Re: NAMD Question: Generating ensemble of structures

From: Brian Radak (
Date: Thu Sep 01 2016 - 08:44:10 CDT

Is the default behavior for DCDFreq > 0 not enough? A DCD trajectory
file is effectively a concatenation of PDB files (but with superior
precision), so long as you first load a PSF with the appropriate atom info.

Alternatively, if you insist on separate files, you can use the options:

restartSave yes
restartFreq <integer for steps between file writing>
restartName <basename for different files>

One way in which this is all /not/ what you want is that regularly
spaced structures will generally not be unique - some of them are
assuredly going to be very similar and thus not look like an NMR
ensemble (since you will have repeats). Presumably you could run a
clustering algorithm on the DCD trajectory and then extract
representative structures, but this is not currently a feature in NAMD.



On 08/31/2016 12:52 AM, Oscar Bastidas wrote:
> Hello,
> I am a novice at using NAMD and I have a question on whether the NAMD
> software can be implemented to provide as output PDB structures at
> regular intervals (specified by me in accordance with my total running
> simulation time) such that I end up with an ensemble of structures
> similar to the ensembles provided by protein NMR structure elucidation
> studies. I am interested in knowing, if this is possible with NAMD,
> how it would be implemented. Would you please tell me if to your
> knowledge this is a possibility with NAMD and how I would go about to
> do it?
> If this is not a possibility, then I am open to the following
> alternative, again, if it is possible: Can I have NAMD produce a final
> PDB structure at the end of a simulation for a certain amount of time
> and, then, once that simulation is finished, feed this output PDB
> structure that shows the final trajectory stopping point as the input
> PDB into another separate simulation run? As you might imagine, I'd
> intend on repeating this multiple times (I'm specifically interested
> in obtaining an ensemble of 10 structures) to obtain superimposable
> structures that would ultimately resemble the aforementioned NMR
> ensemble, but show the actual range of true motion of the protein
> (like strobe photography of a golfer). Thank you for any help you can
> provide.
> Respectfully submitted,
> Oscar B.

Brian Radak
Postdoctoral Appointee
Leadership Computing Facility
Argonne National Laboratory
9700 South Cass Avenue, Bldg. 240
Argonne, IL 60439-4854
(630) 252-8643

This archive was generated by hypermail 2.1.6 : Sun Dec 31 2017 - 23:20:40 CST