Overview
| Name | SaProt · aliases: SaProt, SaProt 650M, SaProt 35M, SaProt 1.3B |
|---|---|
| Full name | Short name only; neither paper nor repository gives an expanded name |
| Ambiguous names (not auto-linked) | None |
| Family | Not a family entry. 35M and 650M are two sizes of one model, not family members |
| Organization | Westlake University, Westlake Representation Learning Lab |
| Organization type | Academic lab |
| Country | China |
| Model class | Protein language model (structure-aware) |
| Domain | Proteins (also: Structure & co-folding) |
| First release | † partly verified · preprint Basis: First public bioRxiv preprint Note: The preprint DOI is not Crossref-confirmed; the date comes from the same-day weights repository and README notes (unverified) source |
| Other milestones | |
| Latest update | · journal Basis: Journal publication in Nature Biotechnology SaProt/SaprotHub published in a journal; 1.3B weights earlier on 2024-12-09, README touched in 2026-03 source The most recent sourced event, not a claim that nothing changed since. |
| Status | Maintained only · Journal publication and repository upkeep, no new generation |
| Openness | Code: Open Weights: Open Code and weights public on GitHub and Hugging Face; licence text not retrieved this round, so terms are left empty |
| Benchmarks reported | 3 counted (board 2, self 1) |
| Wet-lab validation | None recorded |
| Unverified fields | † first_release, benchmarks_reported[2], release_milestones.weights_release, open_status.weights_license, open_status.code_license |
| As of | 2026-11-16 |
What it does
Combines each residue's amino acid with a local structure token into a structure-aware vocabulary, pretrained on a public predicted-structure database, for zero-shot mutation-effect scoring and fine-tuning, with no extra MSA. It is the simpler predecessor of structure-enhanced zero-shot models such as ProSST and VenusREM.
Benchmarks reported
Every row is one pairing from the survey's relation file; a score belongs to the benchmark page's ranking, where it is shown with metric, eval set, source type and date. 3 of 3 pairings are counted.
| Benchmark | Relation | Eval set / protocol | Source type | Note and evidence |
|---|---|---|---|---|
| 02 ProteinGym | Official leaderboard or results | zero-shot substitution DMS summary | Board | Entry on the official summary table · evidence |
| 34 AbBiBench | Official leaderboard or results | antibody binding-affinity correlation, structure-aware LM category | Board | Benchmark-registered score †unverified · no source link |
| 26 FLIP / FLIP2 | Developer report, official protocol | Meltome human-cell (called Thermostability in the paper); AAV 2-vs-rest | Self-reported | Self-reported on two datasets; a third-party re-evaluation also exists · evidence |
Other evaluations (outside the 44)
- Third-party FLIP GB1 extrapolation re-evaluation (third party re evaluation) Third-party: A third-party paper re-ran the low-vs-high split and found it no better than a single-sequence baseline (figures not quoted here)
- Time-dependence of the board position (Public benchmark not among the 44) Third-party: A maintainer blog post discusses it as an example of the structure-token route when weighing scale against gains
Developers report: Developers report that adding structure tokens to the vocabulary lets a mid-sized model beat same-size sequence-only models on zero-shot mutation effects without extra MSA.
Independent evidence: A third-party paper independently re-ran it on a FLIP extrapolation split and did not reproduce the advantage; no wet lab, no independent replication.
Training data
Group in Figure 7: Public data only.
| Class | Value | Detail |
|---|---|---|
| a. Public databases | Yes | About 40M public predicted structures and sequences, encoded into structure tokens with a structure-alignment tool source |
| b. Private or licensed data | No | No private or licensed data reported |
| c. Own wet-lab data used for training | No | |
| d. Wet lab used only for validation | No |
Wet-lab validation
None. The developers report no wet-lab validation.
Background
- Organization: An academic lab with no separate company: a representation-learning lab at Westlake University in Hangzhou working on AI for life science and on deep learning for retrieval and recommendation. The model appeared at a 2024 machine-learning conference and in a journal in 2025.
- Funding: Paper acknowledgements not retrieved; funding left blank
- Products and partnerships: A companion hosting and fine-tuning platform (SaprotHub) shipped with the journal version; the structure-token approach is treated as a peer baseline by ProSST and VenusREM, and third-party inference platforms offer it online.
- Related models: ProSST, VenusREM, ESM3, ESM C
Reading notes
- The "once first" claim is a 2024 snapshot; board position shifts with versions and newcomers and is not a current statement.
- Structure inputs come from public predicted structures, so regions without structure or with low confidence are weak spots.
- A third-party re-evaluation on an extrapolation split shows structure tokens do not guarantee better extrapolation.
- Computational evaluation only, no wet lab: it speaks to ranking, not to designability.
Caveats
- The AbBiBench pairing rests only on this site's page (unverified).
- First-release evidence is thin, taken from the same-day weights repository and README (unverified).
- No wet-lab validation, and no independent replication of its board position.
Sources
- SaProt conference paper (2024) (paper)
- Official repository (repository)
- SaProt/SaprotHub journal version (journal)
- Weights repository (repository)
- ProteinGym official repository and board (leaderboard)
- Third-party FLIP re-evaluation paper (third-party paper)
- Board maintainer blog post (second-hand) (second-hand) second-hand
Corrections
None recorded.
Part of the AI for Bio benchmarks survey · all 112 models · open in the explorer. Record as of 2026-11-16.