Re: [AMBER] Chains missing during the simulation

From: Jason Swails <>
Date: Tue, 21 Oct 2014 07:39:46 -0400

On Tue, 2014-10-21 at 13:26 +0530, Harshavardhan Khare wrote:
> Dear all,
> I am trying to simulate a protein with 6 chains. Everything goes fine up to
> 1 ns. From approximately 1.5 ns onwards, 3 chains are missing from the
> coordinates.

What do you mean when you say 3 chains are "missing"? Their coordinates
are NaN? Inf? Do they move to the other side of the box (appearing to
"disassociate" from the other chains)?

The number of atoms in sander and pmemd _never_ changes, and all
coordinates are _always_ printed out for every frame. The literal
definition of what you said (i.e., that sander or pmemd suddenly stopped
printing out the coordinates for 3 chains such that their coordinates
simply don't exist) seems impossible. For one thing, this would corrupt
the NetCDF trajectory file and cause any programs reading them (like
VMD) to spit out an error and show nothing at all.

You can use the "ncdump" program to look at the actual data stored in a
NetCDF file. That might help you determine whether some of the
coordinates went off to NaNdyland.


Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Tue Oct 21 2014 - 05:00:02 PDT
Custom Search