I am working with a coarse grained system of a multimer of a peptide on top of a lipid bilayer. I see that after some simulation time, some of the peptide molecules are fluctuating too much and moving across the periodic boundary in the z-direction. As a result of this some of the peptides are moving across z-direction so much that their image particles are appearing at the opposite side. Usually this is not a problem, but in this case very high amplitude jumps of the peptides are resulting interactions between the image peptides and the lipid bilayer, which is highly unphysical behavior. For example, the peptides are initially towards the extracellular side and due to large jumps across periodic bounderies, their images are appearing from the opposite side very close to the lipid bilayer, very much within interaction cutoffs. If the peptides move frequently much further away along the z-directions then the image particles will enter from the other side and may have short distances with the lipid bilayer, that are within the intermolecular cutoff values used in gromacs.
I have tried with increasing z box length, but did not work well. Now I am trying one more thing where I have shifted the peptide+lipid-bilayer system from the center of the box to close to ( < 3 nm) one of the edges of the box in z-direction. In this way I can have a large amount of space along z i.e. the gap between the system and one edge of the box along z-direction will be high, which may prevent the image particles appearing from the other side. Does it matter whether we have the system at the center and symmetrical water layers on top and bottom?
Firstly, I would like to understand if the method I current method that I am using is that going to make any difference?
Any other method to restrict the large jumps of peptides, is welcome, thanks in advance.
P.S. I just saw a post talking about application of sometime a flat bottomed potential to restrict large movement of biomolecules sometimes. Is that something we can apply here as well?
I don’t understand exactly what your problem is. Movement across the periodic boundary should not make any differences at all. The location of the peptide+lipid-bilayer system should not affect your results either. The periodic system is repeated continuously.
Is there a problem with the peptide interacting with the bilayer?
If you want to simulate a system where one lipid barrier leaflet is “intracellular” and one leaflet is “extracellular”, the best way is use two separate lipid bilayers, separated by water layers that are thick enough to prevent interactions between the bilayers. If you then put peptides in the “extracellular” water phase they will not diffuse to the “intracellular” water phase.
The system is: 15-16 peptide monomers are positioned on top of the upper leaflet of the lipid. So in this situation they are not supposed to interact with the lower leaflet lipid atoms. The issue is: e.g. if the box size is 10 nm, and one of the peptide molecule atoms which is moving ~10.5 nm further away from a lower leaflet lipid atom, then gromacs will consider interactions between the image particle of that peptide atom and the lower leaflet lipid atom as the distance between the two atoms = 10.5-10=0.5 nm well within the cutoff for various intermolecular interactions. Let me know if I am describing things incorrectly.
The suggestion that you gave that to use two lipid bilayers kind of sandwiched: a total of two separate systems having some separation along z-direction. System-A: bilayer-1+peptides <-------3 nm apart------> System-B:peptide+bilayer-2; however here we see the peptides are interacting way to much and the local density of the peptides around the lipid headgroups are decreasing a lot resulting in very weak interactions between the lipids and the peptides.
If the peptides are not anchored to the upper leaflet of the bilayer they will drift through the water and they will indeed start interacting with the lower leaflet across the periodic boundary.
With a single bilayer you would need large amounts of water to prevent that - enough so that the peptides don’t diffuse close enough to the lower lipid leaflet. But with long simulation times the box size might need to be very large.
I would still argue that the most efficient way would be to use two separate lipid bilayers, to ensure that you isolate the intracellular volume from the extracellular one. In the dual bilayer setup you describe you say that you get too much interactions between the peptides. I don’t understand why you would get more peptide interactions with a setup like that. Are the concentrations of peptides in water the same? Are the concentrations realistic? You say that you have a 3 nm separation on between the bilayers on the “intracellular” side. Do you use the same separation on the “extracellular” side? What about increasing both water layers to at least 4 nm? I.e., 2 nm water (plus suitable concentration of peptides) - lipid bilayer - 4 nm water - lipid bilayer - 2 nm water (plus suitable concentration of peptides)