perses: error while running a multitarget designer

I set up a multitarget designer calculation using an edited form of the Ablaffinitytestsystem class. It looks like the systems are being set up appropriately, but I keep getting the following error:

Minimizing 'explicit-src-imatinib'...
Initial energy is   -58550.422 kcal/mol
Final energy is    -137454.317 kcal/mol
Minimizing 'explicit-abl-imatinib'...
Initial energy is   -31761.928 kcal/mol
Final energy is     -31756.639 kcal/mol
********************************************************************************
MultiTargetDesign sampler iteration        0
================================================================================
SAMS sampler iteration     0
--------------------------------------------------------------------------------
Expanded Ensemble sampler iteration        0
................................................................................
MCMC sampler iteration 0
Taking 5 steps of Langevin dynamics...
potential  = -53268.502 kT
kinetic    = 757989197113.024 kT
force norm = 411813.866 kT/A/dof
Final energy is          nan kT
................................................................................
Updating chemical state with ncmc-geometry-ncmc scheme...
Proposing new topology...
Warning: MCS search truncated
No handlers could be found for logger "alchemy.alchemy"
Proposed transformation: Cc1ccc(cc1Nc2nccc(n2)c3cccnc3)NC(=O)c4ccc(cc4)C[NH+]5CCN(CC5)C => CN1CCC(CC1)COc2cc3c(cc2OC)c(ncn3)Nc4ccc(cc4F)Br
Generating new system...
Performing NCMC annihilation
Traceback (most recent call last):
  File "abl_src_affinity.py", line 215, in <module>
    system.designer.run(niterations=100)
  File "/cbio/jclab/home/albaness/miniconda/lib/python2.7/site-packages/perses-0.1-py2.7.egg/perses/samplers/samplers.py", line 1094, in run
    self.update()
  File "/cbio/jclab/home/albaness/miniconda/lib/python2.7/site-packages/perses-0.1-py2.7.egg/perses/samplers/samplers.py", line 1076, in update
    self.update_samplers()
  File "/cbio/jclab/home/albaness/miniconda/lib/python2.7/site-packages/perses-0.1-py2.7.egg/perses/samplers/samplers.py", line 1040, in update_samplers
    sampler.update()
  File "/cbio/jclab/home/albaness/miniconda/lib/python2.7/site-packages/perses-0.1-py2.7.egg/perses/samplers/samplers.py", line 956, in update
    self.update_sampler()
  File "/cbio/jclab/home/albaness/miniconda/lib/python2.7/site-packages/perses-0.1-py2.7.egg/perses/samplers/samplers.py", line 924, in update_sampler
    self.sampler.update()
  File "/cbio/jclab/home/albaness/miniconda/lib/python2.7/site-packages/perses-0.1-py2.7.egg/perses/samplers/samplers.py", line 799, in update
    self.update_state()
  File "/cbio/jclab/home/albaness/miniconda/lib/python2.7/site-packages/perses-0.1-py2.7.egg/perses/samplers/samplers.py", line 719, in update_state
    [ncmc_old_positions, ncmc_elimination_logp, potential_delete] = self.ncmc_engine.integrate(topology_proposal, positions, direction='delete')
  File "/cbio/jclab/home/albaness/miniconda/lib/python2.7/site-packages/perses-0.1-py2.7.egg/perses/annihilation/ncmc_switching.py", line 315, in integrate
    context.setPositions(initial_positions)
  File "/cbio/jclab/home/albaness/miniconda/lib/python2.7/site-packages/simtk/openmm/openmm.py", line 13580, in setPositions
    return _openmm.Context_setPositions(self, positions)
Exception: Particle coordinate is nan
Compute mode is already set to DEFAULT for GPU 0000:03:00.0.

Looks like I’m having trouble with the NCMC moves. The script that I’m using can be found here