[gmx-users] RMSD Matrix Error

[Apramita Chand]

Dear Mark,
Thanks for your reply. I think you're right about the space needed for
generating RMSD Matrix. I definitely was short of 160GB !!
Further, you've talked about me using highly correlated frames and that a
suitable sub-sampling might solve the problem. How would I know which
frames to use?



Message: 2
Date: Mon, 05 Jun 2017 01:37:37 +0000
From: Mark Abraham <mark.j.abraham at gmail.com>
To: gmx-users at gromacs.org, gromacs.org_gmx-users at maillist.sys.kth.se
Subject: Re: [gmx-users] RMSD Matrix error
Message-ID:
        <CAMNuMATiG=MtnQwgMZF3r+sa-b2qAo38ywgW4x9HgpLUTFuyag at mail.gmail.com>
Content-Type: text/plain; charset="UTF-8"

Hi,














*On Sun, Jun 4, 2017 at 4:08 PM Apramita Chand <apramita.chand at gmail.com
<apramita.chand at gmail.com>> wrote: > Dear All, > > When I'm trying to
construct a RMSD matrix , using the command > g_rms  -s protein_equili.gro
 -f  protein_model1_ut.xtc   -m > rmsd-matrix.xpm  -tu ns > > I get the
error: > Last frame     200000 time   20.000 > > Building RMSD matrix,
200001x200001 elements >  element 28982; time  2.90  Killed*
>

That matrix has 4x10^10 elements, each of which needs 4 bytes of memory.
1GB is about 1x10^9 bytes, so you'd need at least 160 GB of memory.

The real issue is that you are probably using a large number of highly
correlated frames, so even if you could form the full distance matrix, you
would not learn any more than you would from a matrix formed from a
suitable sub-sampling.





*> I have given the reference structure to be the one prior to production
run > and after equilibration step. I have also tried the command with .tpr
file. > Same error! >*

The problem doesn't change with the nature of the reference structure.

Mark



yours sincerely,
Apramita