Anything can move if given sufficient time.  People do things like unbiased micelle assembly all the time.  Your assumptions that larger molecules do not move much is incorrect, or at least perhaps influenced by simulations that are likely very short.  Even over the course of a short simulation (say, 10 ns), small molecules can move quite extensively.

I used to run an unbiased liposome assembly. I was using 3200 molecule phospholipid to simulate the liposome assembly for 40 ns. The liposome itself formed after 16 ns, with the size about 20 nm, but there is a couple stubborn micelle that won't move until the next 24 ns. Marrink et al managed to create a bigger (and perfect) liposome but using some constraint. Right now I'm running a bigger liposome assembly experiment (the liposome size is 32 nm) and even after 250 ns there is some stubborn micelle that won't budge from its position and won't join the liposome. But perhaps we'll leave it that way.

Anyway, those story above is not the experiment where I would apply the extensive solvent movement in MD, just to demonstrate what I mean by "seems to be static to me". The simulation that I would do is to simulate the protein entrapment by polysaccharide nanoparticle. So there are 2 polysaccharide with opposite charge, and then the protein that would be entrapped inside the nanoparticle. So in the real experiment the protein is mixed with the negatively charged polysaccharide then vortex for few second, then added with the positively charged polysaccharide and vortex again. I've already had a hypothetical nanoparticle formation mechanism, judging from the way it is made, the composition, and the TEM image, and I'd like to support the idea using MD simulation.

I have the idea of using "dummy plate" inside the system and move/rotate the plate so the whole molecule inside the system will move by the turbulence/shear force created by the plate. But I'm wondering if this is doable within NAMD or other MD package. I've heard that NAMD could be use to simulate nanodevice so I thought that this is could be possible, but I'm not sure.

This is really just an issue of time scale.  Think about the experiment - by your own description the mixing is done over several seconds.  Then the simulations, even if they are "long" by MD standards (250 ns) are still billions of times shorter than the real physical behavior.  MD simulations can become trapped in local minima; that's a known limitation of unbiased simulations.  Probably you're better off with a coarse-grained force field, for which simulations into the microsecond time scale are relatively cheap and quick to do.  Imposing some shear force and turbulence could be interesting, but you could spend a lot of time tuning this to even be stable, and then all you've shown is that your model does what you're forcing it to do, which really doesn't end up adding much value.

It is coarse-grained, and so far we didn't have the access to supercomputer or HPC (But maybe we'll have soon, we're working on it right now), it even took more than 2 months to complete 250 ns with our server (2 day 10 hours per 10 ns simulation). That's why I resort to such approach, besides it may gives good contribution to science community. Anyway you're quite right in so many ways, I appreciate your answer.

"Imposing some shear force and turbulence could be interesting, but you could spend a lot of time tuning this to even be stable, and then all you've shown is that your model does what you're forcing it to do, which really doesn't end up adding much value."

I'd like to hear your opinion more regarding stability issue. IF it is indeed possible, do you think which one is the weakest link? Is it the pressure instability? Bond potential energy (intramolecular energy)? Atom clash? Or anything else?

Finally found the right keyword, this one is close enough: http://www2.chem.uic.edu/pkral/paper/rotor.pdf

Wondering if large sheet of graphene could do the job.

Update:

And large shear rate is still allowed too:

http://journals.aps.org/pre/abstract/10.1103/PhysRevE.82.051602

Well of course velocity as high as 150 nm/ns could break the membrane stability. So I guess my plan is still make sense (0.2-1 nm/ns).

Dear Muhammad,

perhaps you already found the answers to your questions in the previous replies. Nevertheless,  let me give you my contribution.

In your original message you say something that sounds weird to me, when you say that big molecules don't move because velocities are assigned randomly. Indeed, the memory of these initial velocities is quickly lost, so this has little effect. Indeed, in absence of any external field/drift/etc, you molecules move according to a Brownian dynamics (...assuming that you are in a liquid). Check the mean square displacement (consider, for example, the center of mass of your molecules) vs time and you will discover that the trend is linear (don't forget to properly ensemble average, and average over the various molecules of the same species). The slope of this relation is twice the (self-)diffusion coefficient, which, of course, depends on the chemical species (size, chemical properties etc. of your molecules). So, the phenomenon you observe is perfectly normal, and understood on the basis of macroscopic continuum physics. 

