Rework logic for adding ring-closure torsion biases for fused/bridging ring systems

One of the recurring errors I’ve seen (using Perses 0.8.1, in my slightly janky Docker container)

Result: the atom mapping fails, and the edge does not run

Traceback (most recent call last):
  File "/miniconda/lib/python3.8/site-packages/perses/app/setup_relative_calculation.py", line 849, in <module>
    run()
  File "/miniconda/lib/python3.8/site-packages/perses/app/setup_relative_calculation.py", line 700, in run
    setup_dict = run_setup(setup_options)
  File "/miniconda/lib/python3.8/site-packages/perses/app/setup_relative_calculation.py", line 428, in run_setup
    fe_setup = RelativeFEPSetup(ligand_file, old_ligand_index, new_ligand_index, forcefield_files,phases=phases,
  File "/miniconda/lib/python3.8/site-packages/perses/app/relative_setup.py", line 426, in __init__
    self._ligand_positions_new_solvated, self._solvent_logp_proposal = self._geometry_engine.propose(self._solvent_topology_proposal,
  File "/miniconda/lib/python3.8/site-packages/perses/rjmc/geometry.py", line 235, in propose
    logp_proposal, new_positions, rjmc_info, atoms_with_positions_reduced_potential, final_context_reduced_potential, neglected_angle_terms, omitted_terms = self._logp_propose(top_proposal, current_positions, beta, direction='forward', validate_energy_bookkeeping = validate_energy_bookkeeping)
  File "/miniconda/lib/python3.8/site-packages/perses/rjmc/geometry.py", line 400, in _logp_propose
    growth_system_generator = GeometrySystemGenerator(top_proposal.new_system,
  File "/miniconda/lib/python3.8/site-packages/perses/rjmc/geometry.py", line 2105, in __init__
    self._determine_extra_torsions(extra_modified_torsion_force, reference_topology, growth_indices)
  File "/miniconda/lib/python3.8/site-packages/perses/rjmc/geometry.py", line 2313, in _determine_extra_torsions
    p4_target_growth_index = min(tup[1] for tup in _nbr_to_growth_index_tuple if tup[1] > p1_target_growth_index)
ValueError: min() arg is an empty sequence

[schmolly]

It’s not super high priority

[jchodera]

I’ve updated the issue with a more accurate description of what needs to be fixed.

@schmolly : Can you let us know whether this is a super high priority issue to include before we release 0.9.1, or whether we could include this fix in 0.10.0?