Re: NAMD is too slow?!

From: Peter Freddolino (petefred_at_ks.uiuc.edu)
Date: Sun Mar 29 2009 - 22:37:16 CDT

Hi Dmitry,

DimitryASuplatov wrote:
> To my understanding using PME is sufficient with periodic systems -
> like lipid membranes - and could overstabilize free energy when used
> with proteins.
>

There is some evidence that PME can lead to small overstabilization of
certain conformations, although there are certainly ways to mitigate the
risks (see J. Phys. Chem. B, 2004, 108, 774-788) -- note that this isn't
really PME's problem, but the artificial periodicity of one's system
combined with long range forces. However, it is also well known that
long range electrostatics are very important to the behavior of proteins
and other biomolecules, and Ewald methods are generally the best we can
do currently to treat them (one generally wants to simulate a periodic
system anyway to avoid edge effects). So while using PME or another long
range electrostatics method has some possible artifacts associated with
it, *not* using PME means you're likely to be using a suboptimal
representation of the system itself.

Also, if you do want to do simulations without PME, you should be aware
that if you're using shifted electrostatics (see
http://www.ks.uiuc.edu/Research/namd/2.7b1/ug/node24.html), then the
short range electrostatic interactions will also be altered, which is
likely to severely affect the behavior of the water in your system.

Best,
Peter

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:50:41 CST