S.P. Skinner | 6 Oct 11:33 2011
Picon
Picon

collapse term

Hi All,

I am attempting to perform some SA on a complex of two proteins, one has 
an elongated shape and the other is roughly spherical.  If I do not 
include a collapse term, the two proteins end up inside each other and 
all restraints are thus violated and the VdW energy is huge.  If I 
include a collapse term for the entire complex, the proteins remain far 
apart, as they are randomly placed at the beginning of the run.  I have 
also try to separate the complex into three separate roughly spherical 
domains, to then calculate the Rgyr for each domain and then to include 
these in three separate collapse assign statements in old xplor 
language, when I do this the collapse term becomes NaN as does the 
gradient and the Epot and Ekin terms.  How can I handle this sort of 
complex and get the proteins to move close together without getting too 
close.  Do I need an Rgyr term, if so, what form should it take?  In the 
SA run bonds, angles, impropers, VdW and PRE terms are all included in 
the XplorPot.

Any ideas?

Many Thanks and Kind Regards

Simon Skinner

--

-- 
Simon P Skinner
Protein Chemistry Group
Leiden Institute of Chemistry, Universiteit Leiden
Phone: +31 71 527 6089 / Fax: +31 71 527 4349
E-mail : skinnersp <at> chem.leidenuniv.nl 
(Continue reading)

Charles Schwieters | 6 Oct 15:29 2011

Re: collapse term


Hello Simon--

>
> I am attempting to perform some SA on a complex of two proteins, one has
> an elongated shape and the other is roughly spherical.  If I do not
> include a collapse term, the two proteins end up inside each other and
> all restraints are thus violated and the VdW energy is huge.  If I
> include a collapse term for the entire complex, the proteins remain far
> apart, as they are randomly placed at the beginning of the run.  I have
> also try to separate the complex into three separate roughly spherical
> domains, to then calculate the Rgyr for each domain and then to include
> these in three separate collapse assign statements in old xplor
> language, when I do this the collapse term becomes NaN as does the
> gradient and the Epot and Ekin terms.  How can I handle this sort of
> complex and get the proteins to move close together without getting too
> close.  Do I need an Rgyr term, if so, what form should it take?  In the
> SA run bonds, angles, impropers, VdW and PRE terms are all included in
> the XplorPot.
>

This is not behavior which we usually see. Which protocol are you
using as a starting point? This script is probabably a good starting
point:

/home/schwitrs/xplor/eginput/relaxRatio/dock_rrp.py

(you can remove the relaxation data stuff, if it doesn't apply).

In general, the Vgyr term is safer (than Rgyr) to use on nonglobular
(Continue reading)

Tanya Zaliznyak | 26 Oct 17:29 2011

Shake and nconstraints.

I am trying to run dynamics of DNA oligo ( 11 mer) with explicit water molecules ( about 3000 molecules).

Shake routine is :

shake
        reference = parameters
        bonds (hydrogen) (all)
        tolerance = 1.0e-06
        nconstraints=8000

end

and error message after this is:
**********************************

X-PLOR>shake
 SHAKE>        reference = parameters
 SHAKE>        bonds (hydrogen) (all)
 SELRPN:   7142 atoms have been selected out of  11032
 SELRPN:  11032 atoms have been selected out of  11032
 SHKSET: reference = parameters
 %XREFIN-ERR: allocation for SHAKE-constraints exceeded
              increase NCON parameter and run again
 %XREFIN>-ERR: exceeded allocation for SHAKE-constraints
 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
 BOMLEV=    0 reached.  Program execution will be terminated.
 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
 Subroutine DIE called . Terminating

Supposedly, I have 7142 hydrogen atoms and only constraints which SHAKE routine will take into consideration is bonds between hydrogen and heavy atoms, no more then 7142 .
I tried 8000, 30000, 600000...
Sours code didn't give me any limitation for this parameter, it should be integer, it is not limited. What is wrong with it?
Can somebody give me any advise? Thank you all, Tanya Zaliznyak.
_______________________________________________
Xplor-nih mailing list
Xplor-nih <at> nmr.cit.nih.gov
http://dcb.cit.nih.gov/mailman/listinfo/xplor-nih
Charles Schwieters | 26 Oct 18:24 2011

Re: Shake and nconstraints.


