[gmx-users] R: R: Re: g_cluster error in rhombic dodecahedric system

[Anna Marabotti]

Dear Tsjerk,
once again thank you for your very useful suggestions. I'm not sure that
there are many clusters in my trajectory, so I don't want to lower my
cutoff. Instead I'm quite surprise to find that it should be a problem
related to the absence of transitions, I don't understand why the program
blocks without writing any output (including the average structure of the
cluster) only because it does not find any transition. Anyway, I will take
out the analysis concerning transitions to see what will happen to the
system.
(Probably now I'm too much suspicious about rhombic dodecahedron system:
it's the first time I'm using it and I'm experimenting a lot of problems,
whereas with my beloved tetrahedric cell I made these analyses dozens of
times without too much troubles...)
Anna 

-----Messaggio originale-----

Message: 1
Date: Thu, 21 Jul 2011 18:26:57 +0200
From: Tsjerk Wassenaar <tsjerkw at gmail.com>
Subject: Re: [gmx-users] R: Re: g_cluster error in rhombic
	dodecahedric system	(Mark Abraham)
To: Discussion list for GROMACS users <gmx-users at gromacs.org>
Message-ID:
	<CABzE1SiYof-z_zoKLQ-0tY2iju9VUZqy3cSoi-KDXLVh=7_i=w at mail.gmail.com>
Content-Type: text/plain; charset=ISO-8859-1

Hi Anna,

On Thu, Jul 21, 2011 at 5:11 PM, Anna Marabotti
<anna.marabotti at isa.cnr.it> wrote:
> Thanks Mark for suggestion, I'll try to do a simpler clustering method. In
> the meantime, I only would like to add a further info: I pre-processed my
> trajectory following someone's (Justin, IIRC) suggestion in a previous
> message, using trjconv -pbc mol -ur compact. So, in my trajectory the
> molecules should be whole (at least, I see my protein as a whole). This
> pre-processing was necessary because with trjconv -pbc whole I was not
able
> to have my homodimeric protein in the unit cell: the protein was split in
> two monomers in two adjacent unit cells. Could be the presence of the two
> identical monomers to "confuse" Gromacs that sees an RMSD minimum of 0?
> Anna

No, the output says the RMSD ranges from 0.07 to 0.28. The problem
arises because you only get one cluster using a cutoff of 0.25 nm. If
you're certain there are some clusters to be identified, despite the
low RMSD values that sort of indicate there is not much variation,
then you can lower the cutoff. The error also does not pertain to the
RMSD matrix, but probably to the matrix of transitions, because there
aren't any. Please do note that this has absolutely nothing to do with
using a rhombic dodecahedron :p

Cheers,

Tsjerk



-- 
Tsjerk A. Wassenaar, Ph.D.

post-doctoral researcher
Molecular Dynamics Group
* Groningen Institute for Biomolecular Research and Biotechnology
* Zernike Institute for Advanced Materials
University of Groningen
The Netherlands


------------------------------