Hi Emanuel,

check this guide out. It explains how to remove the contribution of the restraint in your result.

http://www.alchemistry.org/wiki/Example:_Absolute_Binding_Affinity

What you ask is likely to be system specific. In principle it should be possible to dig out equilibrium free energies from geometrically restrained simulations. You may wish to check the following paper, where they kept water molecules inside a protein cavity by positional restraints, and disappear them eventually. http://dx.doi.org/10.1021/ja0377908

http://dx.doi.org/10.1021/ja0377908

i think for a truely rigorous calculation of free energies with constraints you need sometyhing like this ... http://pubs.acs.org/doi/abs/10.1021/jp0217839 ... (McCammons approach referenced by Michal might work for small, rigid, and close to spherical molecules like water, but to me it is not clear how rigorous this method is if you are dealing with anything more complex  than that)

You only need these corrections though if you are interested in ABSOLUTE free energies, if you want to calculate a DeltaDeltaG ... e.g. the difference in binding free energies between two different ligands in a binding site you should get away doing a simple alchemical mutation ... to give you a more precise answer it'd be helpful to know what type of ystem you consider here ...

Michael

Find attached paper it helps you to calculate binding free energy for macromolecules.

Thanks for your replies. Now, as far as I understand it, restraints or constraints affect the system, but I still get e.g. binding free energies, if I am considering everything based on thermodynamic cycles, and I will obtain the finite differences in the free energy when the same constraints are applied in all states of the thermodynamic cycle. 

I think that sounds consistent to me. Maybe probabilistic approaches combined with enhanced sampling are more rigorous, with F = -kT ln P, or the PMF = -kT ln g(r). But that's another discussion. Thanks again.