[gmx-users] Relative Binding Free Energy Calculation

[David Mobley]

Bob,

> When performing a relative binding free energy calculation, I
> understand that it is a good practice to separate the transformation
> of charges of the ligand and the transformation of vdW parameters into
> two calculations. However, is it advisable to completely discharge the
> ligand before transforming vdW parameters? For example, should the
> calculation proceed as:
> 1. Reduce all ligand charges to zero
> 2. Transform vdW parameters for relevant atoms
> 3. Recharge ligand atoms with the new, appropriate values

As far as I am aware there hasn't been a lot of testing here on what's
most efficient.

What I've been doing in some relative free energy studies lately
(which I think is a good idea) is to (1) turn all ligand charges on
atoms which will be inserted/deleted to zero, while simultaneously
changing the charges on any atoms which will undergo charge
modifications; (2) transform vdw parameters for relevant atoms; then
(3) restore ligand charges on any atoms which have been inserted.

Alternatively the modification of charges on changed atoms can be done
in step (3).

> It seems to me that steps 1 and 3 would yield large free energy
> changes, thus making the calculation of free energy differences
> between the bound and unbound states more difficult due to convergence
> issues.

That could be true. On the other hand, my experience with
electrostatic transformations is that it's possible to compute these
quite precisely even if the values are extremely large. That is, I
don't see the magnitude of the uncertainty scaling with the magnitude
of the total. So it might be possible to do this very precisely.
Still, I don't see any reason to change charges on atoms that you
aren't going to be modifying at all.

> Is it more advisable to proceed as:
> 1. Reduce charges to zero on ONLY those ligand atoms that will have
> their vdW parameters transformed
> 2. Transform vdW parameters for relevant atoms
> 3. Transform charges on ligand to their new, appropriate values

Yes, that sounds better.

Where I woudl worry about convergence is not so much in the magnitude
of the charging free energies, but rather that if you're turning off
all the charges on something, the conformational ensembles it will
sample may be quite different from those it would sample with its
charges on. It's probably best to perturb the ensemble as little as
possible unless you have some reason to believe that perturbing the
ensemble will improve your sampling.

David

>
> Proceeding in this way seems like you would obtain smaller free energy
> changes for steps 1 and 3 and could result in better convergence for
> free energy differences. Any suggestions?
>
> Thanks,
> Bob
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at http://www.gromacs.org/search before posting!
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>