To update the TS database, we need to work backwards: generate a TS from a log file. We currently rely on the atom sorting labels to identify the atoms in the reaction center. However, sometimes the sorting labels in the log file don't match up with the TS sorting labels (if using a different version of ase for example). We could make the log to TS method more robust if we include key information in the gaussian log file needed to generate the TS, such as ts.reaction_label, ts.reaction_family, ts.direction, and ts.rmg_molecule.get_all_labeled_atoms(). We could do this by adding these attributes to the title block of the gaussian input file (unfortunately Gaussian doesn't print comments in output). However, ase's gaussian calculator does not have a title parameter (it just writes Gaussian input prepared by ASE), so we would need to change ase Gaussian code
To update the TS database, we need to work backwards: generate a TS from a log file. We currently rely on the atom sorting labels to identify the atoms in the reaction center. However, sometimes the sorting labels in the log file don't match up with the TS sorting labels (if using a different version of ase for example). We could make the log to TS method more robust if we include key information in the gaussian log file needed to generate the TS, such as
ts.reaction_label
,ts.reaction_family
,ts.direction
, andts.rmg_molecule.get_all_labeled_atoms()
. We could do this by adding these attributes to the title block of the gaussian input file (unfortunately Gaussian doesn't print comments in output). However, ase's gaussian calculator does not have atitle
parameter (it just writesGaussian input prepared by ASE
), so we would need to change ase Gaussian code