[Dock-fans] max spheres limit

[Scott Brozell]

Hi,

The DOCK 5 test set used an average of 101 spheres and a
box of size 20000 A^3 (NOT 20A^3 as indicated on page 606).
http://dock.compbio.ucsf.edu/DOCK_6/dock6_manual.htm#Moustakasetal2006
These are larger than for usual docking calcs.

The DOCK 6 test set used an average of 123 spheres and a
box of size 40000 A^3.

Scott

On Tue, 29 Jan 2008, Scott Brozell wrote:

> On Mon, 28 Jan 2008, Ben Keshet wrote:
> 
> > Sudipto recently commented about the limitation of the max number of spheres
> > being 50-75.  I have found a similar comment about dock5.3:
> > http://shoichetlab.compbio.ucsf.edu/pipermail/dock-fans/2005-December/000508
> > .html
> > 
> > Can anyone support or provide any more information about the limitation of
> > Dock with large clusters? Is Dock6.1 any better than 5.3 with regard to
> > that? Is the limit different for rigid and flexible docking? Is there any
> > intelligent way to overcome that limit?  
> 
> Yes, DOCK has practical limitations due to available hardware resources.
> A series of benchmarks would be useful; I'll see what I can find.
> DOCK 6 is much better than DOCK 5 with respect to memory handling
> and consumption because of Terry Lang's work.
> Flexible docking is more memory intensive than rigid:
> http://shoichetlab.compbio.ucsf.edu/pipermail/dock-fans/2006-August/000679.html
> All approaches for circumventing these kinds of limitations are
> based on divide and conquer, and the various common ways of
> getting smaller pieces have been mentioned in this list.
> 
> > 75 spheres are often too small for my ligands.  I could manually edit the
> > SPH file and delete less likely spheres, but I would prefer avoiding such
> > speculations.  
> 
> For large ligands note that the anchor-and-grow
> conformational search algorithm has been validated
> for binding mode prediction on sets of ligands that have
> no more than seven rotatable bonds:
> http://dock.compbio.ucsf.edu/DOCK_6/dock6_manual.htm#Moustakasetal2006
> 
> Scott
> 
> > Date: Fri, 28 Dec 2007 23:17:50 -0800 (PST)
> > From: Sudipto Mukherjee <sudmukh at yahoo.com>
> > Subject: Re: [Dock-fans] bad allocation
> > 
> > Dear Barbara,
> > 
> > Docking works well with about 50-75
> > spheres in the binding site. Could you specify how many spheres you are
> > using? Are all roughly in the binding site? Selecting more spheres
> > would increase the box size for the grid. The size the grid file
> > increases roughly as the cube of the box size i.e. proportional to the
> > volume of the box.
> > 
> > I have tested DOCK successfully with grid
> > files as large as 800MB on a computer with 2GB RAM . However, if the
> > grid file is larger than the amount of available memory, memory
> > allocation errors like the one described is likely.