From: Nicolas Martin (
Date: Fri Feb 19 2016 - 11:48:59 CST

Dear Josh, dear all,

Thanks you for your kind answer. I gave it a try without bigdcd and it
didn't really improve the situation. I have the impression that even the
centering is not working properly. Indeed when running only the first
pbc wrap -compound fragment -center com -centersel "segname A" -now -sel

Segment A isn't in the middle of my box. Segment E is and still moving
quite a lot. I do not understand why the centering algorithm would allow
such a variation in the position of the center of mass of the selection
neither why segE is centered instead of segA.

Although since I have an homopentamere I cannot pick the largest one as
Josh suggested.



On 02/19/2016 04:50 PM, Josh Vermaas wrote:
> Hi Nicholas,
> I've never done it with bigdcd, but another alternative is a set of
> two wrap commands. I think unwrap depends on a previous set of
> coordinates to be loaded, so I'm not sure what it would even do when
> bigdcd only loads one frame at a time.
> #This gets the protein together into one group, assuming each chain is
> its own segment, and isn't too big relative to the periodic box
> pbc wrap -compound fragment -center com -centersel "segname PA" -now
> -sel "protein"
> #This step is optional, and recenters the lipids and waters around the
> protein that is put back together.
> pbc wrap -compound fragment -center com -centersel "protein" -now
> Of course, you'd need to adjust the "segname PA" selection to just
> pick one segment of the multimer, preferably the largest one.
> -Josh Vermaas
> On 02/19/2016 08:30 AM, Nicolas Martin wrote:
>> Dear users,
>> I recently run a simulation in NAMD of a pentameric membrane protein
>> using the wrapping option. Even tough the water and membrane are
>> correctly wrapped, the protein diffuse in the simulation box and end
>> ups crossing the periodic boundaries. Doing so, the protein is not
>> kept full, but 2 chains over 5 jump on the other side of the box. The
>> wrapping in namd is made so there is no long bonds and only full
>> chains move to the other side of the box.
>> I need all the chains to be put back together at the center of the
>> box for further analysis.
>> I tried to follow the documentation for doing so and the best I could
>> end up with is a protein more or less (the com seems to be moving
>> significantly) well centered in the box and several frames over a
>> long trajectory with stretched bonds. Here is the routine I used over
>> all trajectory frames:
>> pbc unwrap -sel protein -now
>> pbc wrap -compound res -center com -centersel "protein" -now
>> I also used bigdcd (that's why the option -now and not -all was used
>> above) for reading trajectories and wrote wrapped dcds every 1000
>> frames to avoid memory problems (otherwise it's a ~30GB trajectory).
>> I also tried using the Join keyword on the first frame which had no
>> significant effect.
>> What procedure would you advice me to use in order to get a perfectly
>> re-centered and wrapped trajectory in the case of my multi-chains
>> system?
>> Thanks you in advance for your help,
>> NM