Considering conformers, cis/trans isomers, and stereoisomers when deducing reactions

[samblau]

By limiting ourselves to isomorphism checking, we are currently unable to account for differences between stereoisomers and cis/trans isomers, which are distinct non-superimposable molecules. We previously sought to use Yu-Hang's graph kernel M3 metric to distinguish between conformers, and had some success with it, but we never got to the point where we distinguished between reactions involving different isomers. In the medium term, this could be important for TFSI, which has a cis/trans isomer, or FEC, which contains stereo centers.