From: "Yong Duan"
Subject: RE: Ionic solutions
Date: Tue, 13 Feb 2001 15:49:58 -0500


Can't stop wanting to add comment. Tom C. made excellent observations. In my
opinion, the equilibration is the key. But if one wants a really short
simulation (<500ps), well, the best bet is to have a little salt and a few
counterions. They should be placed away from solute. The next step is to get
oneself ready for questions, since no matter what one does, there must be a
few questions (sometimes quite serious ones). I also second Tom's opinion
that if one is interested in the properties closely associated with
counterions, extra care should be taken.

In addition to long equilibration, I'd also suggest that a few runs with
different counterion distributions should be considered. The underlying
issue is really sampling, since this would not be a question had we been
able to sample all conformational space. Given that one can not possibly do
that for any macromolecular system with a few nanosecond simulation, the
initial distribution becomes an issue. If one believes that counterions play
a big role, naturally, a little more effort is well justified. So, here
comes an expensive recipe: multiple simulations of extended equilibration.
Just exactly how many and how long? I don't think that I am entirely serious
here (excuse me).

Another related issue that we perhaps should also keep in mind is how
quantitative our present simulations are. Does it contribute a significant
portion to the uncertainties of our results? The answers depend on ones
objective. Therefore, if one wants to answer a highly specific question from
the simulation (e.g. counterion positions), multiple simulations of
different starting counterion distributions are perhaps needed. In the name
of pragmatism, one can certainly reduce the number of simulations to one and
let the referees scratch their heads.

yong