All template generators now take template_generator_kwargs as an optional init parameter. This removes the residue_atoms argument from the generate_residue_template method, see #391 for details. This release updates the CHARMM force field to the July 2024 release. We also updated Amber force fields with latest versions from AmberTools 24. Compatibility with esploma force fields has also been improved.

See more details below 👇

What's Changed

• Remove duplicate atom type handling and re-enable GAFF tests by @epretti in #371
• Update CHARMM force field to July 2024 release by @epretti in #356
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci[bot] in #374
• Improve molecule filtering in tests by @mattwthompson in #370
• set weights_only=False to ensure compatibility with torch>2.6 by @NiklasTR in #375
• Update Amber force fields with latest versions from AmberTools 24 by @epretti in #368
• Pin numpy <2.3 to fix issue with numpy.compat missing by @mikemhenry in #385
• Fixes and improvements for CHARMM force fields by @epretti in #386
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci[bot] in #388
• Update README to reflect default SMIRNOFF force field by @mattwthompson in #389
• adjust ci matrix by @mikemhenry in #394
• run espaloma tests by @mikemhenry in #392
• Fix espaloma model download race conditions by @epretti in #398
• Remove DummySystemGenerator by @epretti in #397
• Fixes for issues and assorted cleanups to generators by @epretti in #391
• Import DGL to avoid race condition writing configuration file by @epretti in #400
• Release 0.15.0 by @mikemhenry in #401

New Contributors

• @epretti made their first contribution in #371
• @NiklasTR made their first contribution in #375

Full Changelog: 0.14.2...0.15.0

What's new?

This release enables compatibility with Interchange version 0.4.2 by better processing CMMotionRemover if present in one of the Systems that is used internally to generate residue templates. The behavior of the removeCMMotion argument to OpenMM's ForceField.createSystem if containing these residue templates is unchanged.

See this pull request and issue for more technical details:
#367
#365

Breaking API Changes?

None!

List of Pull Requests in Release

• run latest version of openmm in CI by @mikemhenry in #345
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in #349
• Bump mamba-org/setup-micromamba from 1 to 2 by @dependabot in #348
• Bump codecov/codecov-action from 4 to 5 by @dependabot in #353
• Downpin away from openff-interchange 0.4.2 in CI by @j-wags in #366
• Test new ambertools 24 build by @mikemhenry in #361
• Remove CMMRemover (if present) in SMIRNOFFTemplateGenerator by @j-wags in #367

Full Changelog: 0.14.1...0.14.2

0.14.1 Bring back GAFFTemplateGenerator for OpenMM >=7.6.0

This release brings back GAFF force feild support for all versions of OpenMM previously supported.
Additionally, we now use the output of parmchk2 for all GAFF parameters.
Previously we used gaff.dat + parmchk2 output to generate forcefield parameters.
Functionally this doesn't change the end user experience but means we do not need to create new forcefield XML files for newer GAFF versions and now support whatever GAFF versions that parmchk2 supports for the installed AmberTools version.

Deprecation Notice

The XML files in openmmforcefields/ffxml/amber/gaff/ffxml may be removed in a future release.

GAFF now works with OpenMM 8.1.2

A release will follow soon that enables GAFF support on older OpenMM releases.

What's Changed

• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in #340
• re-enable gaff by @mikemhenry in #334

New Contributors

• @pre-commit-ci made their first contribution in #340

Full Changelog: 0.13.0...0.14.0

0.13.0 Temporarily remove GAFFTemplateGenerator

This release temporarily removes GAFFTemplateGenerator because of packaging incompatibilities with
AmberTools 23. This functionality is planned to be re-introduced in 0.14.0.

This release is expected to work with Python 3.10-3.12.

Other changes include

• The default force field of SystemGenerator was updated from openff-1.0.0 (code name Parsley) to
  openff-2.0.0 (code name Sage).

What's Changed

• Merge dependabot changes by @mattwthompson in #307
• Update SMIRNOFF/OpenFF docs/examples by @mattwthompson in #305
• fix badges + readme updates by @mikemhenry in #309
• feat: Improve handling of NB 1-4 scaling factors by @LilDojd in #318
• Update partial charge assignment docs by @mattwthompson in #313
• Specify major version of checkout action by @mattwthompson in #314
• Do not generate conformers that will not be used by @mattwthompson in #315
• Temporarily remove GAFF because of parmchk2 issues by @mattwthompson in #329
• Add correct cutoffs to SMIRNOFF example by @mattwthompson in #324
• Bump codecov/codecov-action from 3 to 4 by @dependabot in #322
• Add uniform, automated linting by @mattwthompson in #319
• Remove code paths that use old OpenFF Toolkit API by @mattwthompson in #323
• linted all files by @mikemhenry in #332
• make a new excpetion for dropping gaff support by @mikemhenry in #333

