SaProt
Protein language model integrating amino-acid sequence and Foldseek structural alphabet.
Developer: Westlake REPL; Su et al.
Overview
- Developer
- Westlake REPL; Su et al.
- Category
- protein language models
- Commercial rights
- Unverified
- License fee
- Unverified
- Open weights
- Yes
- Last verified
- 2026-10-18
- Added
- 2026-10-18
Capabilities
Uses combined amino-acid and 3Di structural tokens to produce representations and predictions for protein functions and mutation effects. Repository warns some checkpoints require structure tokens for optimal performance.
When to use it
Use structure-aware protein embeddings or property-prediction research to complement experimental developability assays, not as a direct antibody-affinity/KD measurement.
Compute
Compute category: single-gpu. See original installation instructions or platform terms for specific hardware costs and runtime.
Example use cases
- Use structure-aware protein embeddings or property-prediction research to complement experimental developability assays, not as a direct antibody-affinity/KD measurement.
Source-based workflow description. No experimental performance ranking is implied.
Known limitations
- Protein property predictors and stability variants must be evaluated per downstream task; available source code and pretrained weights alone do not confirm full commercial rights.
Check the original linked sources and the terms of every model dependency.
Commercial licensing
Commercial permission — Unverified
MIT repository code; model checkpoints and pretrained upstream dependency rights not individually reviewed.
Protein property predictors and stability variants must be evaluated per downstream task; available source code and pretrained weights alone do not confirm full commercial rights.
Commercial fee — Unverified
No complete model + checkpoint + cloud compute + service charges have been verified. Public research software or downloadable weights do not imply a free commercial deployment.
Official references
Validate your designed binder.
Seq2Study coordinates expression, affinity, stability and developability testing for AI-designed antibodies and proteins.