bespokefit cannot coexist with pymbar 4

[xiki-tempula]

Description

Another forcebalance dependency issue

Reproduction

mamba create -n bespokefit -c conda-forge -c openeye openff-bespokefit openeye-toolkits xtb-python "libblas=*=*mkl" "pymbar=4"

openff-bespoke executor run --smiles             "CC(=O)NC1=CC=C(C=C1)O"    \
                            --workflow           "default"              \
                            --output             "acetaminophen.json"   \
                            --output-force-field "acetaminophen.offxml" \
                            --n-qc-compute-workers 4                    \
                            --qc-compute-n-cores   1                    \
                            --default-qc-spec xtb gfn2xtb none

Output

cannot import name 'pymbar' from 'pymbar'                                                                                       
 (/exs/shared/collaboration/teams/mdteam/private/envs/bespokefit/lib/python3.8/site-packages/pymbar/__init__.py)", "traceback": "Traceback (most recent call last):
 File "/exs/shared/collaboration/teams/mdteam/private/envs/bespokefit/lib/python3.8/site-packages/celery/app/trace.py", line 477, in trace_task
    R = retval =  
 fun(*args, **kwargs)
  File "/exs/shared/collaboration/teams/mdteam/private/envs/bespokefit/lib/python3.8/site-packages/celery/app/trace.py", line 760, in       
 __protected_call__
    return self.run(*args, **kwargs)
  File                                                                                                    
 "/exs/shared/collaboration/teams/mdteam/private/envs/bespokefit/lib/python3.8/site-packages/openff/bespokefit/executor/services/optimizer/worker.py", line 63, in 
 optimize
    result = optimizer.optimize(
  File                                                                                                                  
 "/exs/shared/collaboration/teams/mdteam/private/envs/bespokefit/lib/python3.8/site-packages/openff/bespokefit/optimizers/model.py", line 232, in optimize
       
 cls.prepare(schema, initial_force_field, ".")
  File                                                                                                             
 "/exs/shared/collaboration/teams/mdteam/private/envs/bespokefit/lib/python3.8/site-packages/openff/bespokefit/optimizers/model.py", line 197, in prepare
        
 cls._prepare(schema, initial_force_field, root_directory)
  File                                                                                                   
 "/exs/shared/collaboration/teams/mdteam/private/envs/bespokefit/lib/python3.8/site-packages/openff/bespokefit/optimizers/forcebalance/forcebalance.py", line 95,  
 in _prepare
    ForceBalanceInputFactory.generate(root_directory, schema, initial_force_field)
  File                                                             
 "/exs/shared/collaboration/teams/mdteam/private/envs/bespokefit/lib/python3.8/site-packages/openff/bespokefit/optimizers/forcebalance/factories.py", line 710, in 
 generate
    target_sections.append(target_factory.generate(".", target))
  File                                                                                
 "/exs/shared/collaboration/teams/mdteam/private/envs/bespokefit/lib/python3.8/site-packages/openff/bespokefit/optimizers/forcebalance/factories.py", line 257, in 
 generate
    cls._generate_target(target, target_records)
  File                                                                                                  
 "/exs/shared/collaboration/teams/mdteam/private/envs/bespokefit/lib/python3.8/site-packages/openff/bespokefit/optimizers/forcebalance/factories.py", line 368, in 
 _generate_target
    super(TorsionProfileTargetFactory, cls)._generate_target(target, qc_records)
  File                                                          
 "/exs/shared/collaboration/teams/mdteam/private/envs/bespokefit/lib/python3.8/site-packages/openff/bespokefit/optimizers/forcebalance/factories.py", line 276, in 
 _generate_target
    from forcebalance.molecule import Molecule as FBMolecule
  File                                                                              
 "/exs/shared/collaboration/teams/mdteam/private/envs/bespokefit/lib/python3.8/site-packages/forcebalance/__init__.py", line 135, in <module>
    from . import   
 parser, forcefield, optimizer, objective, output
  File                                                                                                            
 "/exs/shared/collaboration/teams/mdteam/private/envs/bespokefit/lib/python3.8/site-packages/forcebalance/forcefield.py", line 107, in <module>
    from          
 forcebalance import gmxio, qchemio, tinkerio, custom_io, openmmio, amberio, psi4io, smirnoffio
  File                                                              
 "/exs/shared/collaboration/teams/mdteam/private/envs/bespokefit/lib/python3.8/site-packages/forcebalance/gmxio.py", line 22, in <module>
    from                
 forcebalance.liquid import Liquid
  File "/exs/shared/collaboration/teams/mdteam/private/envs/bespokefit/lib/python3.8/site-packages/forcebalance/liquid.py",    
 line 27, in <module>
    from pymbar import pymbar
ImportError: cannot import name 'pymbar' from 'pymbar'                                                         
 (/exs/shared/collaboration/teams/mdteam/private/envs/bespokefit/lib/python3.8/site-packages/pymbar/__init__.py)

Software versions

• Which operating system and version are you using? Linux
• How did you install BespokeFit? conda
• Are you using Apple Silicon? If so, are you running BespokeFit in Rosetta (osx-64) or natively (osx-arm64)?

[j-wags]

Hi @xiki-tempula,

We've tried to get around this by having ForceBalance 1.9.5 pin to pymbar<4 and BespokeFit 0.2.2 pin to ForceBalance 1.9.5. Could you give more detail about how you installed the environment?

[j-wags]

Oh, my mistake, I see you did put your install instructions - By requiring pymbar=4, you're causing the solver to pull in old versions of everything else. If you drop the version constraint on pymbar you should get a working build.

The pymbar compatibility issue is tracked at leeping/forcebalance#260

[xiki-tempula]

I see. Thanks. I got some other package that relies on pymabr 4, so when I create a conda environment that pins pymbar to 4. It pulls in the old version 0.1.3. I guess I have to create two enviornment for them then.