From: Kenno Vanommeslaeghe (kvanomme_at_rx.umaryland.edu)
Date: Tue Sep 24 2013 - 11:19:21 CDT
On 09/20/2013 01:39 PM, Francesco Pietra wrote:
> Do you have any evidence that T-remd simulations in explicit water are so
> much reliable?
In my understanding, reliability is not the issue, it's the number of
replicas needed and the sampling in function of CPU expenditure that
becomes pathological.
> If so, I could perhaps try to compare GB T-remd with explicit
> water T-remd for one of my peptides (should built-in remd with namd-2.10
> will prove substantially faster than tcl-driven remd with namd-2.9).
I would more think in the lines of clustering the results of implicit
solvent T-REMD, picking a few relevant structures, and running ordinary
explicit solvent MD for a few 10s of ns on those to see how they behave.
> It
> would be a scientifically sound approach, hopefully lending credibility to
> the protein I am attempting to complete after partial failure of X-ray
> diffraction (NMR proved much too complex).
If loops or terminal strands of a protein don't show up on X-ray, that
typically means they're disordered (to use a photography analogy, they're
moving too much to capture in a long-exposure picture). If so, that's not
a failure but reality; it is not reasonable to expect a single
conformation of a disordered region to come out of any experimental or
simulation technique. What we commonly do in such cases is run some loop
modeling software (like Modeller), get one or more representative
conformations, and simultaneously run MD simulations on those
conformations. One could presumably do the same using T-REMD instead of
loop modeling.
Finally, note that the disadvantages of implicit solvent for the purpose
of running T-REMD are exactly the same as the ones for normal MD, and a
large body of literature exists on that subject.
This archive was generated by hypermail 2.1.6 : Tue Dec 31 2013 - 23:23:46 CST