From: Aron Broom (broomsday_at_gmail.com)
Date: Mon May 28 2012 - 08:55:11 CDT
I think perhaps a lot of the lack of response is because simulated
annealing (like just about everything in MD) isn't something that has a set
methodology to follow. The idea is that you have the system at high
temperature where the energy barriers between minima are decreased, and
hopefully as you slowly cool you fall into the lowest one. This becomes
more likely the slower you are cooling. So, yes, you could start with the
high temperature structures, since the heating phase can be fast anyway.
In terms of cooling, you want to go slowly enough to find the global
minimum, but you'd have to look at some successful papers in order to get
any kind of sense for how slow that is. Even then, what works for others
might not work for you. Since the protein folding landscape is thought to
be extremely rugged (many minima), it may be difficult to find the global
minimum, depending on how far from the global minimum structure your high
temperature starting structures are.
On Mon, May 28, 2012 at 2:11 AM, Puspita Halder <puspitah_at_gmail.com> wrote:
> Hi all,
> I have posted an e-mail in the namd mailing list regarding protein folding
> study using simulated annealing technique few days back. But I am yet to
> get any reply from anyone of you. I am writing it once again.
> I have been working on protein folding problem for the last few months. I
> have some high temperature (400K, 500K) euilibrated trajectories of my
> protein with which I would like to adopt the simulated annealing run. Is
> this doable beacause in most of the cases I saw that this technique employs
> the heating as well as the cooling phase. Can I skip the heating phase if I
> start directly from the high temerature structures ? Another thing is what
> should be the ideal step size for temperature decrease and the number of
> iterations for each step in the loop by using this method?
> If you have answers for my queries then please respond.
-- Aron Broom M.Sc PhD Student Department of Chemistry University of Waterloo
This archive was generated by hypermail 2.1.6 : Tue Dec 31 2013 - 23:22:01 CST