[gmx-users] forward and backward states for the binding free energy energy

[Hannes Loeffler]

On Mon, 2 Jan 2017 14:17:56 +0300
Qasim Pars <qasimpars at gmail.com> wrote:

> 1-) The mean of the forward state is ~15 kcal/mol. That is too big,
> right? Maybe the ligand doesn't get decoupled in the forward state?

It will be hard to guess whether this is meaningful or not...  You
switch with 100 ps and carry out 50 replicates.  Is that maybe to fast
or will you need more repetitions?  Experimentation (with monitoring of
the results an convergence), comparison to equilibrium runs, and
literature search should help.

There are many reasons why your current results may not match
expectations. You don't quite say where the "literature" results come
from.  You also don't say if you have closed the thermodynamical cycle
properly or if you take any measurements regarding the vanishing end
state.


> 2-) Is the free energy control stuff used in the em.mdp, nvt.mdp,
> npt.mdp and md.mdp correct for the forward and backward/reversible
> state simulations? I must use the same free energy control parameters
> in both states except init-lambda?

What you need to do is to prepare a number (equal to the number of
trajectories you plan to run in the fast-switching runs) of equilibrated
starting points for each end state. You have various ways to do that
with Gromacs. If you want to use the same control parameters as in the
free energy runs, as you plan to do it, you can use init-lambda=0.0 with
a positive delta-lambda increment and init-lambda=1.0 with a negative
increment.

 
> 3-) Is the free energy control stuff used for 100 ps simulations of
> the forward and backward/reversible state simulations correct? The
> only difference between both states must be delta-lambda?

Plus init-lambda as in answer to 2.


> 5-) I added the state B parameters to the
> [ intermolecular_interactions ] parts. That should be enough to
> define the state B of a molecule in GROMACS2016?

I think the state B topology will be created automatically with
couple-moltype.  Why do you feel like modifying this by hand?