Re: Not all atoms have unique coordinates - revisited

From: Brian Bennion (brian_at_youkai.llnl.gov)
Date: Mon Oct 04 2004 - 10:22:04 CDT

Hello Alberto,
Did you get resolution to this problem yet?

The biggest clue is the VDW energies. Two atoms at least, are sitting on
top of each other.

On Sat, 2 Oct 2004, Alberto Santana wrote:

> Hi Brian,
>
> >You didn't check the hydrated psf file and pdb file together did you?
> I'm not quite sure what u mean here. I did check the pdb file carefully to
> see if there were atoms with identical coordinates. I extracted all the
> coordinates and ran a uniq -d command on it and no output, so NO two entries
> were identical.
>
Float values can never be satisfactorily compared with one another due to
precision. But what I was really getting was this. Did you load your
psf file that describes your protein/water system and then load in the pdb
file of the same system into VMD to visualize the setup?

This would might show which waters are the cause of the problem.

> > Did you make your cell basis vectors insanely large just as a check?
> I did change the cell basis vectors, but I did not change them by more than
> 5 or 6 units. I went from 78 to perhaps 83 or 84. The same to the other 2.
> No more than that. Is it legal to change them insanely? I will make them
> much bigger and try it again.

Just to see if one or two waters is indeed placed incorrectly out of the
box.

>
> > Are all energies 999999 or is it bond, or angles or torsions?
> No, just VDW, TOTAL, TOTAL2, and TOTAL3. I can send you the pdb/psf files if
> this info is not enough.

The log file and pdb and psf files would be nice to see if your
comfortable enough in sending them.

>
> OK, someone suggested a check of the pdb file and that it should not have
> more than 99999 atoms. Ours have 6000 more than that but after the water is
> added. An interesting point is that I compiled NAMD 2.5 on our cluster to
> use Myrinet and also have the downloaded binary (NAMD_2.5b2ss03_Linux-i686).
> The precompiled (downloaded binary) version RUNS NICELY without any
> problems, no warning whatsoever, but the other one gives the warning and the
> funny energies.
>
There might be problems here with the number of atoms. If you used psfgen
to generate your psf and pdb files then the pdbfile will have more than
99999 atoms but they will be grouped into segments that are each less than
the limit.

Its really hard to compare different compiled versions of the code,
especially beta versions. The fact that it runs might be a "bad" thing.
In any case, if you can send your files and log data it would be helpful.

Regards
Brian

> Alberto
>
> -------------------------------------------------------------------
> Alberto Santana, Ph.D.
> Department of Chemistry
> University of Puerto Rico - Mayaguez
> P.O. Box 9019 Phone: (787) 832-4040 x3760
> Mayaguez PR, 00681-9019 http://www.uprm.edu/~asantana
>

*****************************************************************
**Brian Bennion, Ph.D. **
**Computational and Systems Biology Division **
**Biology and Biotechnology Research Program **
**Lawrence Livermore National Laboratory **
**P.O. Box 808, L-448 bennion1_at_llnl.gov **
**7000 East Avenue phone: (925) 422-5722 **
**Livermore, CA 94550 fax: (925) 424-6605 **
*****************************************************************

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 05:18:24 CST