RosettaDock FAQ


  1. Where can I obtain RosettaDock?


    1. RosettaDock is available as part of the Rosetta++ package—see  It has been developed primarily on linux, but it also runs on windows and mac platforms.


  1. I’d like to recalculate the score for certain structures.  Is there an easy way to do that


    1. You can score a pdb by running with the options '-dock -score' and perhaps '-fullatom' for the most-accurate all-atom score.

      Note that I wouldn't compare scores between proteins with different sequences.  The scores are tuned to be relative to other conformations of the same protein pairs.


  1. Do you prefer the dock_mcm option or the more explicit -dock_pert n p r for perturbation?  I noticed that in the example you have -dock_mcm.


    1. -dock_mcm determines the search protocol, and yes, this is the best option for that, it means going through cycles of small perturbations, repacking, minimizationmcm=monte carlo minimization

      -dock_pert specifies the start condition, that is, how is the starting structure perturbed before beginning the whole search process.  Using this option and creating 1000 decoys means you are starting from a range of positions near the input pdb position.

      In the JMB paper fig 1a, -dock_pert is happing in the pink box 'random start position', and ‘-dock_mcm' turns on the 50x loop in fig 1b


  1. Also in the example, -ex1 -ex34 are used.  Why not -ex2 -ex3 as well?


    1. This is a bit of a scientific mystery at this point and probably arises from our imperfect representation of side chain conformational ensemble through rotamers.  In testing, -ex2 improved the scores, but it improved both native-like and false positive scores, making discrimination harder.  -ex34 = -ex3 and -ex4.


  1. Would the scores between a protein and mutants of its interacting partner be meaningful to compare?  We have a bunch of mutants of a pair and measured affinities.  I remember a paper by T. Kortemme & D Baker in PNAS, in which they used a very similar energy function to find out binding energy hot spot.  Basically I also would like to do that for our sets of mutants. 


    1. You could compare the RosettaDock scores, but the dock mode has not really been tested or tuned to give information about mutants.  It is probably qualitatively ok, but I don't know how big the error bars are or how large an energy is significant.

      You can use Kortemme's technique directly through the Robetta interface:  It is designed to analyze interface mutants.

      We are currently working on projects to integrate these modes better.


Last update: 22 March 2006