ADL: autodock Digest, Vol 58, Issue 7

Edelmiro Moman edelmiro.moman at gmail.com
Mon Jun 8 02:22:25 PDT 2009


Hi,

>From some of our results with flexible side-chains, I have the impression
that a ligand hydroxyl group is labeled only as H-bond donor, but not as an
acceptor.

For instance, all other docking programs we have used so far, do not have
any problem for identifying the following H-bond network (even if the
involved residues are flexible):

                    LIGAND
                          |
                       HO
                        /   \
TYROSINE-OH     HO-SERINE

Vina does not (based on the O-O distances, not on the OH dihedrals).

However, Vina can identify this one:

                    LIGAND
                         =
                         O
                        /   \
TYROSINE-OH     HO-SERINE

So, yes if it is a carbonyl, no if it is a hydroxyl group. Empirically, both
should be efficient.

Can this be considered a "bug"? Is there a way to manually label the atoms
as H-bond donor or acceptors or both?

Kind regards,

Miro



>
>
> ------------------------------
>
> Message: 2
> Date: Sun, 07 Jun 2009 13:00:47 -0500
> From: Mark Swingle <mswingle at jaguar1.usouthal.edu>
> Subject: Re: ADL: Vina: Wrong H-bond networks
> To: autodock at scripps.edu
> Message-ID: <200906071300.47267.mswingle at jaguar1.usouthal.edu>
> Content-Type: text/plain; charset=iso-8859-1
>
> Miro,
>
> As I recall from the Vina documentation, polar hydrogens are only used to
> type
> the heavy atoms they are bound to as either h-bond donors or h-bond
> acceptors. As far as Vina's scoring function is concerned, degrees of
> freedom
> that rotate hydroxyl groups are degenerate (i.e. all torsion angles are
> equivalent) and don't need to be set up as rotatable groups in the input
> files. Therefore, the torsion angles of such groups are basically random
> and
> meaningless in the output structures. However, Vina should be able to
> correctly position the hydroxyl oxygen. Your description suggests that is
> the
> case, yes? If so, I think your results are OK.
>
> Regards,
>
> Mark
>
> On Sunday 07 June 2009 05:52:18 am Edelmiro Moman wrote:
> > Hi,
> >
> > I have been trying to reproduce some crystallographic complexes with
> > Autodock Vina. In terms of RMS, the results are very good. In terms of
> > scoring, they are not so good.
> >
> > In a couple of cases, the hydrogen bond networks as identified by Vina
> are
> > obviously wrong. For instance, in one example with not flexibility in the
> > receptor and just one rotatable bond in the ligand (the one corresponding
> > to the dihedral angle of a hydroxyl group), every docking program
> > (including Autodock 3 and 4) finds the right solution for the H-bond
> > network:
> >
> >                               LIGAND
> >
> >                                  HO
> >                                 /     \
> >  GLUTAMATE-C(O)-O-      HN-HISTIDINE
> >
> > However, Vina finds this solution:
> >
> >
> >                               LIGAND
> >
> >                                  OH
> >                                       \
> >  GLUTAMATE-C(O)-O-      HN-HISTIDINE
> >
> >
> > So that the H of the ligand is pointing towards the H of the histidine.
> >
> > I wonder how is this possible. Is Vina reading the polar hydrogens from
> the
> > pdbqt file or is it trying considering different protonation states?
> >
> > Kind regards,
> >
> > Miro
>
>
>
>
> ------------------------------
>
> ________________________________________________
> --- ADL: AutoDock List  ---
> http://www.scripps.edu/pub/olson-web/doc/autodock/ ---
>
> End of autodock Digest, Vol 58, Issue 7
> ***************************************
>



-- 
Edelmiro Moman, PhD
Pharmaceutical and Medicinal Chemistry
Saarland University
D-66041 Saarbrücken
Germany

Phone: +49 (0)681 302 70320 or 70328
Fax: +49 (0)681 302 4386

E-mail: edelmiro.moman at gmail.com
E-mail: e.moman at mx.uni-saarland.de

http://www.edelmiromoman.eu/


More information about the autodock mailing list