robetta.org

What It Is

Robetta is a protein structure prediction server developed by the Baker Lab at the University of Washington, built around the Rosetta macromolecular modeling suite. Its primary goal is to predict 3D protein structures from amino acid sequences, with ongoing evaluation through CAMEO. The site reports an average lDDT of around 69 over one year and more than 900 CAMEO pre-release PDB targets, making it a research-oriented tool with publicly available benchmarking.

Core Features and Ecosystem

Robetta offers five types of structure prediction options: RoseTTAFold, TrRosetta, RosettaCM, RosettaAB, and a pipeline that automatically predicts domains, models them separately, and then reassembles them. Comparative modeling uses the PDB100 template library, the MDB co-evolution model library, and template detection/alignment methods such as RaptorX, HHpred, Sparks-X, and Map align. Users can also upload custom PDB templates and FASTA-format alignments, edit alignments, and add distance restraints.

Multichain Modeling and Limitations

Robetta supports some multichain complex modeling. Homooligomers can be identified in comparative modeling through template biological units. Heterooligomers can be modeled with RoseTTAFold or CM, but RoseTTAFold requires users to provide a paired MSA, while CM supports custom templates, alignments, and inter-chain restraints. Note that heterooligomer comparative modeling has not yet been thoroughly tested. Robetta also cannot directly model protein–ligand complexes at present; ligands are omitted during submission.

Pricing, Ease of Use, and Support

The main documentation does not disclose a pricing model. Users must register and log in before submitting jobs. Through the web-based interactive interface, they can paste or upload sequences, templates, and restraints. Results can be viewed in My queue, are also sent by email, and can be downloaded as an archive containing models, templates, alignments, fragments, restraints, and commands. Jobs usually take from one day to several days, while large multi-domain tasks may take more than a week. To avoid excessively long queues, users can model only one domain at a time.

Pros, Cons, and Who It’s For

Robetta’s strengths include its strong academic foundation, broad range of methods, high level of customization, and support for error estimates and model confidence. Its drawbacks are that it is geared toward research workflows, with no visible API/SDK, self-hosting option, or pricing details, and its queueing and runtime make it unsuitable for strongly real-time use cases. It is best suited to researchers in structural biology, bioinformatics, and protein engineering. For users accessing it from China, the source text does not provide network access or payment information, so it is advisable to keep alternatives such as AlphaFold, ColabFold, and SWISS-MODEL ready as backups.