[gmx-users] Re: How to choose appropriate size for a sliding average for smoothing distances between molecule types

[Justin Lemkul]

On 1/11/13 7:18 AM, anna.duncan wrote:
> Thank you for your reply.
>
> You're right that I want to measure the protein-lipid interaction by
> determining the length of time that the lipid is within a certain distance
> of the protein.  However,  I also want to get a measure of the length of
> time that the lipid is continuously interacting with the protein, that is,
> whether a given lipid has lots of short periods of interaction with the
> protein or just one fairly long interaction time.
>
> For this purpose, the small fluctuations (what you might call thermal
> noise?) in the positions of, and distances between, a given lipid and the
> protein mean that using the raw data gives lots of short interaction times,
> whereas the global motion of the given lipid in relation to the protein may
> be very little.  I've attached an image that will hopefully make this
> clearer.  The image shows the minimum distance between a given lipid
> particle and the protein molecule over the course of the simulation.
>
> I am performing coarse-grained simulations, using MARTINI, so the
> fluctuations in position (and thus inter-particle distance) will be greater
> in magnitude than for atomistic simulations, I would have thought, seeing as
> the timestep is larger.  This may mean that the problem is more (or only)
> pertinent for coarse-grained simulations, although I'm not sure about this.
>
> <http://gromacs.5086.n6.nabble.com/file/n5004497/distance_thermal_noise_example2.jpg>
>

I think it all depends on how you define whether or not two coarse particles are 
in contact.  For atomistic simulations, the value of 0.6 nm is generally used, 
but for coarse-grain simulations I suspect the value can be much larger.  I 
still don't think smoothing the data is a valid approach, since doing so is 
rather specific to your system and dependent upon the interval at which frames 
were saved, but you need some reasonable criterion that defines a contact.  It 
seems to me from the plot that you should have reasonably long contact periods, 
even using a value only slightly larger than the atomistic standard.

Maybe someone with actual CG experience can comment.

-Justin

-- 
========================================

Justin A. Lemkul, Ph.D.
Research Scientist
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================