[gmx-users] Free energy perturbation with hamiltonian replica exchange (FEP - HEX)

[Julian]

I would like to add that the manual is not correct for expanded free energy
calculations (under Gromacs 5.0.6).

The manual states the parameters with the following options:

*lmc-stats*: no, metropolis-transition, barker-transition, wang-landau,
min-variance
*lmc-mc-move*:
no, metropolis-transition, barker-transition, gibbs, metropolized-gibbs

I have been trying it on Gromacs 5.0.6 and the parameters and options
should be named as this:

*lmc-stats*: no, metropolis-transition, barker-transition, wang-landau,
weighted-wang-landau, minvar
*lmc-move*: no, metropolis, barker, gibbs, metropolized-gibbs


2015-09-22 13:25 GMT+02:00 Julian <julimoxx at gmail.com>:
>
> Dear Gromacs Users,
>
> I am running relative binding free energy calculations with free energy
perturbations (FEP) but the lambda points do not converge within 10ns of
simulation time so I am thinking of using hamiltonian replica exchange
(HEX, HREX or H-REMD) with FEP as described in this paper on FEP/H-REMD
using CHARMM.
>
> As I understand, Gromacs offers termperature replica exchange (T-REMD),
which does not work for me as my system is too big and furthermore contains
a membrane, and λ-dynamics in the "free-energy expanded"  module.
>
> On alchemistry.org it states that Expanded Ensemble or λ-dynamics are NOT
recommended for beginners - does any body have any experience with this?
> If so, could anybody recommend me any of the parameters, ... as I cannot
find any example, tutorial, and the manual is very short on it.
>
> An alternative would be the HREX-Plumed plugin by Giovanni Bussi. However
it seems to work only for Gromacs 4.6 (not sure about this) and I would
prefer to use Gromacs 5.0 as Gromacs 4.6 does not allow the LINCS algorithm
on free energy calculations. Furthermore Giovanni Bussi states that he is
not sure whether his HREX plugin works properly with FEP calculations.
>
> This makes me think that maybe I should give it a try with λ-dynamics?
> Any suggestions? Should I try something completely different?
>
> Best regards,
> Julian