REXzyme
REXzyme is a protein structure & design tool, first public in 2026, described in bioRxiv preprint (2026). Reaction-to-enzyme translation model (Efficient T5, ~770M) that generates amino-acid sequences from a user-specified chemical reaction in SMILES; Colab notebook for inference.
| Category | Protein structure & design |
|---|---|
| Task | Reaction-conditioned enzyme sequence generation (SMILES reaction -> AA sequence) |
| Interface | Hugging Face Transformers; Google Colab notebook |
| Website | huggingface.co/AI4PD/REXzyme |
| Released | 2026-11-05 (source) |
| Paper | https://www.biorxiv.org/content/10.64898/2026.11.04.764151v1 · bioRxiv preprint (2026) |
| Code license | Apache-2.0 |
| Weights license | model card and weights Apache-2.0 on Hugging Face |
| What this allows | Permissive (commercial OK) Use, modify and ship it inside closed-source commercial products; keep the license text and copyright notices. What do these mean? |
| License source | huggingface.co/AI4PD/REXzyme/raw/main/README.md · checked 2026-11-06 |
| Pricing | Open source |
| Tags | #enzyme-design#generative#reaction-conditioned#protein-language-model |
| Last checked | · links verified |
Can I use REXzyme commercially?
Yes. The recorded license is permissive, so closed-source commercial use is allowed if you keep the notices.
- Code: Apache-2.0
- Weights: model card and weights Apache-2.0 on Hugging Face
- Source: huggingface.co/AI4PD/REXzyme/raw/main/README.md · license checked 2026-11-06
What the license types mean · not legal advice.
Specs
| Task | Reaction-conditioned enzyme sequence generation (SMILES reaction -> AA sequence) |
|---|---|
| Modalities | — |
| MSA required | — |
| Open weights | Yes (Hugging Face AI4PD/REXzyme checkpoints) |
| GPU need | — |
| Interface | Hugging Face Transformers; Google Colab notebook |
Compare with…
Pick 1–3 other Protein structure & design tools to compare side by side with REXzyme.
Related
DeepMind's open-source protein structure prediction model (monomer and multimer).
Structure prediction of proteins, nucleic acids, ligands and their complexes; hosted web server plus open code, weights on request for non-commercial use.
Open biomolecular interaction models (Boltz-1, Boltz-2) for complex structure and binding-affinity prediction.