Multi-Modal Foundation · bjing-mit
Promera
Generative model for biomolecular co-folding, binder filtering, and miniprotein and nanobody design.
What Promera is
Promera is a unified generative model developed by researchers at the Massachusetts Institute of Technology and the University of Texas at Austin (Jing, Bafna, Diaz, Klivans, and Berger) for biomolecular structure prediction, filtering, and molecular design.
The model unifies all-atom co-folding with confidence-based binder filtering and sequence generation. It addresses therapeutic design challenges by supporting controllable generation of miniproteins and nanobodies using masked-sequence prediction.
How it works
Promera couples all-atom co-folding mechanisms with masked-sequence prediction. This unified framework enables both the evaluation of biomolecular complexes and the controllable generation of binding sequences, accepting conditioning inputs such as specific target epitopes.
Key facts
- Release year
- 2026
- Category
- Biomolecular Structure Model
- Core tasks
- Structure prediction, binder filtering, sequence design
What you can use it for
- Therapeutic Structure Prediction
- Predicting all-atom co-folded structures across therapeutic biomolecular categories.
- Candidate Binder Filtering
- Filtering potential binding candidates using internal confidence assessments.
- Controllable Binder Generation
- Designing nanobodies and miniproteins via masked-sequence prediction conditioned on specified target epitopes.
Run it here
We list this model but cannot run it yet: no pipeline_tag — needs a human.
Everything on this page still applies. Running it needs a loader for this repository, not a bigger machine.
Strengths and limits
Good at
- Exceeds the performance of OpenFold3-p2 and Boltz-2 on therapeutic structure prediction categories.
- Unifies all-atom structure co-folding, confidence-based filtering, and molecular design within a single generative architecture.
- Supports epitope-conditioned design for targeted miniprotein and nanobody development.
Watch out for
- Available training details, exact parameter scale, and full architecture descriptions are not documented in the source record.
- Input modalities and sequence length limits are not specified in the source.
Inputs and outputs
You provide
Sequence or text
A protein or nucleotide sequence, or free text — whichever this model was trained on.
You get back
Mean-pooled representation, one vector per input.
Licence
Check the licence before commercial use · mit
The mit licence carries terms worth reading before you depend on it.
Questions
Can I use Promera commercially?
Check the licence before commercial use. The mit licence carries terms worth reading before you depend on it.
What formats of binders can Promera generate?
Promera is designed for the controllable generation of miniproteins and nanobodies.
How does Promera perform against other co-folding models?
Promera surpasses OpenFold3-p2 and Boltz-2 on therapeutic co-folding categories.
Can Promera target specific epitopes?
Yes, it supports controllable masked-sequence prediction with optional epitope conditioning.
How to cite it
Jing, Bafna, Diaz, Klivans, Berger. Unified generative model for biomolecular structure prediction, filtering, and design. bioRxiv 2026.
Written from Promera's own documentation (model card). Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.