First Commit - "Removing existing SmilesGenerator implementation."
The two before it are from another patch - but made sense to continue from there.
Couple of regressions all fixed I think. This is the initial behaviour I still might change this to turn on/off canonicalisation - the existing CanonicalLabeller shouldn't be used for isomeric SMILES but I've left it for now to match the tests assertions.
Couple of regressions elsewhere but mainly issues with the new generator not changing the aromatic input. Also the fragmenters weren't updating the hydrogen counts so the SMILES generated weren't canonical and also wrongly interrupted. The fix is currently to do the atom-typing at every point before generating SMILES. Rajarshi might want to look at these though as it exposed some quirks, I left one of them in the FragmentsUtilsTest...
With that we should be able to round trip stereo between InChI/SMILES.
Log in to post a comment.