Datasets:
| pretty_name: AlphaEnzyme data | |
| tags: | |
| - enzyme | |
| - protein | |
| - biology | |
| # AlphaEnzyme data | |
| Released input tables and supporting assets for the AlphaEnzyme specificity, kinetics, and mutation tasks. The corresponding code is available in the [`liuyong` branch of the AlphaEnzyme repository](https://github.com/yliust/AlphaEnzyme/tree/liuyong). Model checkpoints are published separately at [GENTEL-Lab/AlphaEnzyme](https://huggingface.co/GENTEL-Lab/AlphaEnzyme). | |
| ## Contents | |
| - `inputs/specificity/`: training, validation, and evaluation tables; reference banks and reaction fingerprints. | |
| - `inputs/kinetics/`: training, validation, and test tables; row-level auxiliary data, normalization statistics, and descriptor banks. | |
| - `inputs/mutation/`: variant tables, fold definitions, fold row/index files, normalization files, and component inputs. | |
| - `inputs/CONTRACT.json`: machine-readable file layout. | |
| - `inputs/README.md`: detailed directory instructions. | |
| Download into the AlphaEnzyme project root: | |
| ```bash | |
| hf download GENTEL-Lab/AlphaEnzyme-data --type dataset --local-dir data | |
| ``` | |
| This recreates `data/inputs/`. Large per-pair Pairformer/ESMC feature caches are not included; see `docs/NEW_DATA.md` in the code repository for feature-generation commands. The repository contains task input files, not a single table directly loadable by the Hugging Face `datasets` library. | |
| `SHA256SUMS` records the SHA-256 digest of each released file. | |