[Dock-fans] problem selecting spheres

[Scott Brozell]

Hi,

On Sun, 27 Apr 2008, Carlo Zambonelli wrote:

> after successfully completing all the tutorials using L-arabinose 
> binding protein I started working on my first real docking. Everything 
> went smoothly until I reached the point where I have to select spheres. 
> Following the instruction in the Generating Spheres tutorial I start 
> having problems in "STEP 3: Select a subset of spheres to represent the 
> binding site(s) - Option 1". How do I create a selected_spheres.sph 
> file? This is clear with option 2 using sphere_selector (which is what I 

Create a copy of the .sph file from STEP 2;
edit the copy deleting all but the first cluster.

For example,
install/test/sphere_generation/struct.sph
DOCK 3.5 receptor_spheres
cluster     1   number of spheres in cluster    54
   35  26.47111  20.05715  60.11750   2.478  353 0  0
...
  362  24.77408  23.16160  67.02017   3.114   61 0  0
cluster     2   number of spheres in cluster    26
    1  17.23107   5.40138  69.54200   2.620  187 0  0
...

delete starting at the line cluster     2 ...
to the end of the file.


> did in the tutorial), but what should I do if I have no idea where is 
> the binding site on my protein? Since I do not know where is the binding 

Usually one does have clues to the binding site(s): experiment,
visualization, intuition, etc.
Search the archives; there are a number of posts related to binding sites;
e.g.,
http://blur.compbio.ucsf.edu/pipermail/dock-fans/2007-December/001337.html


> site on my molecule, or if there is only one binding site, can I create 
> a selected_spheres.sph file containing all spheres in rec.sph?

Yes, the last cluster contains all the spheres.
So follow the recipe above to select that cluster; e.g.,

...
cluster     0   number of spheres in cluster  1298
    1  17.23107   5.40138  69.54200   2.620  187 0  0
    1  18.01832   6.16253  69.54200   2.310  187 0  0
...


Note that using too many spheres will consume lots of machine
resources.  The fundamental paradigm to produce tractable computations
is divide and conquer: partition the receptor into small enough
pieces and perform multiple dockings.  In addition, the basic approach
to checking for tractable calculations is to start small (perhaps
with scientifically useless calcs) and scale up.

Scott