# [gmx-users] Re: RMSD VS. parallel simulation

Mark Abraham Mark.Abraham at anu.edu.au
Sat Apr 5 13:31:50 CEST 2008

```Dechang Li wrote:
>> Send gmx-users mailing list submissions to
>> 	gmx-users at gromacs.org
>>
>> To subscribe or unsubscribe via the World Wide Web, visit
>> 	http://www.gromacs.org/mailman/listinfo/gmx-users
>> or, via email, send a message with subject or body 'help' to
>> 	gmx-users-request at gromacs.org
>>
>> You can reach the person managing the list at
>> 	gmx-users-owner at gromacs.org
>>
>> than "Re: Contents of gmx-users digest..."
>>
>>
>> Today's Topics:
>>
>>   1. RMSD VS. parallel simulation (DeChang Li)
>>   2. Re: RMSD VS. parallel simulation (Mark Abraham)
>>
>>
>> ----------------------------------------------------------------------
>>
>> Message: 1
>> Date: Sat, 5 Apr 2008 11:48:15 +0800
>> From: "DeChang Li" <li.dc06 at gmail.com>
>> Subject: [gmx-users] RMSD VS. parallel simulation
>> To: gmx-users at gromacs.org
>> Message-ID:
>> 	<bbfc6b070804042048s297a545fie0d4122ae70d19a3 at mail.gmail.com>
>> Content-Type: text/plain; charset=ISO-8859-1
>>
>> Dear all,
>>
>>  I used Gromacs-3.3.1 to simulate a small protein in water.
>> I have used 2 and 16 CPUs to do the simulation respectively. But
>> I got the RMSD of the protein equilibrated to 0.15 nm in the 2 CPUs
>> simulation and 0.20 nm in the 16 CPUs one. Are these differences
>> reasonable?
>>  In the 16 CPUs simulation, the RMSD of protein at t=0 was about
>> 0.1 nm, why not equal to zero? I used the initial structure for the
>> least squares fit.
>>
>>
>>
>>
>> Best regards,
>> 2008-4-5
>>
>>
>> =========================================
>> Dechang Li, PhD Candidate
>> Department of Engineering Mechanics
>> Tsinghua University
>> Beijing 100084
>> PR China
>>
>> Tel:   +86-10-62773779(O)
>> Email: lidc02 at mails.tsinghua.edu.cn
>> =========================================
>>
>>
>> ------------------------------
>>
>> Message: 2
>> Date: Sat, 05 Apr 2008 14:02:29 +1000
>> From: Mark Abraham <Mark.Abraham at anu.edu.au>
>> Subject: Re: [gmx-users] RMSD VS. parallel simulation
>> To: Discussion list for GROMACS users <gmx-users at gromacs.org>
>> Message-ID: <47F6F9D5.8090305 at anu.edu.au>
>> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
>>
>> DeChang Li wrote:
>>> Dear all,
>>>
>>>   I used Gromacs-3.3.1 to simulate a small protein in water.
>>> I have used 2 and 16 CPUs to do the simulation respectively. But
>>> I got the RMSD of the protein equilibrated to 0.15 nm in the 2 CPUs
>>> simulation and 0.20 nm in the 16 CPUs one. Are these differences
>>> reasonable?
>> MD allows you to observe an ensemble (usually at or approaching
>> equilibrium) over time. Any single point of that time isn't any more
>> significant than any other.  So you should expect any pair of points in
>> different simulations (which had their velocities generated with
>> different random numbers, right?) to generally give different values for
>> observables, and for that not to mean anything much. Distributions of
>> observables over long enough periods of time should be the same, however.
>
>    I have done the two simulation last 1 ns long. For the last 500ps,
> the RMSD of the protein is 0.21nm (0.016nm) in the 16 CPUs simulation, but
> 0.15nm (0.011nm) in the 2 CPUs simulaiton, respectively. (The value in
> parentheses reflects the standard deviation). I think at the last 500ps the
> system should be equilibrium. Why the RMSD so different? Does it mean I should
> simulate much longer?

These numbers don't mean anything out of context. Even the most stable
protein structure of a thousand residues might have an all-atom average
RMSD of this size. In a decapeptide it might mean there isn't a single
conformer, or that equilibrium is not yet reached.

>>>   In the 16 CPUs simulation, the RMSD of protein at t=0 was about
>>> 0.1 nm, why not equal to zero? I used the initial structure for the
>>> least squares fit.
>> If you've done an equilibration or EM, the structure can have changed
>> during that.
>
>    I mean the initial structure is the structure I used to generate the
> .tpr file for mdrun_mpi.

What were your command lines, choices for groups and outputs? The most
likely simple explanation is that you haven't done what you think you've
done :-)

Mark

```