Re: [AMBER] Query about checking of convergence in REMD

From: Daniel Roe <daniel.r.roe.gmail.com>
Date: Fri, 26 Apr 2013 10:47:38 -0600

Hi,

There is an extensive discussion of determining convergence in REMD
simulations in this publication:

Henriksen et al, J. Phys. Chem. B, 2013, 117 (15), pp 4014–4027
http://pubs.acs.org/doi/abs/10.1021/jp400530e

-Dan

On Fri, Apr 26, 2013 at 7:14 AM, gargi borgohai <gargib2011.gmail.com> wrote:
> Dear AMBER Users,
>
> Your suggestion regarding checking of convergence of REMD was quite
> helpful. But still one more confusion arises for confirmatrion of
> convergence. The system I am simulating has total 44 number of replicas.
> While checking convergence (RMSD plot) it is observed that out of 44
> replicas 8 replicas are showing much more fluctuations in comparision to
> the rest. The remaining 36 replicas can be concluded to be converged. Now
> my question is: do I have to run the system further so that the remaining
> replicas also get converged ?
>
> Thanking You...
> Sincerely
> Gargi Borgohain.
> PhD Scholar.
> Indian Institute of Technology, Guwahati.
> India
>
>
>
>
>
>
> Message: 15
> Date: Wed, 6 Feb 2013 07:32:35 -0500
> From: Carlos Simmerling <carlos.simmerling.gmail.com>
> Subject: Re: [AMBER] Query about checking of convergence in REMD
> To: AMBER Mailing List <amber.ambermd.org>
> Message-ID:
> <CAGk3s-SQFhMbJWM0dngkX4vSR1+_=X0_5cuFYE0zOkcLcZyNyw.mail.gmail.com>
> Content-Type: text/plain; charset=ISO-8859-1
>
> you can do a few things:
> - perform 2 runs, one starting from all replicas folded and 1 starting from
> all replicas unfolded. All measures should match between the 2 runs.
> anything that doesn't match is unreliable.
> - do analysis on each replica (not each temperature) separately. Every
> replica should have the same properties (rmsd histograms, fraction folded,
> fraction of time at each temperature, and so on). If they don't, it's not
> converged.
>
> remember that "converged" is a bit vague since different properties
> converge at different rates. Potential energy converges quickly, while
> structure sampling does not. Every thing you measure needs error bars. Then
> you know the precision, which is much more quantitative than a binary idea
> of "converged" or not.
>
> On Wed, Feb 6, 2013 at 7:21 AM, gargi borgohai <gargib2011.gmail.com> wrote:
>
>> Dear AMBER users,
>>
>> I am using REMD method to study protein folding/unfolding behavior. I
> have
>> run the system for 35ns. Can anyone help me regarding how can I confirm
>> whether my system has converged or not?
>>
>>
>> Thanking You..
>> Sincerely
>> Gargi Borgohain.
>> PhD Scholar.
>> Indian Institute of Technology, Guwahati.
>> India.
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber



-- 
-------------------------
Daniel R. Roe, PhD
Department of Medicinal Chemistry
University of Utah
30 South 2000 East, Room 201
Salt Lake City, UT 84112-5820
http://home.chpc.utah.edu/~cheatham/
(801) 587-9652
(801) 585-9119 (Fax)
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Apr 26 2013 - 10:00:03 PDT
Custom Search