[gmx-users] Gromacs Relative Free Energy Calculation Issue

[Hannes Loeffler]

Hi,

you could have a look into http://www.hecbiosim.ac.uk/fesetup to set up
your system.

It is late evening here so I will answer in more detail tomorrow.

Cheers,
Hannes.


On Mon, 3 Oct 2016 19:57:07 +0000
Guanglin Kuang <guanglin at kth.se> wrote:

> Dear Gromacs users,
> 
> Has any of you managed to use Gromacs to do relative free energy
> calculations?  I have some technical questions that would need your
> suggestions.
> 
> I am trying to reproduce the Amber free energy calculation using
> PMEMD (http://ambermd.org/tutorials/advanced/tutorial9/#overview).
> 
> First, I generated the topology and coordinate files of the ligands
> (benzene and phenol) using alchemistry_setup.py, developed by Dr.
> David Mobley (https://github.com/MobleyLab/alchemical-setup), as well
> as antechamber with AM1/BCC charge.
> 
> Then I set up the mdp file and topology file for relative free energy
> calculations. I used the single topology together with the one-step
> and three-step transformations, respectively. Please have a look at
> the relevant files  attached for the details.
> 
> Finally, I used g_bar to calculate the free energy
> (alchemistry_analysis.py can give similar results). From the results
> I found that one-step transformation can give good results, only if
> some restraints are applied to the protein-ligand complex, because I
> found that benzene/phenol are not strong binders to lyszome, and
> would drift around in the binding pocket, thus give poor results
> deviating from experimental data. However, in the three-step
> strategy, the results are much worse, this is probably due to the
> fact that a dual topology is used in Amber but a single topology is
> used in Gromacs (my case), we may not be able to do the changes
> directly (as shown in the attached topology files).
> 
> Even though one-step transformation seems to give a better result in
> this case, it may be due to luck, because it is suggested that
> one-step transformation is usually not as good as three-step
> transformation, and can even give misleading results in some
> occasions (what occasions?).
> 
> Therefore, could any experienced users of relative free energy
> calculation have a look at my set-up and point out the problems? I
> would appreciate any improvement to the set-up!
> 
> Thank you!
> 
> -----------------------------------------------------------------------------------------------------------------------------------------------------------
> Best Regards!
> Guanglin Kuang
> KTH-Royal Institute of Technology
> School of Biotechnology
> Division of Theoretical Chemistry & Biology
>