OpenMMSolutionGen().get_input_set(...) ValueError

[xperrylinn]

Hey @orionarcher,

I'm running the code below and where I get a ValueError from within OpenFF stating there aren't any registered toolkits can provide the capability "assign_partial_charges":

>>> from pymatgen.io.openmm.schema import InputMoleculeSpec
>>> from pymatgen.io.openmm.generators import OpenMMSolutionGen
>>> 
>>> water_input_mol_spec = InputMoleculeSpec(name="water", smile="O", count=1)
>>> 
>>> input_set = OpenMMSolutionGen().get_input_set(input_mol_dicts=[water_input_mol_spec])
/Users/xperrylinn/miniconda3/envs/atomate2-openmm/bin/wrapped_progs/antechamber: Fatal Error!
Cannot properly run "/Users/xperrylinn/miniconda3/envs/atomate2-openmm/bin/sqm -O -i sqm.in -o sqm.out".
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Users/xperrylinn/miniconda3/envs/atomate2-openmm/lib/python3.8/site-packages/pymatgen/io/openmm/generators.py", line 196, in get_input_set
    openff_mol.assign_partial_charges(self.default_charge_method)
  File "/Users/xperrylinn/miniconda3/envs/atomate2-openmm/lib/python3.8/site-packages/openff/toolkit/topology/molecule.py", line 2503, in assign_partial_charges
    toolkit_registry.call(
  File "/Users/xperrylinn/miniconda3/envs/atomate2-openmm/lib/python3.8/site-packages/openff/toolkit/utils/toolkit_registry.py", line 370, in call
    raise ValueError(msg)
ValueError: No registered toolkits can provide the capability "assign_partial_charges" for args "()" and kwargs "{'molecule': Molecule with name '' and SMILES '[H][O][H]', 'partial_charge_method': 'am1bcc', 'use_conformers': None, 'strict_n_conformers': False, 'normalize_partial_charges': True, '_cls': <class 'openff.toolkit.topology.molecule.Molecule'>}"
Available toolkits are: [ToolkitWrapper around The RDKit version 2022.09.5, ToolkitWrapper around AmberTools version 22.0, ToolkitWrapper around Built-in Toolkit version None]
 ToolkitWrapper around The RDKit version 2022.09.5 <class 'openff.toolkit.utils.exceptions.ChargeMethodUnavailableError'> : partial_charge_method 'am1bcc' is not available from RDKitToolkitWrapper. Available charge methods are ['mmff94', 'gasteiger'] 
 ToolkitWrapper around AmberTools version 22.0 <class 'subprocess.CalledProcessError'> : Command '['antechamber', '-i', 'molecule.sdf', '-fi', 'sdf', '-o', 'charged.mol2', '-fo', 'mol2', '-pf', 'yes', '-dr', 'n', '-c', 'bcc', '-nc', '0.0']' returned non-zero exit status 1.
 ToolkitWrapper around Built-in Toolkit version None <class 'openff.toolkit.utils.exceptions.ChargeMethodUnavailableError'> : Partial charge method "am1bcc"" is not supported by the Built-in toolkit. Available charge methods are ['zeros', 'formal_charge']

I poked through the files referenced in the traceback and it seems like I need to have one of the supporting dependencies installed for one of the toolkit wrappers, so I conda installed rdkit. I then realized the OpenMMSolutionGen.default_charge_method default parameter is not compatible with all possible tool toolkit wrappers. I updated the instantiation to look like OpenMMSolutionGen(default_charge_method="mmff94") and this seems to have unsnagged this bit. Is this expected?

Following that snag, I'm now snagging on this:

