[gmx-users] Separate lambdas for repulsive and dispersive free energies?

[Nuo Wang]

Hi all,

I am trying to calculate the hydration/solvation free energy for proteins
using free energy perturbation:
E_hydration = E_vdw + E_elec
E_vdw (WCA decomposition) = E_vdw_repulsive + E_vdw_dispersive

However, unlike small molecules, if I turn off both elec and vdw for a
normal-sized protein, a huge cavity will be created and it will take
unfeasible amount of time to equilibrate the system. For this reason, I
only plan to calculate E_vdw_dispersive and E_elec for my system.

I have read the mailing list posts of implementing the repulsive WCA
potential in Gromacs, but I have not found any discussion that try to
implement the WCA-decomposed potential, i.e. repulsive + dispersive, and
further scaling them separately in free energy calculations.

According the the manual, it seems that it is impossible to implement the
decomposition with the Gromacs function form g(x) + f(x). On top of this,
vdw is scaled together by one lambda. I wonder if I am right on this and if
there is an alternative way that I am not aware of?

FYI, the WCA decomposed scaling protocol has been implemented in CHARMM:
http://pubs.acs.org/doi/abs/10.1021/jp048502c

Thanks,
Nuo