Open pgrinaway opened 7 years ago
This issue has been addressed for the nonequilibrium switching project by subclassing SmallMoleculeSetProposalEngine
and skipping the round-trip through a SMILES string. I'm leaving this issue open, however, since it still needs to be resolved for the general case.
(will raise issue there).
Could you clarify which codebase has the primary issue?
The long-term fix for OpenMM Topology
to OEMol
conversion is to migrate to using the OpenMM-OEChem topology conversion functions in https://github.com/oess/oeommtools
This is being moved to a generally-available conda package that we will be able to use in the longer term, but it would be useful to see if we can temporarily borrow the code in the meantime (license permitting): https://github.com/oess/oeommtools/issues/7
In #426 I was able to switch to a scheme where we always use canonical isomeric SMILES strings with explicit hydrogens so that we can be explicit in which protonation/tautomer states to consider for relative free energy calculations.
Let's keep this issue open, since we need to eventually support two modes of operation:
Changed label to enhancement
since the bug has been addressed in #426
Right now, the SmallMoleculeProposalEngine does not correctly respect protonation states, due to code that converts the OpenMM topology to an oemol (will raise issue there).