * test docker image and fix missing deps * install ps * also push latest tag * don't build on tag since the conda-forge package won't exist yet * don't test the examples * Remove docker deb build, we can do these ourselves * better document container use * build a latest version for apptainer * build with 11.2 to make things more compatable * skip docker test to see if other bits build okay * see if I can get the step to fail if there is an error * skip docker tests to make sure apptainer builds okay * Add mpiplus and mpi4py to docker image * give correct path to oe license * add mpi stuff to docker * clean up diskspace before build * skip tests for singularity now that the only failures come from a package bug
* Improving examples dir structure and readme * Adding Tyk2 CLI example * removing new-cli/ripk2 example (deprecated) * fix typo for link * Clarify tyk2 cli example docs
* CI miscellaneous fixes (#1217) * CI minor fixes. Allow codecov to fail. * bump ci --------- Co-authored-by: Mike Henry <11765982+mikemhenry@users.noreply.github.com> * Changing offline freq default to checkpoint interval * Fixing input yaml for example * commenting offline-freq param --------- Co-authored-by: Mike Henry <11765982+mikemhenry@users.noreply.github.com>
* peptide mutation MPI example added * better documentation of motivation
* set cutoff distance in sterics_custom_nonbonded_force * matching cutoff for custom forces. Improving logic. * test for HTF nonbonded cutoff --------- Co-authored-by: Iván Pulido <2949729+ijpulidos@users.noreply.github.com>
* CI miscellaneous fixes (#1217) * CI minor fixes. Allow codecov to fail. * bump ci --------- Co-authored-by: Mike Henry <11765982+mikemhenry@users.noreply.github.com> * added dels to contexts * Update perses/app/setup_relative_calculation.py --------- Co-authored-by: Iván Pulido <2949729+ijpulidos@users.noreply.github.com>
* fix spectator support. Enabling test. * Test to run on GPU CI.
|
Getting some interesting errors from pymbar https://github.com/choderalab/perses/actions/runs/6160980084/job/16719308140#step:9:818 I had thought we fixed that in openmmtools, it is pulling in the latest version |
|
@mikemhenry it should be fixed, it should catch the attribute error and handle that in https://github.com/choderalab/openmmtools/blob/main/openmmtools/multistate/multistateanalyzer.py#L1924 |
|
Looks like the real error was https://github.com/choderalab/perses/actions/runs/6160980084/job/16719308140#step:9:977 |
|
It seems like we do need to use pymbar 3.1.1 for GPU tests to pass |
|
We can keep that in mind for the conda-forge recipe, is this good now to merge in @ijpulidos ? |
| - pdbfixer | ||
| - pip | ||
| - progressbar2 | ||
| - pymbar <4 | ||
| - pymbar |
There was a problem hiding this comment.
Do we want to pin the pymbar version here as well or only in the conda-forge recipe?
There was a problem hiding this comment.
I rather do it in the conda-forge recipe since if we don't have a pin in the yaml we use for tests, it is really easy then in CI to make a matrix testing different versions or like the way we do it now, since there is no pin, cpu tests get pymbar 4, but in our GPU CI yaml, we use the extra build args to then pin pymbar<4. If we put it in the env yaml, we would get an error because micomamba doesn't know what to do when the yaml says >3 but the cli command says <4
Description
Motivation and context
Resolves #???
How has this been tested?
Change log