Recently, I had an idea to simulate hydrated bilayers. The model that I am interested on is a double bilayer with a water layer in-between the bilayers. Something like (bilayer-water-bilayer). I have attached the graphic of the model for better view.
But I have some doubts related to the model that I am interested on and I hope I could get some help to shade some light on this. I would like to know if this model is technically correct concerning the periodic boundary condition.
Extra information (optional):
Actually, I have simulated a test run and I got the bilayer thickness value which is almost 3 angstrom higher than the experimentally reported value for multi-bilayer in hydrated condition. I used the definition given by Nagle, (Nagle et al., 2000, Current opinion in structural biology, 10(4): p. 474-480.) to calculate the bilayer thickness as the sum of (Luzzati thickness) and (steric water thickness).
My focus on this model system is primarily concern the water and head group interactions. As the initial guess, I feel this model does not represent the actual system. But, on the other hand, I feel this model could give some insight relating to the water and lipids from the two mono-layers along the water layer between two bialyers.
Even though I gave the details at some extend, I appreciate if someone could help to give some feed back if this model is wroth simulate.
The protocol for model building of lipid-containing systems and their simulations are not well established in my opinion, at least in comparison to protein or DNA simulations. You are right to question if your model represents the "actual" system that you are studying. There are many things that go into this, including how you set up your PBC, as Uttam pointed out, to the choice of force-field equation and parameters. I know in the AMBER community, many researchers are applying a constant surface tension to keep the surface area of lipids to experimental values, while I think the GROMACS community does not do this. In our work, we found that our bilayer model does converge to experimental values without the application of an artificial surface tension, but the time scale to achieve this if VERY long.
Since this is not standard protocol for lipid modeling, I think you will need to try different things (both with model construction and simulations settings) until you obtain results that you can validate using experimental findings. Statistical convergence in your MD observables will be an important thing to consider, especially since such a large system will take a while to reach equilibration. I would suggest measuring additional observables (e.g. head group surface areas) to help validated your model and simulations.
Best regards,
K
The protocol for model building of lipid-containing systems and their simulations are not well established in my opinion, at least in comparison to protein or DNA simulations. You are right to question if your model represents the "actual" system that you are studying. There are many things that go into this, including how you set up your PBC, as Uttam pointed out, to the choice of force-field equation and parameters. I know in the AMBER community, many researchers are applying a constant surface tension to keep the surface area of lipids to experimental values, while I think the GROMACS community does not do this. In our work, we found that our bilayer model does converge to experimental values without the application of an artificial surface tension, but the time scale to achieve this if VERY long.
Since this is not standard protocol for lipid modeling, I think you will need to try different things (both with model construction and simulations settings) until you obtain results that you can validate using experimental findings. Statistical convergence in your MD observables will be an important thing to consider, especially since such a large system will take a while to reach equilibration. I would suggest measuring additional observables (e.g. head group surface areas) to help validated your model and simulations.
Best regards,
K
A lot of this depends on the quality of the force field you are working with. If you want to match experimental data, you should be using the latest revision of the CHARMM force field, version 36, or the modifications proposed as "Stockholm Lipids", or perhaps even our model (if you want to use a united-atom representation). The AMBER lipid parameters, in my opinion, have not undergone the same level of optimization as the other models I described.
CHARMM 36
Klauda, J. B.; Venable, R. M.; Freites, J. A.; O’Connor, J. W.;
Tobias, D. J.; Mondragon-Ramirez, C.; Vorobyov, I.; MacKerell, A. D.;
Pastor, R. W. J. Phys. Chem. B 2010, 114 (23), 7830−7843.
Stockholm Lipids
Jambeck, J. P. M.; Lyubartsev, A. P. J. Phys. Chem. B 2012, 116
(10), 3164−3179.
TraPPE-UA Lipids
Bhatnagar, Kamath and Potoff, J. Phys. Chem. B 2013, 117, 9910−9921
Regarding stacked bilayers, we simulate them all the time, and one has to careful with the protocol or you can get all kinds of artifacts. One needs to start with a very well equilibrated single bilayer (50ns +), stack it, hydrate, run NVT to get things to settle down (1-2ns), and then anisotropic NPT (where the box lengths can fluctuate independently). If done correctly, we find that the structural properties, such as deuterium order parameter, of the stacked bilayers are very close to those of single bilayers. Water structure may be different, hence the interest in stacking bilayers...
So much thanks for those given some points to think on this issue.
While, I should explain a bit from where I get idea for building this model . Recently, one of the reviewer of our journal pointed out that bilayer simulations in anhydrous condition (without water layers) is technically inappropriate concerning the PBC application. He emphasized that for anhydrous systems, the model should contain two bilayers (double bilayers) to overcome artifacts cause by PBC.
Taking his point of view I tried to model double bilayer with water in between. (I have attached a newer version of graphic to illustrate this more clearly). The simulation box is shaded with yellow color. The experimental data I wish to calculate is the "d".
As I said, my concern is to study the water and headgroup interaction of these lipids in the middle of the bilayer, L2 and L3.
I appreciate if any one could explain a little on why this model possibly go wrong.
- Badges
- Science topic
- Similar topics
- Chemistry
- Theoretical Chemistry
- Computational Chemistry
- Molecular Dynamics