AMBER Archive (2008)

Subject: Re: AMBER: FW: MD run time inquiry

From: Carlos Simmerling (
Date: Thu Mar 27 2008 - 05:41:02 CDT

it depends very much on what you want to study and what actually
happens. A good rule of thumb is that whatever happens, it should be
Also you need to know something about different timescales of motion- if you
to look at folding, then clearly 1ns is not enough. if you want to look at
side chain
motion, it may be. Another point is whether you want your data to give a
of what might happen, or if you want to calculate actual populations of
structures. The latter probably requires observation of multiple
If your goal is to study loop dynamics as in the example, then I would say
run it
until whatever you are measuring or reporting gives the same results in two
runs. I would expect loop motions to take multiple ns at minimum, but some
move more slowly than others so if you don't have experimental data you will
just need
to run it out and see.

On Thu, Mar 27, 2008 at 6:13 AM, Campbell, Patrick <>

> ------------------------------
> *From:* Campbell, Patrick
> *Sent:* Wed 3/26/2008 10:18 PM
> *To:*
> *Subject:* MD run time inquiry
> A pleasant good day to you all,
> I am currently doing an MD simulation, following on the example of Dr.
> Matt Lee (Loop dynamics of the HIV-1 integrase core domain) on the AMBER
> scripps site. Thus far, I have done a simulation for 300 picoseconds and
> will be looking to extend this time to 1.1 nanoseconds. My question is
> this - is there an established amount of time that would be deemed as
> sufficient for this simulation to run, such that the results will be
> acceptable for publication?
> Thanks again for all your assistance and do have a great day!
> Patrick

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to