Hi Anu,
Have you plotted the MM/GBSA calculated energy (kcal/mol) along MD
frames (ns or frame number) to check if the MM/GBSA results are converged?
Did you include the entropy calculation? (very likely not very accurate but
computationally expensive in this case) Do you have experimental results
to compare/optimize the implicit solvent model?
I suggest to run several short single trajectory MD simulations with
random seed ig=-1 and the following MMPBSA.py energy calculation (e.g. six
or more short MD simulations with ig=-1, and do MM(P)GBSA calculation for
each short MD simulation trajectory). We found this "multiple independent
sampling" approach allow the simulations to sample more meaningful phase
space than one long MD simulations (e.g. one 6 ns long MD simulations)
+MM(P)GBSA calculation, and it might avoid the system sensitivity on
different MM(P)GBSA settings such as various GB methods (igb=1,2,5,7,8) .
Here <
http://onlinelibrary.wiley.com/doi/10.1002/jcc.24011/abstract> is
the link to the article: J. Comput. Chem. 2015, 36, 1859–1873.
3. If one long simulation is still your choice, using MM-PBSA or
MM-GBSA and igb = 7 or 8 might be a better choice for protein-protein/
peptide-peptide interactions, as discussed in this article (H. Nguyen, D.
R. Roe, C. Simmerling, J. Chem. Theory Comput. 2013, 9,2020.) and our
article above. However, like Dr. Luo said, the results might be system
dependent, so it will be interesting to see you results!
Btw, how accurate do you need to quantify the differences between with
and without domain B?
How big is the domain B? If the domain B is only a few peptides,
perhaps try the explicit solvent methods such as TI, FEP, which are
computationally expensive but can quantify free energy differences more
accurately? (though the TI/FEP results might hardly converge even with only
a few peptides)
HTH,
Pin-Chih
On Fri, Oct 23, 2015 at 2:02 PM, Carlos Simmerling <
carlos.simmerling.gmail.com> wrote:
> It might seem like a strange opinion from a GB model developer, buy I don't
> usually recommend GB for binding affinity post-processing. We use GB where
> speed is crucial, such as in md. For post-processing you're not doing it
> for each time step, so I prefer using a better PB. GB just isn't as good as
> PB, and if you have the choice then why use something less accurate? It's
> not likely that the speed of the PB calculation will be the limiting favor
> in this project. For md, it definitely is so we use GB there. GB also tends
> to be not as well trained for chemical diversity.
>
> Others may have different opinions, this is just mine...
> Carlos
> On Oct 23, 2015 10:02 AM, "anu chandra" <anu80125.gmail.com> wrote:
>
> > Dear Amber users,
> >
> > I am working with a protein-ligand system, where the protein comprise of
> > two domains, say A and B and the ligand bind to domain A. I am trying to
> > calculate the free energy of ligand binding in presence and absence of
> > domain B ( I have carried two different simulations, with and without
> > domain B).
> >
> > With igb=2, I got binding free energy of 85.6 without domain B and
> > 91.7 with domain B, and difference is ~6 kcal/mol. With igb=5, the
> binding
> > free energy is 110.5 without domain B and 121.6 with domain B, and
> > difference is ~11 kcal/mol. Which values are more reliable, with igb=2 or
> > with igb=5? Will it be possible to consider the difference of 6 or 11 (
> > either of it depends to the chosen igb value) as a measurement of
> influence
> > of domain B on ligand binding?
> >
> >
> > Many thanks
> > Anu
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
> >
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
Pin-Chih Su (Henry Su)
Ph.D.
Center for Pharmaceutical Biotechnology (MC 870)
College of Pharmacy, University of Illinois at Chicago
900 South Ashland Avenue, Room 1052
Chicago, IL 60607-7173
office 312-996-5388
fax 312-413-9303
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Oct 24 2015 - 13:30:03 PDT
