ADL: Docked binding energies - how accurate are they?

Mark Swingle mswingle at
Fri Nov 4 16:54:20 PDT 2011

A 200-fold difference between estimated and experimental affinities is not 
unexpected. Current docking software is reasonably good at predicting binding 
modes of ligands but is much less successful at predicting affinity. Errors of 
even three orders of magnitude or more can be seen.

Most scoring functions display a bias toward higher MW ligands, although 
Autodock and Vina do try to correct for this.

One should also keep in mind that Km does not necessarily correlate well with 
Kd - depends upon the mechanism and the values of the various rate constants.

Another thing to consider is the ionization states of receptor residues. Was 
the pH of the Km determination mimicked reasonably well by your choice of 
ionization states?

On Thursday, November 03, 2011 07:59:41 pm Meg wrote:
> Hi all,
> I have a question about the binding energies that come out of autodock
> vina.
> I have a sugar-cleaving enzyme, and I have docked a known inhibitor
> (monosaccharide analog) and a trisaccharide (the substrate) to its active
> site.  I made all non-redundant bonds rotatable.  I am getting the
> following results for binding energies:
> Inhibitor: -5.5 kcal/mol (equates to 130 micromolar affinity)
> --- The wetlab-determined Ki is 50 micromolar  (so, within a similar range)
> Trisaccharide substrate: -7.5 kcal/mol (equates to 5 micromolar affinity)
> --- However, the wetlab-determined Km is 1 millimolar. (3 orders of
> magnitude away)
> I'm calculating the affinity using Gibbs' Free energy equation, as AutoDock
> does.
> The fact that the trisaccharide is a substrate, and therefore I'm expecting
> to see some conformational change upon binding, may relate.  I was
> surprised though that the difference (micromolar to millimolar) was so
> big.  Could the size of the ligand relate to the accuracy of the binding
> calculation?  Have any other people seen this?
> Does anyone have insight onto what I can make of this?
> Thanks in advance,
> Megan
> ________________________________________________
> --- ADL: AutoDock List  --- ---

More information about the autodock mailing list