[gmx-users] inexplicable rmsf error - Too many iterations in routine JACOBI

[Mala L Radhakrishnan]

Hi all,

I am trying to run gmx RMSF using.a particular snapshot as my reference
structure (-s option).

When I extract a gro file to make my reference structure that has only the
molecule that INCLUDES the alpha-carbons in my index file that I want to
use for my RMSF calculation (I checked), then midway through its reading in
the frames (and not always at the same point) I get the following error:

Program:     gmx rmsf, version 2016.3
Source file: src/gromacs/linearalgebra/nrjac.cpp (line 165)

Fatal error:
Error: Too many iterations in routine JACOBI

When I extract a gro file that has an even larger subset of the atoms in
the frame than the one requested previously, it seg-faults right away.

But when I extract a snapshot that has the *entire* system to use as my
reference structure (all atoms, including solvent and ions), it seems to
work. I am using the same .xtc file and index file in all cases, just
changing the -s file.

Googling the error suggests that I have a mismatch somewhere in my index
file and/or structure file, but (a) I have checked this by eye and can't
find any mismatches and (b) when I do he same sort of extraction of a
subset of the system on several other trajectories it works just fine -- in
other words, this seems to be the only trajectory where I need to extract
the entire system (including solvent, ions, etc.) as my reference frame
just to get an rmsf of alpha carbons in the protein molecule of interest. I
am using version 2016.3.  I have tried multiple gromacs versions to extract
and run just in case but got the same error.  The reference structure also
appears to be structurally fine (no broken molecules, etc.)

What could be different about this trajectory that is causing the problem?
Again, I looked by eye at the gro file I'm using as my reference structure
and made sure that it contains all the atoms in the index file selection
I'm using for my RMSF calculation...

Thanks for any help. I am happy to send you the files for which it works
and for which it doesn't...just very confused!

Mala



-- 
Mala L. Radhakrishnan
Associate Professor of Chemistry
Director, Biochemistry Program
Wellesley College
106 Central Street
Wellesley, MA 02481
(781)283-2981