Protein Structure · Baker Lab / UW
RoseTTAFold All-Atom
Full-atom structure prediction for proteins, non-canonical residues, glycans, metals, and small-molecule cofactors.
What RoseTTAFold All-Atom is
RoseTTAFold All-Atom is a biomolecular structure and complex prediction model developed by the Baker Lab at the University of Washington. It generates full-atom structural coordinates from protein sequence and small-molecule inputs.
While many predictors focus solely on standard protein backbones, this model addresses the wider chemical context of macromolecular systems. It models modified residues, metal ions, glycans, and small-molecule cofactors alongside the protein chain, catering to biological questions where surrounding chemistry is critical.
Key facts
- Version
- 1.0
- Domain
- Protein structure
- Tasks
- Structure prediction, complex prediction
- Modalities
- Protein sequence, small molecule
- GPU memory
- 40 GB
What you can use it for
- Cofactor and Metal-Binding Prediction
- Model protein complexes containing coordinate metal ions and small-molecule cofactors alongside the polypeptide chain.
- Modified Residue and Glycan Modelling
- Predict full-atom structures that incorporate non-canonical amino acids, post-translational modifications, and glycans.
- All-Atom Complex Prediction
- Generate coordinated three-dimensional atomic coordinates in PDB format to examine chemical interactions beyond simple backbone contacts.
Run it here
No installation and no GPU of your own: fill in the inputs and it runs on our servers. Results stay in your browser unless you sign in.
Not runnable yet
This model isn't available to run right now.Run this model
RoseTTAFold All-Atom
Strengths and limits
Good at
- Predicts full-atom representations including non-canonical residues, glycans, metals, and small-molecule cofactors.
- Provides residue-level confidence metrics via per-residue pLDDT scores.
- Outputs structural coordinates in standard, viewable and downloadable PDB format.
Watch out for
- Requires substantial hardware resources, specifically 40 GB of GPU memory.
- Residue predictions scoring below 70 on the pLDDT scale are generally not reliable.
Inputs and outputs
You provide
Protein sequence
Single-letter amino acid code. Whitespace and FASTA headers are stripped automatically.
You get back
All-atom coordinates in PDB format, viewable and downloadable.
pLDDT score per residue (0–100). Above 70 is generally reliable.
Licence
Academic and non-commercial use only · MIT code / non-commercial weights
Code is MIT but the released weights are non-commercial. Commercial tier routes to Boltz-2 instead.
Questions
Can I use RoseTTAFold All-Atom commercially?
Academic and non-commercial use only. Code is MIT but the released weights are non-commercial. Commercial tier routes to Boltz-2 instead.
How do I run RoseTTAFold All-Atom without installing anything?
Use the form under “Run it here” on this page: fill in the inputs and press Run model. It runs on our servers on 2× A100, typically in 5.0min, and a run costs about 1,680 credits. Visitors without an account get 1,000 credits a month.
What inputs are required to run the model?
The model accepts protein sequences and small molecules as inputs.
What outputs does RoseTTAFold All-Atom generate?
It outputs all-atom three-dimensional coordinates in PDB format alongside a vector of per-residue pLDDT confidence scores ranging from 0 to 100.
How are prediction confidence scores evaluated?
Confidence is assessed via the pLDDT score for each residue, with values exceeding 70 generally regarded as reliable.
Written from RoseTTAFold All-Atom's own documentation. Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.