search

chenruduan / OAReactDiff

An object-aware diffusion model for generating chemical reactions
MIT License
100 stars 24 forks source link

About the Transition1x given in this repository #5

Open allblueJT opened 3 months ago

allblueJT commented 3 months ago

Hello, there seems to be some additional properties in the Transitions1x given in this repo (train_addprop.pkl) compared to the origin, e.g. SMILES of the reactant/product. Could you tell me how to get such SMILES strings? And some of the SMILES strings caused a failure while using rdkit.Chem.MolFromSmiles(), which means an illegal SMILES string was provided.

zEuler1027 commented 3 months ago

SMILES comes with the trainsition1x dataset and I think most likely it comes from GDB7 dataset.

allblueJT commented 3 months ago

SMILES comes with the trainsition1x dataset and I think most likely it comes from GDB7 dataset.

Only rxns are provided in the Transition1x without correspondence between rxn and SMILES. I wonder how to get this correspondence..

Moreover, not all SMILES strings in the train_addprop.pkl are in the GDB7, e.g. O[C]1=NOC=[N]=1