Re: [AMBER] diffusion path length

From: newamber list <>
Date: Fri, 11 Jul 2014 21:41:00 +0100

Hi Daniel

Thanks for your help and suggestions that you always render. So saying this
I think following approach should work:

Actually I have some volume selection criteria that my script can handle
and tell which waters are there in that volume in each frame, thus having
this info I should do this:

Lets say at time t1 I know one water molecule in some volume I am
interested in. Then I should find time t2 such that this water molecule
stays continuously in that volume. This will give small chunk (t2 - t1),
may be only few ps, of trajectory during which that water stays
continuously. Now I can calculate diffusion using cpptraj using specific
mask on that water. Yes also I will consider if that water molecule falls
back in that volume.

Similarly this will be done for other waters starting at time t1 and thus
finally I will average and if standard deviation is very large then I
should consider changing volume size criteria.


On Fri, Jul 11, 2014 at 8:57 PM, Daniel Roe <> wrote:

> Hi,
> On Fri, Jul 11, 2014 at 1:11 PM, newamber list <>
> wrote:
> > I have some general query regarding diffusion. cpptraj diffusion command
> > says
> > 'In order to correctly calculate diffusion molecules should take
> > continuous paths'
> >
> This just means that you should not have any imaging artifacts in your
> trajectory (i.e. when your molecule "wraps" from one side of your cell to
> the opposite side), otherwise you will artificially inflate the "velocity"
> of that molecule for that frame.
> >
> > Ideally how long this path should be? infinite or some finite (if finite,
> > then how much?). I am asking because somehow I am interested to know
> local
> > diffusion in very small volume and as its a very small volume so one can
> > not expect the same molecule to stay longer inside that volume and
> > molecules will be exchanging quickly with molecules outside the
> considered
> > box. Thus smaller will be the continuous path it will travel if moves out
> > volume quickly.
> >
> If I understand what you are proposing to do, then it is currently very
> difficult to do with cpptraj. You can calculate diffusion from the
> mean-square displacement (MSD) using the Einstein relation:
> 2*n*D = lim(t->inf) MSD / t
> The key is that this relation is valid as t approaches infinity, otherwise
> your data will be too noisy; this is why the manual recommends averaging
> over many atoms (in other words calculate diffusion for all waters, not
> just one). What you would need to do is somehow filter out all waters
> except those in the area of interest. You could *maybe* do this if the area
> of interest is near some solute which can be used as a reference point,
> then you use the "closest" command to retain a certain number of waters
> near that area of solute. You would need to make sure you have enough atoms
> and enough frames so that you are getting a good estimate of the diffusion
> constant though; you would probably want to do something like see how your
> answer changes vs the number of waters you add in etc.
> Hope this helps,
> -Dan
> >
> > Any suggestions will be really helpful
> >
> > Thanks
> > JIom
> > _______________________________________________
> > AMBER mailing list
> >
> >
> >
> --
> -------------------------
> Daniel R. Roe, PhD
> Department of Medicinal Chemistry
> University of Utah
> 30 South 2000 East, Room 201
> Salt Lake City, UT 84112-5820
> (801) 587-9652
> (801) 585-6208 (Fax)
> _______________________________________________
> AMBER mailing list
AMBER mailing list
Received on Fri Jul 11 2014 - 14:00:02 PDT
Custom Search