I have using Amber 11 to carry out MD (via pmemd) on an array of five small (74 atoms each), self-assembling molecules in order to gain insight as to the self assembly process and its energetics. In the lab, I can induce these molecules to form vesicles or organic nanotubes. I have compared three slightly different approaches to the model:
1) MD, with iwrap=0
2) MD, with iwrap=1
3) MD, with iwrap=0, but re-imaged with ptraj post-simulation.
Each run is carried out in explicit water solvent, isothermally at 300K, and each run has been repeated twice…a total of 6 experiments.
The results are what I would expect intuitively…the energies look good, each dropping to about the same equilibrium value after 2ns of simulation. I observe groups of two or three of the molecules aggregating during the run, as would be expected in the actual self assembly process. My request for a little hand-holding is in the interpretation of these observations:
A. Although the results are intuitively correct, none of the 6 experiments shows an identical result when viewed in VMD. Some runs end with a group of three molecules aggregated. Others end up with two. But the conformations are different in each. Given the flexibility of these molecules, I am tentatively concluding that I am modeling a system with a number of thermodynamically accessible minima that accommodate aggregation, and am seeing several of them in these experiments. But is it possible that this is also a result of having ig=-1, as recommended in the MD protocols?
B. With iwrap=1, or using ptraj to re-image the initial run, I can observe the molecules jumping from one side of the water droplet to another as they are re-imaged around the center of gravity. Again, this is a reasonable picture for a molecule moving from one periodic box to another (i. e., moving through solvent and finding other similar molecules). However, some runs pull a molecule away from an aggregating group and place it at the back of the solvent box while the rest of the conformation is unchanged. Other aggregates result from a molecule jumping into the vicinity of an associated pair during the re-imaging process. Is this a reasonable representation of what might be happening as a molecule moves through the solution? I. e., per the manual, iwrap=1 "can mess up diffusion and other calculations". It's not clear what types of diffusion experiments are meant, but I am certainly looking at small molecules traveling through solution.
C. For experiments 1 and 2 above, I have had the .mdcrd files written in the NetCDF format, per an earlier discussion. I also used NetCDF for experiment 3. However, when I use ptraj for post processing, it appears to convert the files back to ASCII format, since VMD now fails to reload the re-imaged .mdcrd file in NetCDF format. The manual (p. 212) implies that this is what should happen, but is there a flag that could/should be set to keep the files in NetCDF format after the ptraj step? I mention this because it means that my "handling" of each file after the simulation isn't quite identical.
Again, undoubtedly very basic questions…I have been learning to speak Amber for about 3 months now, and my knowledge is clearly swiss-cheesed. Thanks for any guidance, as I'm reaching a point where I want to explain this to my group without making a total fool of myself!
Joe Bozell
University of Tennessee
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Dec 20 2010 - 14:00:02 PST
