[gmx-users] Re: gmx-users Digest, Vol 43, Issue 81

[servaas michielssens]

  Date: Wed, 21 Nov 2007 14:03:53 -0800
> From: "David Mobley" <dmobley at gmail.com>
> Subject: Re: [gmx-users] restart an amber run in gromacs
> To: "Discussion list for GROMACS users" <gmx-users at gromacs.org>
> Message-ID:
> <bc2c99750711211403oc4bd410g8eeda8a2c25ca216 at mail.gmail.com>
> Content-Type: text/plain; charset=ISO-8859-1
>
> Hi,
>
>> servaas michielssens wrote:
>> > I used amb2gmx.pl (from http://chemistry.csulb.edu/ffamber/) to convert
>> > an amber topology and restart file to a gromacs topology and structure
>> > file and tried to continue the simulation in gromacs. I got the
>> > following error message:
>> >
>> > Constraint error in algorithm Lincs at step 0
>> >
>> > t = 0.000 ps: Water molecule starting at atom 4642 can not be settled.
>> > Check for bad contacts and/or reduce the time-step. Wrote pdb files 
>> > with
>> > previous and current coordinates
>> >
>> > It seems strange to me that this error occurs as there was no problem
>> > with this simulation in amber. (it ran a few ns in amber).
>> > I tried to run it after minimalization and this worked fine. So there
>> > seems to be a problem while converting the files, I am new to gromacs
>> > and I don't really have a clue how to find the cause of this problem,
>> > perhaps some of you also tried to continue a amber run in gromacs?
>
> The script is not intended for continuing AMBER runs in GROMACS. As
> far as I know no one has ever tested it on this. There are a number of
> reasons why this would not be a good idea, not the least of which is
> that you are changing simulation parameters/algorithms when you do so,
> as Mark notes in the following paragraph.
>
>> Judging by my interpretation of your last email, you probably did a more
>> radical change of ensemble than you think you did. Both AMBER and
>> GROMACS are approximating some true ensemble, and there's no reason to
>> assume that a configuration from one will be equilibrated in the other.
>
> Also, I personally have never used the script to convert an entire
> system including solvent from AMBER to GROMACS -- just proteins and
> ligands, for example, One might worry that the simulation box will not
> be converted correct, which could certainly lead to the problems
> you're seeing.
>
> Short answer: Even if you think the script is working perfectly,
> converting the box properly, etc., then you should at the very least
> minimize and equilibrate your amber end state before doing production
> in gromacs.
>
> David
>
>
>> Mark

Thanks for your help, I checked the perl script (amb2gmx.pl)  it should work 
for a solvated protein. The solvend and box are converted to gromacs but the 
box information is not converted as it should, amber only defines d and 3 
angels (+information in the topology that it deals with a truncated 
octahedron), while gromacs needs 3 vectors to defines a truncated octahedron 
and this information is only in the .gro file not in the topology. The 
amb2gmx.pl script only copies tree values to the .gro file for an truncated 
octahedron so gromacs thinks it is a cubic box.
Normally the ensembles are both NTP, I can hardly believe that differnces in 
algoritm would cause a "Constraint error in algorithm Lincs at step 0" and 
make the simulation crash. But I agree that this procedure is not the way to 
go for a good simulation, the only intention of this was doing a speed 
comparision.

kind regards,

servaas