Hello,
On Thu, Jun 24, 2010 at 1:18 PM, Steve Seibold <seibold.chemistry.msu.edu>wrote:
> Hello
> Sorry, I don't mean to inundate the web site, but I need help if possible.
> I finally got AMBER11 and AmberTools1.4 compiled (serially) but, I am having
> problem in their performance. I have been using AMBER8 in the past, but have
> just started using AMBER11.
>
> AMBER11 problem:
> I serially compiled AMBER11 and it did pass all its tests. However, when I
> do a MD run Sander does start the run and finish with a complete output file
> that shows no problem. I have, in the command line, tagged to save a restart
> file and a mdcrd file. It does save a restart file, but it DOES NOT save a
> mdcrd file. That is, my file "test.mdcrd" is empty although it writes the
> file's name and places it in my directory.
>
We would need to see your input file that you used. I'm guessing you did
not specify ntwx (which is a number that specifies the interval of snapshots
that is saved to the trajectory file) in your &cntrl namelist, which
defaults to a value of 0. This means that coordinates are never written to
the mdcrd, even though you may specify (-x test.mdcrd) on the command-line.
The restart file will be written at the end of the simulation regardless.
Another variable that I would check out for Amber11 is ioutfm. Setting this
to 1 will cause trajectories to be written out in netCDF format, which is a
binary format that stores higher precision coordinates, is faster to write,
faster to read, and is a highly compressed file, so it is a fraction of the
size of an ascii mdcrd (very few downsides to it). Of course, ntwx must be
non-zero (and indeed smaller than the total number of steps you're
simulating).
>
> AmberTools1.4 problem:
> I have serially compiled AmberTools1.4 (with all bugfixes) and ran the
> tests without any errors. However, when I run ptraj to combine multiple
> trajectories and then run "rdparm" to "checkcoords" get the following error:
>
>
> RDPARM MENU: checkcoords RNAP_openMD34_35.mdcrd
>
> checkCoordinates(): Could not predict number of frames for AMBER trajectory
>
ptraj recently underwent a major overhaul. One thing that was added was
file size prediction, so based on the size of the file ptraj can determine
how many frames are in the trajectory. This is based on the current
formatting (F8.3 or something like that). Since I was not around for
Amber8, I have no idea if the formatting is different between then and now.
Someone else would have to comment here.
One thing I would suggest is to visualize your trajectory with your prmtop
file in a program like VMD. If it looks weird and distorted, it means that
your prmtop and mdcrd are incompatible, and this warning is justified. If
it's not distorted (and it remains not distorted throughout the entire
trajectory!), then I'm guessing it's just a different format that's
triggering that error, and it can likely be ignored. You may consider just
using the simple
trajin mdcrd
trajout new_mdcrd
to write out an updated-format trajectory file (you can even add netcdf at
the end of the trajout statement to use that format instead). This may
work, but I'm not sure.
>
> file: RNAP_openMD34_35.mdcrd
>
> If this is not a compressed file then there is a problem
>
> Rank: 0 Atoms: 230188 FrameSize: 5593594 TitleSize: 30 NumBox: 3 Seekable 0
>
> File (RNAP_openMD34_35.mdcrd) is an AMBER trajectory with box coordinates
>
>
> If I use a single trajectory file (i.e. one from a MD run) to obtain the
> RMS there is no problem. It is only when I COMBINE the trajectory files.
>
How do you combine the trajectory files? Using cat to append one file to
another? This would be bad, since there may be a title line at the top of
each trajectory that ptraj won't know how to handle. I would use ptraj to
combine the trajectory files as such
trajin mdcrd1
trajin mdcrd2
...
trajout combined_mdcrd (netcdf)
where the netcdf is optional. This is the better way to combine trajectory
files.
Hope this helps,
Jason
>
> I am using AMBER8 trajectories and a prmtop file from AMBER8..Should this
> make a difference??
>
>
>
>
> Can someone please give me some adivce.
>
> Thanks Steve
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Graduate Student
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Jun 24 2010 - 11:00:03 PDT
