A chemical synthesis route dataset constructed from the USPTO reaction dataset (1976-Sep2016) and a list of commercially available
building blocks from eMolecules (~23.1M molecules). After processing, the dataset has 299202 training routes, 65274 validation routes, 190 test routes, and the corresponding target molecules.
Variants: USPTO-190
This dataset is used in 1 benchmark:
| Task | Model | Paper | Date |
|---|---|---|---|
| Multi-step retrosynthesis | PDVN | Retrosynthetic Planning with Dual Value … | 2023-01-31 |
| Multi-step retrosynthesis | RetroGraph | RetroGraph: Retrosynthetic Planning with Graph … | 2022-06-23 |
| Multi-step retrosynthesis | EG-MCTS | Retrosynthetic Planning with Experience-Guided Monte … | 2021-12-11 |
| Multi-step retrosynthesis | Retro* plus | Self-Improved Retrosynthetic Planning | 2021-06-09 |
| Multi-step retrosynthesis | Retro* | Retro*: Learning Retrosynthetic Planning with … | 2020-06-29 |
Recent papers with results on this dataset: