Re: Avoiding Rotation and translation of protein

From: Marc Q. Ma (
Date: Fri Jun 10 2005 - 11:35:04 CDT

Hi Blake,

This is important and I would like to share it on the namd-l. Thanks
for sending it to me off the line first :)

> Are you sure about this? From my understanding of Brownian motion / the
> random walk / diffusion, I would not expect the protein to remain fixed
> indefinitely in and NVE simulation without restraints. Even if the
> system
> has zero momentum in the lab frame for the entire simulation, and even
> if
> the protein starts with zero momentum, I would not expect the protein
> (or
> the solvent) to have zero momentum at any instant or on average. I
> would
> expect slow drift with a squared COM displacement (averaged over an
> ensemble
> of simulations) given by {6 x (the diffusion constant for the protein)
> x
> (the time interval)} in accordance with the Einstein relation.

You are right. Due to the solute/solvent interaction, the protein will
diffuse and thus will not stay at where it was. If the protein has an
initial non zero angular momentum, then it will rotate, with varying
angular momentum due to solvent/solute interactions.

If you simulate protein in vacuum in NVE, starting with zero linear
momentum, and assuming you have a very accurate representation of real
numbers in computers, the COM should not move at all, at least for the
first part of simulation. As the simulation goes on and on, the
rounding errors kick in and the momentum will not remain zero, and the
COM will move. The angular motion is very complicated since proteins
are flexible.


> Blake
> -----Original Message-----
> From: [] On
> Behalf
> Of Marc Q. Ma
> Sent: June 8, 2005 5:58 PM
> To: Blake Charlebois
> Cc: 'Leonardo Sepulveda Durán';
> Subject: Re: namd-l: Avoiding Rotation and translation of protein
> If the linear momenta are all zero in x, y and z directions, then the
> protein should not be translocated if the system is conservative. You
> should remove the COM motion of the protein anyway so that you have a
> zero momentum to start with.
> NPT systems are not conservative. You may try NVE microcanonical
> simulations after you use NPT simulation to get the system into an
> equilibrated state -- size, density, COM, energy, temperature .... all
> stable.
> Meanwhile, you may increase your water box size just to stay on the
> safer side.
> Hope this helps.
> Marc
> On Jun 8, 2005, at 1:39 PM, Blake Charlebois wrote:
>> The following thread may help:
>> There are several related (and possibly more useful) threads. You can
>> find
>> them with a Google search specifying namd-l as one of the search
>> terms.
>> You should read the section of the NAMD manual on wrapping
>> coordinates. You
>> may find some useful scripts on the VMD site for reversing coordinate
>> wrapping in a dcd file.
>> Does your protein consistently rotate in the same way? Does it have a
>> large
>> dipole moment? If so, you may want to increase your box size.
>> Blake
>> -----Original Message-----
>> From: [] On
>> Behalf
>> Of Leonardo Sepulveda Durán
>> Sent: June 8, 2005 11:59 AM
>> To:
>> Subject: namd-l: Avoiding Rotation and translation of protein
>> Hello!!!!!
>> I am performing a NPT dynamics of a protein. I use a 10A waterbox, and
>> 2 ions (neutral system), PBC and PME. At 400 ps, the protein begins to
>> rotate (I dont Know if its center of mass translates too),
>> furthermore, one residue comes dangerously close to the border of the
>> box. I would like to ask if there is a way to avoid such protein
>> movements in the simulation, or whether there is a way to reorganize
>> trayectory files later in order to obtain a dcd where this movements
>> are taken out. And if that is posible, what happens if one of my
>> protein's residue steeps through th boundary? Is that Dynamics lost???
>> Thanks

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:40:49 CST