You are here

Local docking: "manually place the two proteins facing each other" ?

6 posts / 0 new
Last post
Local docking: "manually place the two proteins facing each other" ?

This is one of the first sentences in the docking tutorial.  I'm new to Rosetta, and to chem software in general, and I just have no idea how to do this.  I'm trying to dock two identical proteins that will eventually build up into a capsid structure.  My first attempt to do this was with Chimera, but Chimera always overlays the two structures and I can't figure out how to move them independently.  I don't know if I should use a different program, but I'm mostly limited to the programs on the computers in my lab.  I really wish this tutorial had better information on this necessary first step, or a link to a relevant Chimera tutorial.  Just when I thought I was about to make some progress...

Post Situation: 
Fri, 2018-07-20 10:24

Having more information in the tuorial probably wouldn't have helped you, as they probably would have written it with PyMol and not Chimera.

I believe there's a number of ways to do this in Chimera, but I might recommend using Tools ->"Structure Editing"- >"Movement Mouse Mode". This will pop up a dialog box which allows you to change how the mouse behaves. You can choose "Move molecule" or "Move chain" from the dropdown, and then you can left-click-drag to rotate the molecule/chain, and middle-click-drag to translate the protein into the appropriate location.  (Click-dragging on the background will do the standard view changes.

If those controls aren't working for you (e.g. if you don't have a middle button) you can change how the mouse buttons behave with Favorites->Preferences and select "Mouse" from the dropdown.

Fri, 2018-07-20 11:39

Thank you for the reply, I have now gotten the two protein chains to face each other and have moved them close together.  However, when I run the docking protocol, they do not actually dock together.  The output pdb shows the two chains in essentially the same position.  The energy score isn't bad, but I don't think it reflects the interaction of the two chains, unless the output pdb simply isn't representative of what Rosetta actually scored?

Mon, 2018-07-30 11:57

If the docking protocol is behaving appropriately, you should get a number of structures, each with slightly different orientations of the two binding partners. (If you're only getting a single output structure, I'd definitely recommend increasing the value passed to -nstruct and checking the variation amoung the multiple outputs.)

How much movement you'll get depends on the settings you put in the protocol. You'll expect less movement from the local refinement protocol than with a global docking protocol.

Have you successfully run the tutorial with the example system provided with the tutorial? If that run is successful, what changes did you make when moving over to your own system? Are there any errors and warnings in the log which might point to some issue with your protein?

Fri, 2018-08-03 07:46

Hi rmoretti,

I was working with the tutorial example, to make certain I could recreate it before trying my own molecule.  The tutorial had set -nstruct to 1 and used local docking.  I don't really know if the tutorial intended to get a good result, or just demonstrate the setup.  It used the following flags:

-in:file:s input_files/col_complex.pdb
-in:file:native input_files/1v74.pdb
-unboundrot input_files/col_complex.pdb

-nstruct 1

-partners A_B
-dock_pert 3 8


-out:path:all output_files
-out:suffix _local_dock

I used my own placement of the two proteins, rather than the provided file.  I spent some time aligning them to match the input file in the tutorial.  The output pdb simply doesn't show a docked pair; possibly slightly rotated, but the two chains are the same distance apart.  I've been using local refine in my actual project (I have a native structure), and it seems to work well, but I've avoided using the local or global docking protocols to try to "recreate" the native structure.

Fri, 2018-08-03 09:42

You did set the -nstruct flag to only 1.


To get a reliable result you should run at least 1,000 decoys.




Tue, 2018-10-16 09:05