ROSIE Rosetta VIP Server Documentation


This app is provides an interface to the RosettaVIP program. RosettaVIP identifies cavities in protein structural models using the RosettaHoles program and attempts to identify mutations that result in a better packed protein core. Although the original motivation for developing RosettaVIP was to 'rescue' designed proteins that exhibited poorly packed cores, it has also proven itself useful for stabilizing native proteins by filling in residual cavities.

The selection of mutations proceeds in two steps. First, a candidate list of mutations are identified based on rough geometric complementarity for the cavity to be filled. Next, each mutation is modeled on the structure and a local relaxation is performed to verify that the mutation can be accomodated without clashes. The best scoring mutation is retained, and if desired, further mutations may be identified.

The expected input is a structural model in pdb file format. Because the relaxation step of the RosettaVIP program is computationally expensive, the model should be trimmed to the smallest relevant domain.

  1. The "number of cycles" parameter controls the maximum number of iterations (mutations) that the program will attempt to find.
  2. The program will terminate if no beneficial mutation can be found, regardless of whether the specified number of cycles have been performed.
  3. The default number of iterations is 1, the maximum allowed on the web server is 5. If further iterations are desired, the Rosetta code can be downloaded, and the RosettaVIP program can be run on a local machine.
  4. The procedure used to identify cavities is stochastic. If the program fails to find a beneficial mutation, rerunning with a different random seed may succeed. The "number of attempts" parameter instructs the program to allow multiple attempts before considering a given iteration a failure. The default value for this parameter is 1.

A structural model in pdb file format. The file must be uploaded - files may not be directly taken from the PDB because these models can have multiple copies of monomers in the asymmetric unit, which will result in redundant (and slow!) calculations. The expected input for the VIP protocol is a model that is the result of a design calculation, or a native structure that has been pruned to the minimal relavant domain.

Interpreting Results:

The output from the "reports.txt" file is shown on the web page with the results. The following information is available in this file:

  1. Candidate mutations for an iteration: these are the mutations that score as favorable using the approximate scoring function that searches for amino acids that can improve packing. This doea not mean that these mutations can be accommodated with relaxation - that comes later. If more than one mutation is found at a particular position, only the best is reported. If no mutations at any positions scored as favorable, the message "No candidate mutations found!" will be reported.
  2. Mutations that were accommodated after relaxation. Each of the candidates identified above are subjected to relaxation using the full Rosetta scoring function to rule out mutations with clashes that cannot be eliminated by structural relaxation. If none of the candidate mutations pass this check, a message to that effect is reported.
  3. Mutations that are accepted. The best scoring mutation after relaxation is incorporated into the structure. If multiple iterations are requested, this is the model that serves as the input model for the next round.
The final model is output with the name "final_mutant..pdb" can be found by following the link on the results page entitled "[Full set of decoy structures created on this run (FS view)]". Intermediate structural models are also written whenever a new mutation is incorporated into the model. These are found by following the same link as the final structure, and are named "vip_iter_X.pdb", where X is the iteration number. The intermediate structure with the highest number is identical to "final_mutant.pdb".

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

VIP modeling tools developed by the Jim Havranek lab in Department of Genetics at Washington University St. Louis

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