[Dock-fans] Critical points docking

[Dmitry Osolodkin]

Scott Brozell wrote:
> Yes the sphere file format is correct according to
> http://dock.compbio.ucsf.edu/DOCK_6/dock6_manual.htm#SphgenCriticalPoints
> Yes I see no reason to rebuild grids.
>
> Dock 6.2 contains code to read and check critical points.
> It does not appear to be obviously wrong.
> However, there are no tests or examples that use critical points.
> I'm contacting other developers for the status of critical points.
>
> In the mean time you could try the alternative mentioned in
> http://dock.compbio.ucsf.edu/DOCK_6/dock6_manual.htm#CriticalPoints
> to determine whether your system is conducive to critical points.
>   
Okay, I have deleted some critical points which seem unreasonable to use 
for docking. Nevertheless, I haven't rebuilt the grids, so in the 
docking run I have used a rather big box which only had spheres in one 
corner. The best-scored orientation of the ligand was rather reasonable, 
but the minimal distance between my ligand and critical points was about 
6 Å. This distance seems unreasonable. See 
http://qsar.chem.msu.ru/~dmitry_o/cp.png (spheres are shown as balls and 
transparent spacefill, critical spheres are orange, and the best scored 
orientation is ball-and-stick (a part of it is hidden behind the spheres)).
Moreover, the primary conformations are positioned in the different 
places of the box, and this positions are not directed by spheres. As 
far as I understand the DOCK concept, it should suggest only 
orientations defined by spheres, so if there is an empty box region 
without the spheres and receptor atoms, there shouldn't be any ligand 
atoms in this region. But in my case there are a lot of ligand 
orientations which are not intersected with any sphere. Maybe the reason 
is that my molecule is small and no anchor is defined during docking?
Third, after activation of the chemical matching feature on the same 
four spheres, there were no difference between results obtained with 
critical points, without critical points and with chemical matching. Is 
it an algorithm problem?

Dmitry Osolodkin.