Protein-protein binging energy not realistic

[Bernard Hehlen]

Hello,

When I run gmx_MMPBSA in my protein-protein complex in water solvent, I find Delta_H ~=-130 Kcal/mol which seems quite unrealisic as compared to estimation of ~-30 to -40 Kcal/mol.  I played with parameters such as indi, surf_ten, and with different models, gb5, gb_IE, and the closests values are -60 Kcal/mol for  indi=1, surf_ten=0.0005 which look like extreme parameter values.  My data are very long NPT trajectories (1mus) obtained with Andersen thermostat.

I am wondering 

1/ is it realistic to reduce further indi (<1??) and surf_ten, and are there other parameters that could be changed? 

2/ are there some check I could do to  fix this Delta_H problem? 

Thank you in advance for your help.

Bernard

[marioe911116]

MMPB(GB)SA method tends to overestimate the energy values; that's why it is better in relative comparisons context... indi < 1 is not recommended, nor is changing the surften parameters unless you have a valid justification.