Re: [AMBER] mmpbsa discrepancies

From: Jason Swails <jason.swails.gmail.com>
Date: Tue, 22 Mar 2016 21:45:18 -0400

On Mon, Mar 21, 2016 at 1:05 PM, George Tzotzos <gtzotzos.me.com> wrote:

> I’ve been trying to work out the binding energy of a homodimer, with 2
> ligands and 2 water molecules (one per subunit).
>
> I’ve generated two independent 100 ns trajectories.
>
> I calculated the binding energy using the following input script
>
> Input file for running PB and GB
> &general
> strip_mask = :WAT,Na+,&!:271-272
> endframe=10000, verbose=1, interval=20,
> # entropy=1,
> /
> &gb
> igb=2, saltcon=0.100
> /
> &decomp
> idecomp=1,
> dec_verbose=1,
> /
>
> 271 and 272 are the said two water molecules. They do not exchange with
> bulk solvent (occupancies > 85%)
>
> In both cases mmpbsa.py.MPI through WARNINGS
>
> GENERALIZED BORN:
>
> WARNING: INCONSISTENCIES EXIST WITHIN INTERNAL POTENTIAL
> TERMS. THE VALIDITY OF THESE RESULTS ARE HIGHLY QUESTIONABLE
>
> the origin of these warning as it has been discussed previously in the
> list and my understanding is that they are not deemed important for single
> trajectories.
>
> Differences (Complex - Receptor - Ligand)
> DIHED 0.0001 0.0054
> 0.0002
> DIHED 0.0000 0.0056
> 0.0003
> (For BOND and ANGLE) the corresponding differences are zero.
>

​These are *tiny* differences. It wouldn't surprise me if there was some
kind of difference in improper torsion ordering between the bound and
unbound states. That's really the only thing that makes sense here, and I
know that, while uncommon, it *does* happen. I think it's important when
potential problems are flagged and brought to your attention (which is why
MMPBSA.py warns in cases like these), but in this instance I think the
warning is benign.


> BINDING ENERGY RESULTS OBTAINED
>
> For trajectory 1, the binding energy = -54 kcal/mol
> For trajectory 2, the binding energy = -47 kcal/mol
>
> I stripped the water molecules and generated new topologies with
> ante-mmpbsa.py
> I run mmpbsa.py using the same script as above with the only difference
> that the strip_mask was quoted out.
>
> For trajectory 1, the binding energy = -49 kcal/mol
> For trajectory 2, the binding energy = -37 kcal/mol
>
> I don’t think these discrepancies are due to the water molecules, (1)
> because they are not at the binding interface although they do form h-bonds
> with the ligands part of which is at the binding interface and (2) because
> mmpbsa-decomp gives
> for water1 -0.14 kcal/mol (trajectory 1); -0.22 kcal/mol (trajectory 2) &
> for water2 —0.03 kcal/mol (trajectory 1); -0.11 kcal/mol (trajectory 2)
>
> I’d appreciate any suggestions explaining this behaviour.
>

​It's unclear to me what you are defining as your "ligand". Is the water
included in your receptor or ligand topology? An extra water molecule can
make a fairly substantial difference in calculated binding energies using
MM/PBSA or MM/GBSA. Both due to polar and nonpolar solvation
contributions. Particularly if the water molecule is "floating out in
space" in some of the frames. (It's not clear what you mean by "does not
exchange", but only "85% occupancy"...)

At the end of the day, MM-PBSA and MM-GBSA are very simple analyses, which
limits the possible explanations that can exist for the behavior you're
seeing.

HTH,
Jason

-- 
Jason M. Swails
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Mar 22 2016 - 19:00:06 PDT
Custom Search