[gmx-users] rationale behind tcoupling ligand with protein instead of with SOL

[chris.neale at utoronto.ca]

You didn't state your usage, but if you're doing US or decoupling, etc  
(some method where you lose the dynamics anyway) I suggest that you  
use Langevin dynamics. You will get the correct ensemble. Separate  
temperature coupling groups is a trick that helps in some cases, but  
it still does not give you the correct ensemble.

Chris.

-- original message --

Any rationale behind the thermostat coupling of a ligand with the protein
instead of the ligand with the solvent (as shown in Justin's T4 Lysozyme
binding example)? Especially with small drug-type molecules as generally the
ligand might/would take the usual place of solvent within a binding region
or other solvent accessible surface and it might be more realistic to
simulate the ligand as part of the solvent's ensemble (one might run two
simulations in order to compare the ligand-protein interaction to the
water-protein interaction at the interaction interface...)