ADL: Vina and multiple receptor conformations

Oleg Trott trott at scripps.edu
Thu Apr 9 11:25:10 PDT 2009


On Thu, Apr 9, 2009 at 9:50 AM, Ciaran Carolan
<ciaran.carolan at embl-hamburg.de> wrote:
> Hi all,
> While it is extremely convenient and rapid to use vina for molecular docking
> on account of its calculation of docking grids automatically (ie without the
> use of autogrid as for autodock), sometimes I would find it useful to have
> access to the grid maps that are used. For example, I have used weighted
> average maps to calculate average affinities for different ligand atoms
> across a series of receptor conformations.
> In the video tutorial, it is stated that vina uses grid maps that it
> calculates itself at the beginning of the runs. I wonder can vina be fed
> pre-calculated grid maps (for example, calculated with autogrid) that have
> been modified according to the users needs?
>
> Any help would be greatly appreciated.
>
> Kind regards,
>
> Ciaran Carolan
>

Ciaran,

A "grid" is a computational trick used by AutoDock, Vina and other
docking programs to speed up the calculation of the scoring function
at some expense of memory and accuracy. It uses trilinear
interpolation between the precalculated values.

In Vina, this fact is mostly hidden, because firstly, Vina is not
always using grids. Sometimes it "switches the grids off" and does the
calculation directly. Secondly, I believe grids and trilinear
interpolation are not a good abstraction for the user to be dealing
with. AD4 requires N+2 grids, where N is the number of AD4 atom types
in the ligand and flexible residues. Vina uses a different atom typing
scheme internally (more about this in the paper, when it comes out).
Some future version might have a yet another atom typing scheme and
scoring function, or not use trilinear interpolation at all. The users
would not suffer because they don't deal with the grids directly. [1]

A better software design might be if Vina read multiple receptor
structures and docked to an "averaged" receptor. We might look into
this in the future.

Another approach you could use is the "relaxed complex" method[2],
where one simply docks the ligand to multiple rigid receptors
conformations and then analyzes the results.

Best of luck,
Oleg

[1] http://en.wikipedia.org/wiki/Information_hiding
[2] Lin JH, Perryman AL, Schames JR, McCammon JA. J Am Chem Soc. 2002
May 22;124(20):5632-3.


-- 
Oleg Trott, Ph.D. (Columbia University)

http://olegtrott.com


More information about the autodock mailing list