ADL: Calculation of the free energy of binding using epdb or AutoDock4.1Score

Mark Swingle mswingle at jaguar1.usouthal.edu
Mon Apr 6 09:40:30 PDT 2009


On Monday 06 April 2009 07:48:36 am annalisa bordogna wrote:
> Hi everybody!
>
> I tried to calculate the Free Energy of Binding on a pose generated by
> AutoDock using both the command epdb and AutoDock4.1Score and I found that
> the results were different for the three cases:
> *
> 1) epdb:*
> Estimated Free Energy of Binding    =   -8.42 kcal/mol  [=(1)+(2)+(3)-(4)]
> epdb: USER    (1) Final Intermolecular Energy     =  -20.50 kcal/mol
> epdb: USER        vdW + Hbond + desolv Energy     =  -19.52 kcal/mol
> epdb: USER        Electrostatic Energy            =   -0.98 kcal/mol
> epdb: USER    (2) Final Total Internal Energy     =   +0.00 kcal/mol
> epdb: USER    (3) Torsional Free Energy           =  +12.07 kcal/mol
> epdb: USER    (4) Unbound System's Energy         =   +0.00 kcal/mol
> *
> 2) AutoDock4.1Score:*
>    AutoDock4.1Score     estat      hb            vdw      dsolv    tors
>        -9.2842             -0.9357 -4.0624 -24.7674  7.3561  13.1252
>
> *3) energy calculated during docking:*
> DOCKED: USER    Estimated Free Energy of Binding    =  -14.60 kcal/mol
> [=(1)+(2)+(3)-(4)]
> DOCKED: USER
> DOCKED: USER    (1) Final Intermolecular Energy     =  -23.84 kcal/mol
> DOCKED: USER        vdW + Hbond + desolv Energy     =  -22.94 kcal/mol
> DOCKED: USER        Electrostatic Energy            =   -0.90 kcal/mol
> DOCKED: USER    (2) Final Total Internal Energy     =   -1.72 kcal/mol
> DOCKED: USER    (3) Torsional Free Energy           =   +9.88 kcal/mol
> DOCKED: USER    (4) Unbound System's Energy         =   -1.08 kcal/mol
>
> Does anybody know why?
> Now, I don't know which of the three methods is reliable for such a
> calculation!

Annalisa,

compute_AutoDock41_score.py uses autodock4.1 parameters, which differ from 
autodock4 with regard to the weights on the free energy components 
(electrostatic, vdW etc.) and also uses Unbound System's Energy = Ligand 
Internal Energy of bound conformation (autodock4's default parameter set 
assumes Unbound System's Energy = Internal Energy of the unbound extended 
conformation (determined from an LGA search prior to docking). The python 
script also does not use grid maps like epdb does, so there would be some  
differences due to that as well. I believe you can force 
compute_AutoDock41_score.py  to use autodock 4 parameters by using the -p 
flag and pointing the script to an appropriate parameter file (comes with the 
autodock source code) like so:

compute_AutoDock41_score.py -p AD4_parameters.dat -r receptor.pdbqt -l 
ligand.pdbqt 

OTOH, epdb should give the same result as in your .dlg file (assuming the 
conformations and grid maps are the same) except for the unbound energy, 
which needs to be determined separately. I see that your torsional free 
energies are different which suggests that the TORSDOF values in your input 
files differ by ~8. Still, your Final Intermolecular Energy should be the 
same for epdb and your docked conformation - not sure why they're different.

Regards,

Mark


More information about the autodock mailing list