Traceback (most recent call last):
  File "/Users/xperrylinn/miniconda3/envs/atomate2-openmm/lib/python3.8/site-packages/jobflow/managers/local.py", line 98, in _run_job
    response = job.run(store=store)
  File "/Users/xperrylinn/miniconda3/envs/atomate2-openmm/lib/python3.8/site-packages/jobflow/core/job.py", line 544, in run
    response = function(*self.function_args, **self.function_kwargs)
  File "/Users/xperrylinn/Documents/Academics/MSSE/spring_2023/Capstone/atomate2-openmm/src/atomate2/openmm/jobs/core.py", line 63, in make
    input_set = OpenMMSolutionGen(default_charge_method="mmff94").get_input_set(
  File "/Users/xperrylinn/miniconda3/envs/atomate2-openmm/lib/python3.8/site-packages/pymatgen/io/openmm/generators.py", line 222, in get_input_set
    coordinates = get_coordinates(
  File "/Users/xperrylinn/miniconda3/envs/atomate2-openmm/lib/python3.8/site-packages/pymatgen/io/openmm/utils.py", line 202, in get_coordinates
    pw = PackmolBoxGen(seed=random_seed).get_input_set(
  File "/Users/xperrylinn/miniconda3/envs/atomate2-openmm/lib/python3.8/site-packages/pymatgen/io/packmol.py", line 225, in get_input_set
    return PackmolSet(
TypeError: Can't instantiate abstract class PackmolSet with abstract methods from_directory

How can I understand what's happening here?

[orionarcher]

I am quite confused by this. PackmolSet is not an abstract class nor is it being instantiated with from_directory.

Are you getting the tests to pass locally or are they currently failing?

I ran the same code and ran into two errors, though not the one you described. First, the density argument was missing in get_input_set and second the box was too small, I believe it has to be around 15 Angstrom to be large enough to be instantiated by OpenMM. I made these modifications and it started working for me

    from pymatgen.io.openmm.schema import InputMoleculeSpec
    from pymatgen.io.openmm.generators import OpenMMSolutionGen

    water_input_mol_spec = InputMoleculeSpec(name="water", smile="O", count=100)

    input_set = OpenMMSolutionGen().get_input_set(input_mol_dicts=[water_input_mol_spec], density=0.2)

[xperrylinn]

The tests are failing for me. I created a new Python 3.8 conda environment dedicated for the pymatgen-io-openmm repo and after working through installing the required dependencies, I ran the tests and they are failing: 28 failed, 28 passed, 24 warnings, 2 errors in 25.02s. I've sampled some of the failures below. I've attached my conda environment YAML export (as a .txt file because GitHub doesn't allow you to upload YAML). This feels like an issue related to setup.
environment.txt

E       TypeError: get_disconnected_fragments() got an unexpected keyword argument 'return_index_map'

E       TypeError: Can't instantiate abstract class PackmolSet with abstract methods from_directory

E       ValueError: No registered toolkits can provide the capability "assign_partial_charges" for args "()" and kwargs "{'molecule': Molecule with name '' and SMILES '[H][O][H]', 'partial_charge_method': 'am1bcc', 'use_conformers': None, 'strict_n_conformers': False, 'normalize_partial_charges': True, '_cls': <class 'openff.toolkit.topology.molecule.Molecule'>}"
E       Available toolkits are: [ToolkitWrapper around The RDKit version 2022.09.5, ToolkitWrapper around AmberTools version 22.0, ToolkitWrapper around Built-in Toolkit version None]
E        ToolkitWrapper around The RDKit version 2022.09.5 <class 'openff.toolkit.utils.exceptions.ChargeMethodUnavailableError'> : partial_charge_method 'am1bcc' is not available from RDKitToolkitWrapper. Available charge methods are ['gasteiger', 'mmff94'] 
E        ToolkitWrapper around AmberTools version 22.0 <class 'subprocess.CalledProcessError'> : Command '['antechamber', '-i', 'molecule.sdf', '-fi', 'sdf', '-o', 'charged.mol2', '-fo', 'mol2', '-pf', 'yes', '-dr', 'n', '-c', 'bcc', '-nc', '0.0']' returned non-zero exit status 1.
E        ToolkitWrapper around Built-in Toolkit version None <class 'openff.toolkit.utils.exceptions.ChargeMethodUnavailableError'> : Partial charge method "am1bcc"" is not supported by the Built-in toolkit. Available charge methods are ['zeros', 'formal_charge']

[xperrylinn]

@orionarcher, steps for creating the Python 3.9 environment that is failing at running pytest:

1. conda create -n pymatgen-io-openmm python=3.9
2. conda activate pymatgen-io-openmm
3. pip install pymatgen
4. conda install -c conda-forge openmm parmed numpy pytest openbabel rdkit pydantic openff-toolkit packmol openmmforcefields
5. conda install -c conda-forge MDAnalysis
6. pytest .