Hello Tanya--

>
> I am trying to run dynamics of DNA oligo ( 11 mer) with explicit water
> molecules ( about 3000 molecules).
>
> Shake routine is :
>
> shake
>         reference = parameters
>         bonds (hydrogen) (all)
>         tolerance = 1.0e-06
>         nconstraints=8000
>
> end
>
> and error message after this is:
> **********************************
>
> X-PLOR>shake
>  SHAKE>        reference = parameters
>  SHAKE>        bonds (hydrogen) (all)
>  SELRPN:   7142 atoms have been selected out of  11032
>  SELRPN:  11032 atoms have been selected out of  11032
>  SHKSET: reference = parameters
>  %XREFIN-ERR: allocation for SHAKE-constraints exceeded
>               increase NCON parameter and run again

Does it help if you place an additional
    nconstraints=8000
line at the later location (where the error occurs)?

best regards--
Charles
--
Charles Schwieters     email:   Charles <at> Schwieters.org
                       www:     http://schwieters.org/cds
phone: (301) 402-4914  PGP key: http://schwieters.org/cds/pgp.txt

_______________________________________________
Xplor-nih mailing list
Xplor-nih <at> nmr.cit.nih.gov
http://dcb.cit.nih.gov/mailman/listinfo/xplor-nih
Jurgen F. Doreleijers | 28 Oct 16:29 2011
Picon

i5 core parallel ?

Hi guys, Charles,

So e.g. in: xplor-nih-2.27/eginput/gb1_rdc/.runAll

the command: 

xplor -quick -py -parallel anneal.py 

should already use multiple cores?

It doesn't inside my Ubuntu virtual machine with 4 cores. Should I mention the -smp 4 to xplor or add another setting somewhere? I don't see how the class StructureLoop can be instructed to take advantage. Below I've tried with an environment setting.

We're trying to run a little faster today...

Cheers,
Jurgen

Some tests:
iLvc/~/ echo $XPLOR_NUM_PROCESSES
4
iLvc/~/ xplor -py
                       Xplor-NIH version 2.27 
 
 C.D. Schwieters, J.J.  Kuszewski,       Progr. NMR Spectr. 48, 47-62 (2006). 
 N. Tjandra, and G.M. Clore              J. Magn. Res., 160, 66-74 (2003). 
 http://nmr.cit.nih.gov/xplor-nih        based on X-PLOR 3.851 by A.T. Brunger 

 User: i            on: vc        (x86/Linux     )  at: 28-Oct-11 16:27:08
No entry for terminal type "xterm";
using dumb terminal settings.
python> import xplor
python> xplor.p_numProcs
1

_______________________________________________
Xplor-nih mailing list
Xplor-nih <at> nmr.cit.nih.gov
http://dcb.cit.nih.gov/mailman/listinfo/xplor-nih
Charles Schwieters | 28 Oct 17:10 2011

Re: i5 core parallel ?


Hello Jurgen--

> 
> So e.g. in: xplor-nih-2.27/eginput/gb1_rdc/.runAll
> 
> the command: 
> 
> xplor -quick -py -parallel anneal.py 
> 
> should already use multiple cores?
> 

It will use multiple cores if you specify which nodes to run on using
the -machines option, or you use the -smp option (in which -parallel
is redundant). e.g.

  xplor -quick -py -smp 4 anneal.py

best regards--
Charles 

--
Charles Schwieters     email:   Charles <at> Schwieters.org
                       www:     http://schwieters.org/cds
phone: (301) 402-4914  PGP key: http://schwieters.org/cds/pgp.txt

_______________________________________________
Xplor-nih mailing list
Xplor-nih <at> nmr.cit.nih.gov
http://dcb.cit.nih.gov/mailman/listinfo/xplor-nih
Jurgen F. Doreleijers | 28 Oct 19:36 2011
Picon

Re: i5 core parallel ?

On Fri, Oct 28, 2011 at 5:10 PM, Charles Schwieters <charles <at> schwieters.org> wrote:

Hello Jurgen--

>
> So e.g. in: xplor-nih-2.27/eginput/gb1_rdc/.runAll
>
> the command: 
>
> xplor -quick -py -parallel anneal.py 
>
> should already use multiple cores?
>

It will use multiple cores if you specify which nodes to run on using
the -machines option, or you use the -smp option (in which -parallel
is redundant). e.g.

 xplor -quick -py -smp 4 anneal.py

Excellent, I can reproduce this on the virtual machine (VM).
Well, of course we tested that before on a different install but not on my VM.
We'll see what went wrong on the other install.

Thanks for the quick & easy response Charles!

Cheers,
Jurgen
_______________________________________________
Xplor-nih mailing list
Xplor-nih <at> nmr.cit.nih.gov
http://dcb.cit.nih.gov/mailman/listinfo/xplor-nih

Gmane