Different in Binding energy, mmpbsa and gmx energy

[bta...@googlemail.com]

Hi All,

I am new to use this software. I have generated the energy_MM under this command: 

 g_mmpbsa -f traj.xtc -s topol.tpr -n index.ndx -mme -mm energy_MM.xvg -decomp -mmcon contrib_MM.dat -pdie 2

However the total binding energy is really high: 

@ s0 legend "chA VdW Energy"

@ s1 legend "chA Elec. Energy"

@ s2 legend "chB VdW Energy"

@ s3 legend "chB Elec. Energy"

@ s4 legend "chA-chB VdW Energy"

@ s5 legend "chA-chB Elec. Energy"

@ s6 legend "chA-chB Total Energy"

          0.000          0.000          0.000          0.000          0.000       -351.811      -1498.856      -1850.667

         30.000          0.000          0.000          0.000          0.000       -300.090      -1366.446      -1666.536

         60.000          0.000          0.000          0.000          0.000       -325.273      -1372.878      -1698.151

         90.000          0.000          0.000          0.000          0.000       -334.094      -1435.629      -1769.724

However, if I used gromacs gmx energy, it gives me energy much lower and sensible:

Coul = -357.213
LJ = -136.393

Hence can someone explain to me why do I have such significant difference?

Best,

Ben

[g_mmpbsa Mailing List]

Hi,

When gmx energy is used, it does not calculate energy, it only extract energy from edr file. In the edr file, energy is calculated according the simulations setting provided in mdp input file. Therefore, it has separate short-range and long range terms for electrostatic and van der Waals energies. Moreover, if cut-off is used, long-range electrostatic is calculated using PME while there is no long-range van der Waals energy (depending on the settings in simulation mdp file). Also, electrostatic energy is calculated with a dielectric constant 1.  However, g_mmpbsa does not use any cut-off and calculate energy for all atom-pairs. Also, by using pdie 2, you changes the dielectric constant to 2, which in turn changes the electrostatic energy. We had reproduced and matched energy values from gmx_energy and g_mmpbsa, however it requires to set-up a mdp files with different settings than the settings used for MD simulations.

Thanks

[bta...@googlemail.com]

Hi,

Thank you very much for your insight. For a protein-protein binding, which dielectric value should I use? as the original paper only mentioned the dielectric for measuring solvent is 2 to 4. Or do I need to find out by changing the constant until the energy is constant? 

Best,

Ben