[gmx-users] Relative binding free energy

Hannes Loeffler Hannes.Loeffler at stfc.ac.uk
Mon Mar 14 16:02:20 CET 2016

On Mon, 14 Mar 2016 14:08:32 +0000
Stefania Evoli <stefania.evoli at kaust.edu.sa> wrote:

> Please, could you give a look at that? I obtained it by using the
> useful tool described in the article 'A Python tool to set up
> relative free energy calculations in GROMACS¹ Pavel V. Klimovich and
> David L. Mobley. Probably I¹m doing something wrong in the MCSS map.

As far as I know Pavel and David's tool reads in a ready-made MCS and
creates a Gromacs topology from it.  You seem to indicate the same.  I
trust that their tool works so you would now need to make sure that
your MCS map is what you think it should be.  I have indicated earlier
what I think it should look like.

I'm writing a tool my self, FESetup
(http://www.hecbiosim.ac.uk/fesetup), which can also create Gromacs
topologies for relative free energy calculations.  I must say though
that at the moment I do not support both appearing and disappearing
atoms in one for Gromacs (and I do no recommend you to that anyway).
Also, the MCSS is based on rather loose rules so may not give the
expected results in your case. But eventually it is based on the
algorithm in RDKit which is fairly easy to use.

If you need a map only once though it shouldn't be too difficult to do
that by hand as indicated earlier.

> You mentioned that I can simultaneously switch off the charges for
> disappearing groups and switch on for the appearing groups with a
> single lambda path. Could you clarify this point and if it is
> possible suggest me someway to do that, please?

No you can't.  That is exactly why you need _two_ topology files e.g. in
the first do q_off followed by vdw on/off.  The second topology file
takes care of q_on for the appearing group (see previous email for
details and the attachment).  You cant switch the electrostatics in the
same topology because you, crucially, have to do it in that specific
order and Gromacs does not allow you to say that a certain lambda path
only applies to a subset of atoms.

But I anyway do not recommend this for your case.  Make your life
easier and just to a maximal map and you will only have disappearing
atoms (probably best to set up your system to mutate from the larger
to the smaller molecule).


More information about the gromacs.org_gmx-users mailing list