A problem with the optimization_schemas_from_molecules function

[ognjenperisic]

Hi everybody,

I am having a problem generating an optimization schema from the list of molecules.

Here are the key elements of the code:

1. Loading a set of molecules:
   local_file_name = 'ligands.sdf'
filename = directory + set_name + '/' + local_file_name
molecules = Molecule.from_file(filename)

2. generating a schema:
   workflow_schema = factory.optimization_schemas_from_molecules(molecules)

3. Calling the executor:
   from openff.bespokefit.executor import BespokeExecutor, BespokeWorkerConfig, wait_until_complete
with BespokeExecutor(
n_fragmenter_workers = 1,
n_optimizer_workers = 1,
n_qc_compute_workers = 2,
qc_compute_worker_config=BespokeWorkerConfig(n_cores=1)
) as executor:
# Submit our workflow to the executor
task_id = executor.submit(input_schema=workflow_schema)
# Wait until the executor is done
output = wait_until_complete(task_id)

This is what I get:

_Deduplication                : 100%|██████████| 24/24 [00:00<00:00, 462.61it/s]
_Building Fitting Schema: 100%|██████████████████| 24/24 [00:05<00:00,  4.08it/s]

worker: Warm shutdown (MainProcess)

worker: Warm shutdown (MainProcess)

worker: Warm shutdown (MainProcess)
Traceback (most recent call last):
  File "test_fep_benchmark_set_1.py", line 83, in <module>
    task_id = executor.submit(input_schema=workflow_schema)
  File "...../site-packages/openff/bespokefit/executor/executor.py", line 372, in submit
    data=CoordinatorPOSTBody(input_schema=input_schema).json(),
  File "pydantic/main.py", line 342, in pydantic.main.BaseModel.__init__
pydantic.error_wrappers.ValidationError: 1 validation error for CoordinatorPOSTBody

input_schema
  value is not a valid dict (type=type_error.dict)__

However,
when I replace step 2. with
workflow_schema = factory.optimization_schema_from_molecule(molecules[0])
(processing just one molecule) the code runs with no issues.

My environment:
openff-bespokefit 0.1.3 pyhd8ed1ab_0 conda-forge
openff-forcefields 2.0.0 pyh6c4a22f_0 conda-forge
openff-fragmenter-base 0.1.2 pyhd8ed1ab_1 conda-forge
openff-qcsubmit 0.3.2 pyhd8ed1ab_0 conda-forge
openff-toolkit 0.10.6 pyhd8ed1ab_0 conda-forge
openff-toolkit-base 0.10.6 pyhd8ed1ab_0 conda-forge
openff-utilities 0.1.7 pyh1a96a4e_0 conda-forge

[jthorton]

Hi @ognjenperisic, thanks for trying out BespokeFit! The issue here is that the Executor.submit method expects a single BespokeOptimizationSchema object. In this case, you could submit each one and keep track of the ids like so

with BespokeExecutor(
n_fragmenter_workers = 1,
n_optimizer_workers = 1,
n_qc_compute_workers = 2,
qc_compute_worker_config=BespokeWorkerConfig(n_cores=1)
) as executor:
    ids = []
    for optimization_schema in workflow_schema:
        task_id = executor.submit(optimization_schema)
        ids.append(task_id)

When running multiple fits in parallel however it is often better to use the server and submit methods via the CLI as explained in the docs here. I hope this helps!

[ognjenperisic]

Hi @jthorton. Thank you very much for the advice. It fixed my issue.