AlphaFold 2 vs RoseTTAFold

AlphaFold 2: DeepMind's open-source protein structure prediction model (monomer and multimer). RoseTTAFold: Baker lab's three-track neural network for protein structure and protein-protein complex prediction.

Facts below come from each project's own website, repository and paper. We don't publish benchmark claims here; see the papers for evaluations.

FieldAlphaFold 2RoseTTAFold
TaskStructure predictionStructure prediction
ModalitiesMonomer, multimerProtein sequence, complexes
MSA requiredYes (genetic databases)Yes
Open weights—Yes
GPU needNVIDIA GPUGPU required
LicenseApache-2.0MIT (code; trained weights non-commercial, Rosetta-DL Software license)
PricingOpen sourceOpen source (non-commercial)
Released2021-07-152021-06-29
PaperLinkLink
GitHubGitHubGitHub
GitHub stars14,8902,273
InterfaceDocker, Colab—

License and commercial use

FieldAlphaFold 2RoseTTAFold
Code licenseApache-2.0MIT
Weights license—trained weights non-commercial, Rosetta-DL Software license
Commercial usePermissive (commercial OK)Mixed (code vs weights differ)
AnswerYes. The recorded license is permissive, so closed-source commercial use is allowed if you keep the notices. DetailsPartly. The code and the weights (or hosted service) are under different terms; the restrictive part does not allow commercial use. Details

Add more Protein structure & design tools to this comparison →

More Protein structure & design comparisons

← All Protein structure & design tools