New Contributors

• @LilDojd made their first contribution in #318

Full Changelog: 0.12.0...0.13.0

What's Changed

Improvements

• Convert all amber ions (see openmm/openmm#3663 for more detail) one-to-one using Amber ion frcmod files (from AmberTools 22) into OpenMM xml files with the same names and the same contents. by @mattwthompson & @aizvorski in #242 & #239
• SystemGenerator should only add barostat if system is periodic. Solves #252 by @jchodera in #253
• Fix OPC3 bond length. In the Amber implementation of the OPC3 water model at ffxml/amber/opc3.xml, the H-O bond length was too large by a factor of 10. by @mattwthompson in #273
• raise ForceException if a custom force is found. by @mattwthompson in #292
• Allow an offxml string to be used as a small molecule force field for the SystemGenerator and SMIRNOFFTemplateGenerator for example:

"Try and load a OFFXML string into a system generator object"
from openmmforcefields.generators import SystemGenerator
from openff.toolkit.typing.engines.smirnoff import ForceField
from openff.toolkit.topology import Molecule
import openmm

# load the ff
ff = ForceField("openff-2.0.0.offxml")
# create a system generator
system_gen = SystemGenerator(
    forcefields=['amber/ff14SB.xml', 'amber/tip3p_standard.xml'],
    small_molecule_forcefield=ff.to_string()
)
mol = Molecule.from_smiles("CC")
system = system_gen.create_system(topology=mol.to_topology().to_openmm(), molecules=mol)
with open("system.xml", "w") as output:
    output.write(openmm.XmlSerializer.serialize(system))

by @jthorton in #288

• Support system and template generator for Espaloma 0.3.* (espaloma 0.2.* is no longer supported) by @ijpulidos in #293

CI

• Bump codecov/codecov-action from 3.1.0 to 3.1.1 by @dependabot in #238
• Switch to setup-micromamba by @mattwthompson in #274
• Scattered fixes to get CI running by @mattwthompson in #290

Docs

• Update docstrings, run in CI by @mattwthompson in #240

New Contributors

• @jthorton made their first contribution in #288

Full Changelog: 0.11.2...0.12.0

Quick Look

@mikemhenry bungled the support of the new OpenFF Toolkit. This release fixes that.

What's Changed

• add back in openeye toolkit by @mikemhenry in #226
• Tweak CI: Switch to Micromamba, pin to supported OpenFF Toolkit versions by @mattwthompson in #232
• Bump codecov/codecov-action from 2.1.0 to 3.1.0 by @dependabot in #197
• Bump actions/checkout from 2 to 3.0.2 by @dependabot in #196
• Fix toolkit tests by @mikemhenry in #227
• make sure we don't install the newest toolkit by mistake by @mikemhenry in #237

Full Changelog: 0.11.1...0.11.2

Quick Look

This bug fix release mostly fixes regressions when introduced during development of 0.11.1, hence many of these bugs never made it into a release.
The main highlight is we now support the new openff-toolkit 0.11 and have backwards compatible support for older versions.
See changelog below for more details.

What's Changed

• Remove espaloma debug line by @mikemhenry in #201
• Remove debug prints. by @ijpulidos in #198
• Fix CI by @mikemhenry in #202
• Fix accidental elimination of torsions after 0.11.0 release by @jchodera in #208
• Fix #210: gaff atom type name is now correctly passed by @jchodera in #212
• Fix #211 : Use basename instead of full path for cache database table names by @jchodera in #213
• Update for upcoming OpenFF Toolkit API breaks by @mattwthompson in #191
• Remove oe license check by @mikemhenry in #220
• Fix #214: Leave temp dir before raising exceptions by @rosemary-cresset in #218
• Fix #216: Remove % sign from parmchk2 command by @rosemary-cresset in #217
• Fix #215: Copy gaff dat file to temp dir by @rosemary-cresset in #219

New Contributors

• @ijpulidos made their first contribution in #198
• @mattwthompson made their first contribution in #191
• @rosemary-cresset made their first contribution in #218

Full Changelog: 0.11.0...0.11.1

This release adds experimental support for generating small molecule parameters with the graph convolutional model espaloma, which must be installed separately to use this capability. This feature is documented in the README.

Bugfixes:

• Some fixes to GLYCAM support (#180)

• Update imports and tests for OpenMM 7.6 #173
• Fix tests to test GLYCAM and use new way of importing for openmm #177
• Enable GLYCAM06j-1 forcefield conversion #156
• Bump codecov/codecov-action from 1 to 2.0.3 #171
• fix dependabot so it can see our secrets #167
• Testing CI #155