MCL1 perturbation 23~66

Per our conversation during the meeting, I think we should also use MCL1 as a test system for ‘validation’ because it almost 3x smaller than CDK2. The perturbation I am working with is simple and is located at indices 19 > 13 in MCL1_ligands.sdf.zip. MCL1_protein_fixed.pdb.zip is the protein that I prepared using pdbfixer.

However after preparing the topology proposal with NonequilibriumFEPSetup, I get an error when trying to build the HybridTopologyFactory:

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-17-9565cf892879> in <module>()
      1 snefep = NonequilibriumSwitchingFEP(sfesetup[0], sfesetup[1], sfesetup[2], ncmc_nsteps=1000000, 
      2                                     nsteps_per_iteration=1000, trajectory_directory='mcl1_solvated_011118',
----> 3                                     trajectory_prefix='traj')

~/PycharmProjects/perses/perses/dispersed/relative_setup.py in __init__(self, topology_proposal, pos_old, new_positions, use_dispersion_correction, forward_functions, n_equil_steps, ncmc_nsteps, nsteps_per_iteration, temperature, trajectory_directory, trajectory_prefix, atom_selection, scheduler_address)
    362 
    363         #construct the hybrid topology factory object
--> 364         self._factory = HybridTopologyFactory(topology_proposal, pos_old, new_positions, use_dispersion_correction=use_dispersion_correction)
    365 
    366         #use default functions if none specified

~/PycharmProjects/perses/perses/annihilation/new_relative.py in __init__(self, topology_proposal, current_positions, new_positions, use_dispersion_correction, functions)
    146 
    147         #verify that no constraints are changing over the course of the switching.
--> 148         self._constraint_check_fast()
    149 
    150         #construct dictionary of exceptions in old and new systems

~/PycharmProjects/perses/perses/annihilation/new_relative.py in _constraint_check_fast(self)
    455         for constraint_pair in overlapping_constraints:
    456             if old_system_constraints[constraint_pair] != new_system_constraints[constraint_pair]:
--> 457                 raise ValueError("There is a changing constraint length in this system.")
    458 
    459     def _determine_interaction_group(self, atoms_in_interaction):

ValueError: There is a changing constraint length in this system.

I am currently trying to debug this. @pgrinaway can you give it a try on your system and see if you get the same error?

[pgrinaway]

One big issue I spotted was that the ligand in solvent TopologyProposal construction code doesn’t specify the old_alchemical_atoms argument, which means that all atoms are assumed to be alchemically changing. We should fix this throughout any drivers you have.

Ah, right–I had written that code before you changed the TopologyProposal to contain the old_alchemical_atoms argument. You’re right, we should update the construction code.

[pgrinaway]

Sounds good