On Thu, Jun 09, 2011, caobb0214 wrote:
> I'm dealing with simulations of structurally similar ligands (~100
> atoms) bound at the surface of a receptor (~ 2600 atoms). I calculated
> binding free energies with MM-GBSA method and performed entropy
> calculations with Nmode module of Amber 9. GB calculation was applied
> for 1000 snapshots while Nmode calculation was applied for 20 snapshots.
> One ligand remained rigid at the binding interface, and the absolute
> binding free energy was close to the experimental value (less than 0.5
> kcal/mol). Another ligand fluctuated over the interface and not bound as
> tightly as the former, but its absolute binding free energy was lower
> than the former's and deviated from its experimental value by almost 2
> kcal/mol. Is it because that Nmode calculation failed to give reasonable
> estimation for the latter ligand-receptor complex?
I don't think anyone can answer this question. If you get results within 2
kcal/mol of experiment, you are (probably by chance) doing much better than
average, or that you have any reason to expect. You could use more snapshots
to see if the entropy results are likely to be converged, but the general
errors of the mm-gbsa method are larger than those cited above.
....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Jun 09 2011 - 02:00:04 PDT
