ADL: parameter optimization of vina scoring function

Oleg Trott trott at scripps.edu
Thu Jan 16 14:05:51 PST 2014


Hi

We plan to but don't currently distribute any software that can
automatically tune the parameters for you. However, the users could
try do it themselves. Parameter tuning is a kind of machine learning,
and so it requires some understanding of statistics.

Tuning the parameters for just one ligand probably doesn't make sense.
If you have a set of systems of a certain class, then you could try
tuning the parameters for it, and might be able make conclusions about
the suitability of the new scoring function for systems of the same
class.




On Thu, Jan 16, 2014 at 2:40 AM, meginir urestal
<meginir.urestal at gmail.com> wrote:
> Hello all,
> I'm currently docking a series of nucleotides in a certain protein using
> Vina. In order to enhance the results (number of good poses -compared to a
> crystal structure- and their relative positions in a given docking result)
> I'm currently tweaking the scoring function through the hydrogen bond
> weight parameter.
> Using the same protein and ligand, I realized several docking runs with
> everything identical except for the hbond weight (and using the same random
> seed for comparability). I then compared the results between these
> different runs. However, I'm having trouble getting something out of this
> modus operandi. In fact several questions have come up while considering
> this:
>
>       - is parameter optimization a useful thing to do?
>       - Is this a good way to optimize parameters?
>       - is parameter optimization valid only for the ligand it has been
> done for, or can it be extrapolated to a family of similar ligands or
> perhaps all ligands for a given receptor? A priori, the answer is no, but
> I'd like a different opinion.
>       - Is parameter optimization in the way that I perceive it (changing
> only one parameter at a time) any good considering that a scoring function/
> energy function is multidimentional?
>
> Thanks,
> meginir
> ________________________________________________
> --- ADL: AutoDock List  --- http://autodock.scripps.edu/mailing_list ---



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

Staff Scientist in the Olson Lab
The Scripps Research Institute

http://olegtrott.com


More information about the autodock mailing list