Concerning the application of a shear force, I don't understand the phenomenon you expect to see. Perhaps I'm naive but you should preserve the equilibrium concentration of your heavy molecules (apart fluctuations) and so I won't expect any speed up of the aggregation. Moreover, if a speed up is observed, this would be an artifact of a unnatural increase of concentration.

In your original message you despise "constrained MD". Indeed, this is just one of the many possible acceleration techniques, which have a solid theoretical foundation, and are not mere tricks to force a system to do something it apparently does not like to do. These techniques have been already used to study crystallization, which is a particular, and particularly difficult, kind of aggregation. If you want to know more just let me know, and I'll suggest you some paper about rare events in general, and their application to nucleation/aggregation.

Ciao

Dear Simone,

Thank you very much for your comprehensive insight. Yes, I just recently found that you and Justin were right. All this time I thought that diffusion could be much faster, then I found that the displacement rate for 1 - 1000 nm particle is in the order of micron/s. No wonder that the movement is hardly apparent in 250 ns timescale.

Regarding the application of shear force, what I expect to see is the different velocity between polysaccharide and protein (because of the different surface to volume ratio), and thus a higher rate of collision and interaction between polysaccharide and protein. Hopefully the 1st polysaccharide will create the first layer around the protein, then the 2nd polysaccharide (with opposite charge) is added and given the same treatment and hopefully create the second layer and bind all of them together as one nanoparticle. The shear force is applied to address the mixing process between all components.

Whoa, "despise" is a strong word. No I'm not despise it. And I understand this statement of yours "and are not mere tricks to force a system to do something it apparently does not like to do.". I just thought that giving constraint to polysaccharide and protein so they would come together could possibly lead to "premature aggregation (forming nanoparticles)" because of the unnatural localized higher concentration. There are two level of aggregation, first is between polysaccharide and protein, and the second one is between all of the first aggregate (comprised of protein and 1st polysaccharide) which is facilitated by the 2nd polysaccharide, the second level aggregation is the one which would produce the nanoparticles. Or perhaps there is another way of applying constraint that could prevent this from happening?

Regards,

Radif

Dear Radif,

I still have a couple of questions to ask you, if I wish. Usually, shear flow is imposed via Lees-Edwards boundary conditions and the related machinery. This, I believe, does not change the average concentrations of the two species, thus won't increase the chances of sticking of one on the other. Moreover, when stationarity is reached, which happens rather soon, their drift velocity is the same. This is just the opposite of what you thought but I don't see how it can happen that the two species have different drift velocities, it would be against standard (proven and intuitive) conclusions of fluid dynamics: all the species in a liquid flowing with a drift velocity v_d have a velocity v = v_T + v_d, where v_T is the (random) fluctuation of the velocity of the particle off of the drift velocity, which determined (or is determined) by the  temperature of the system. Am I wrong??? Where??? Thus, I don't see how you can reach your objective following your plan.

Concerning enhanced sampling techniques, you might increase the concentration of one or the other species, or both of them, in a given domain of your computational sample, e.g. using as collective variable the smoothed number of particles in the domain. This amounts to fixing the (smoothed and coarse grained) density field of that species. The rate of the agglomeration process follows, I imagine,  something close to the classical nucleation theory (or something similar - you can give a look to the book of Kelton and Greer, Nucleation in condensed matter: applications in materials and biology). If this is true, you can extrapolate the rate at more realistic concentrations using (more or less) analytical formulas.

Actually, nucleation of multicomponent crystals is often related to the fluctuations of concentrations of their components (see, for example, our article on methane cleathrate hydrates: http://pubs.acs.org/doi/abs/10.1021/jp5052479). In other words,  investigating the Landau free energy as a function of the density/concentration of the two species might be a key element in elucidating the mechanism and kinetics of the process you are interested in.

Ciao,

Simone

Probably your problem is that you haven't reached equilibrium. Therefore, you need more simulation time, or you have to use more convenient initial conditions that will lead to equilibrium as quick as possible.

Dear Muhammad,

Making simulations at higher temperatures will increase speed of molecules. Controlling density can also have some effect. But I have to agree with the majority that more time is required to do such simulations. To my knowledge most molecular dynamics of macromolecules are performed with presumption of known interaction site or prior molecular docking studies.