ROSIE Antibody Server Documentation


This app is the ROSIE interface to the RosettaAntibody3 program, which models the 3- D structure of an antibody Fv region starting from the sequences of the light chain and heavy chain variable domains.

RosettaAntibody3 consists of two stages. The first stage is canonical template selection and assembly. Templates are chosen using BLAST from antibody crystal structures in the PDB. Template searches are independent for each CDR (L1, L2, L3, H1, H2, and H3) and for each framework region (FRL and FRH), respectively. They are assembled using a Rosetta protocol, resulting in a crude structure. If multiple-template mode is enabled, 10 crude structures, each with a different VL/VH orientation, will be created. In the second stage, CDR-H3 is remodeled de novo and VL/VH orientation is optimized using a Rosetta protocol. Paratope side chains and loop backbones are refined simultaneously.


  1. Only Fv region can be modeled.
  2. Six CDRs are automatically identified based on an expert knowledge-based system. Hence some conserved residues are assumed to be present in the input sequences. For example, Cys residues at position L22, L92, H22 and H92, and Trp residues at position L35, H36 and H103 must be present in input sequences.
  3. All CDRs other than H3 are modeled using templates in the PDB whereas CDR-H3 is re-modeled de novo using a kinematic loop modeling algorithm in a Rosetta protocol.
  4. If there is no reasonable template of a CDR based on a bit score in BLAST, a random template of the same length as the query is chosen.
  5. For CDR-L3, if the length is larger than 10, it is less likely that reasonable templates are available in the current database.
  6. Multiple-template mode is the recommended setting
  7. The accuracy of CDR-H3 is strongly length dependent. If the length of CDR- H3 is less than 10, it is more likely that the model is accurate (RMSD ~1.5A can be expected; see Sivasubramanian PROTEINS 2009).
  8. The PDB files of the 10 output models are numbered by the Chothia scheme.
  9. The ten output models can be used together in subsequent antibody-antigen docking simulations, such as SnugDock or EnsembleDock.
  10. Non-standard amino acids are not supported.

The mandatory inputs are sequences of a light chain and a heavy chain of Fv region.

Interpreting Results:

The server returns templates information, one (10) a crude assembled model(s), 10 low-energy refined models as PDB files and the score file. The reliability of models is strongly depends on the length of CDR-H3. See Tip 7 above.

Please cite the following article when referring to results from our ROSIE server:

  1. Sivasubramanian A, Sircar A, Chaudhury S, Gray JJ., "Toward high-resolution homology modeling of antibody Fv regions and application to antibody-antigen docking". Proteins. 2009 Feb 1;74(2):497-514. doi: 10.1002/prot.22309. Link

  2. Marze NA, Gray JJ, "Improved Prediction of Antibody VL–VH Orientation". Under review.

  3. Lyskov S, Chou FC, Conchúir SÓ, Der BS, Drew K, Kuroda D, Xu J, Weitzner BD, Renfrew PD, Sripakdeevong P, Borgo B, Havranek JJ, Kuhlman B, Kortemme T, Bonneau R, Gray JJ, Das R., "Serverification of Molecular Modeling Applications: The Rosetta Online Server That Includes Everyone (ROSIE)". PLoS One. 2013 May 22;8(5):e63906. doi: 10.1371/journal.pone.0063906. Print 2013. Link

Additional references of interest:

  • A. Sircar, E. Kim, & J. J. Gray, "RosettaAntibody: Antibody Variable Region Homology Modeling Server," Nucleic Acids Research 37 (Web Server Issue), W474-W479 (2009). Link

We welcome scientific and technical comments on our server. For support please contact us at Rosetta Forums with any comments, questions or concerns.