On Mon, Jul 2, 2012 at 8:13 AM, Dilraj LAMA <dilrajl.bii.a-star.edu.sg>wrote:
> Dear Amber users,
> I ran a REMD simulation of a 16mer peptide in
> explicit solution using 24 replicas in the temperature range between
> 300-400K for 40ns. The simulation ran properly with a exchange
> frequency between 20-30%. I then tried to extract the temperature
> trajectory at 300K using the script provided in the amber tutorial
> http://ambermd.org/tutorials/advanced/tutorial7/
>
> Here is the scriptI used:
>
> ----------------------------------------------------------------------------
> #!/bin/csh
>
> ptraj ../../peptide.parmtop << EOF
> trajin ../../production/remd.mdcrd.001 remdtraj remdtrajtemp 300.0
> trajout remd.300K.mdcrd nobox
> go
> EOF
> ---------------------------------------------------------------------------
>
> I am expecting 40000 (since the simulation is for 40ns) coordinate
> sets for each of the 24 replicas. I am however not able to extract the
> entire coordinate sets
>
> For the replica at 300K, I get only 63 sets and there is an error message:
>
> checkCoordinates(): Could not predict number of frames for AMBER
> trajectory file: remd.300K.mdcrd
> If this is not a compressed file then there is a problem
>
> I am not exactly sure the cause of this problem.
>
This is worrying. It suggests to me that your mdcrd file and your prmtop
file are incompatible (or that the mdcrd file is corrupted). My suggestion
is to use VMD or some other similar program to visualize your trajectory
files to make sure everything looks OK and you get the number of frames you
are expecting.
In the future, I strongly suggest using "ioutfm=1" to trigger NetCDF
trajectory files instead of ASCII ones. NetCDF files are safer, smaller,
higher precision, and about an order of magnitude faster to read.
> For your information, I ran the simulation using 192 processors (8
> processor per replica). Prior to running the REMD, I also did a 200 ps
> equilibration for each of the 24 replicas at their specified
> temperatures using 96 processors (4 processor per replica).
>
> Could this change in processor number of the separate run be a
> possible cause of the error?
>
No. The error is caused by something else.
HTH,
Jason
--
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 Mon Jul 02 2012 - 09:00:02 PDT
