Re: [gmx-users] Free energy calculation problem with too high dG

[Hannes Loeffler]

On Tue, 4 Oct 2016 15:27:25 +0300
Vlad  wrote:

> I tried to do the decoupling and strangely the result is not the same
> Below is the end of the log
> The total dG is significantly less then previous and the last dG
> value is particularly small I think total dG will get less with more
> precise lambda sampling (by 0.05 instead of 0.1) The thing which
> still is strange is the value of entropy for the last lambda (in the
> previous case it was so high as well for the first lambda)
> 
> So you suggest to use restraints
> Througout the simulation or also gradually switched off?

I suggest that you have a careful look at that web page and related
literature to understand how this type of simulation is supposed to
work. As per my previous explanation I do not see anything particularly
strange at this point.  Making a molecule disappear is just so much
simpler than making it appear.  In the latter case you may get away
with (much) longer equilibration (did you actually do prior
minimisation?).  You may also disrupt your system quite severely esp. in
the case of a protein environment.  Maybe a proper choice of positional
restraints may help.  But I am nor sure if going this route is really
worthwhile except for better understanding the method.  You will
however not get around a careful analysis of your simulations.


Cheers,
Hannes.
-- 
Gromacs Users mailing list

* Please search the archive at 
http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting!

* Can't post? Read http://www.gromacs.org/Support/Mailing_Lists

* For (un)subscribe requests visit
https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a 
mail to gmx-users-requ...@gromacs.org.

Re: [gmx-users] Free energy calculation problem with too high dG

[Hannes Loeffler]

On Tue, 4 Oct 2016 14:01:52 +0300
Vlad  wrote:
> No reason. I just thought it would give the same result.  

In theory both directions should be symmetrical and give the same
result.


> In the trajectory the ligand diffuse to the end of the cell like a
> ghost through the water molecules. And could the reason lie in the
> fact that in the initial structure during all preparations the ligand
> is "fully existing" and in the first step of simulation suddenly is
> turned off?  


I would suggest to have a look through
http://www.alchemistry.org/wiki/Example:_Absolute_Binding_Affinity and
other material on that web page.  In particular reference 1 on that
page should be of interest to you in how to use restraints within a
general scheme.

In a homogeneous environment like water it doesn't matter that your
"ghost" molecule drifts because on average the interactions with its
environment will be the same regardless of location.  In an
inhomogeneous environment, however, the nearly non-interacting ligand
will have different probabilities of where it is located.  In
principle, you would have to sample _all_ possible locations which in
practice is hard or impossible.  Hence the restraint scheme as
suggested above.

Your ligand should exists as a bonded network throughout the
simulation.  What you are doing however is "decoupling" which means
that the non-bonded interactions between the ligand and its environment
and vice versa a gradually switched off.


Cheers,
Hannes.
-- 
Gromacs Users mailing list

* Please search the archive at 
http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting!

* Can't post? Read http://www.gromacs.org/Support/Mailing_Lists

* For (un)subscribe requests visit
https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a 
mail to gmx-users-requ...@gromacs.org.