On Tue, Sep 18, 2012 at 10:25 PM, Fabrício Bracht <bracht.iq.ufrj.br> wrote:
> Hello. Thank you for the reply. It really was kind of stupid to me to
> leave the double-quote open. As for the / character missing, there is
> no second line in my file. The email server must have formatted the
> text, and the result was a second line. But anyway, thank you for the
> tip.
> I have corrected the input and the dist.RST file now looks like this
>
> # Distance restraint for chain A and B (center of mass)
> &rst restraint="distance (com(1-33.N,CA,C),com(34-66.N,CA,C))", r1=0.,
> r2=11., r3=11., r4=35.0, rk2 = 30., rk3 = 30.,
>
> When I type the pmemd command, a warning appears on screen stating that:
>
> At line 990 of file nmr_calls.F90 (unit = 33, file = 'dist.RST')
> Fortran runtime error: Cannot match namelist object name restraint
>
> I have simply copied and modified a bit the lines on the amber12
> manual that state:
>
> The following are all examples of valid restraint definitions:
> restraint = "distance( (45) (49) )"
> = "angle (:21.C5’ :21.C4’ 108)"
> = "torsion[-1,-1,-1, com(67, 68, 69)]"
> = "angle( -1, plane(81, 85, 87, 90) )"
> = "angle(plane(com(9,10),:5.CA,31,32),plane(14,15,15,16))"
> = "coordinate(distance(:5.C3’,:6.O5’),-1.0,distance(134,-1),1.0)"
>
This syntax is specific to sander. Look at page 202 of the Amber 12 manual
where it describes the "restraint" variable in the &rst namelist. In that
description it says:
New as of Amber 10 (sander only), users may now use a "natural language"
system to define restraints by using the RESTRAINT character variable.
If you want to use pmemd.cuda and/or pmemd with NMR restraints, you will
need to use the "iat" variable (and "igr1" and "igr2" variables for COM
distances). In light of this, I actually never use the "natural language"
input myself.
Note that pmemd supports only a relatively small subset of the Amber nmropt
functionality, and pmemd.cuda supports an even more limited set. As a
result, if you think you have an input file that should work for pmemd but
doesn't, try it on sander. If it works, then most likely you've hit a
pmemd limitation.
HTH,
Jason
P.S. As another note -- check out the $AMBERHOME/test/nmropt/ test cases
(only present after you've applied all Amber12 bug fixes). The Makefile
lists all of the tests that work with pmemd (and you can find the
pmemd.cuda-compatible ones in $AMBERHOME/test/cuda/nmropt). My suggestion
is to adapt one of these input files to your system (since these are known
to work for pmemd and pmemd.cuda, respectively).
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Sep 18 2012 - 20:00:02 PDT
