[Dock-fans] Anchor size and clashes elimination

[Scott Brozell]

Hi,

Here is an improved description of clash_overlap.
http://dock.compbio.ucsf.edu/DOCK_6/dock6_manual.htm#TimeRequirements

* use_clash_overlap [no] (yes, no):
  #Flag to check for overlapping atomic volumes during anchor and grow
      o clash_overlap [0.5] (float):
        #The relative threshold for overlapping atomic volumes;
        #a clash exists if the distance between a pair of atoms is less than
        #the clash_overlap times the sum of their atom type radii;
        #thus, a clash_overlap of 0.75 allows 25% (1 - 0.75) of relative overlap


Regarding worse docking with clash filtering than without,
the usual explanation is poor radii.

Scott


On Tue, 7 Aug 2007, Sudipto Mukherjee wrote:

Min Anchor size:
The default size of 40 atoms will cause Dock to select the largest anchor only for most molecules. Using a smaller anchor size rapidly increases the execution time though. min_anchor_size of 6  means the anchor can be as small as a single benzene ring. This parameter is somewhat analogous to the multiple_anchors parameter present in DOCK 4. Using multiple anchors can improve sampling in some cases, but mostly results in sampling similar conformers multiple times. Also, I would recommend using clustering when using a small min_anchor_size to prune the number of similar conformers sampled.

Clash_overlap:
DOCK computes clash overlap with the following formula:
if ( distance(atom_A, atom_B) < clash_overlap *( radius_atom_A + radius_atom_B ) ) assume there is a clash

See the segment_clash_check() function in conf_gen.ccp for details.

Regards
Sudipto Mukherjee
Graduate Student, Robert C. Rizzo Lab
Dept. of Applied Math & Statistics, Stony Brook University


----- Original Message ----
From: Grzegorz Popowicz <popowicz at biochem.mpg.de>
To: dock-fans at docking.org
Sent: Monday, August 6, 2007 5:39:21 AM
Subject: [Dock-fans] Anchor size and clashes elimination

Hi all!
1. The default size of an anchor in DOCK is 40 atoms this generates quite  big
fragments as for beginning of docking (at least 480Da). I have tried smaller
anchors (8 atoms) with good results. Are there any rules governing selection
of anchor size?
2. Dock run with use_clash_overlap=no generates solutions with terrible
clashes (of course), however the predictions of known ligands binding are
very good in my case. With use_clash_overlap=yes and clash_overlap=0.5 (BTW.
What does percent of overlap mean?) there are no clushes but known ligands
are badly docked and scored much much worse than without clashes filter. Is
there any resonable explanation to this fact?