Dear @obZehn, thank you very much for your thoughtful and detailed feedback.
I definitely do not expect water molecules to enter the bilayer core. I tried removing them manually after solvation, but they reappear during equilibration. I’ve read about possible strategies like:
Flat-bottomed position restraints on water molecules,
Or using SETTLE constraints in the topology file (e.g., for water hydrogens),
But I’m unsure whether either method would introduce artifacts or negatively affect the membrane dynamics. Do you have experience with such techniques or any recommendations for keeping water out of the bilayer?
Do you think I may have contributed to this by setting compressibility = 0.0 in the z-direction? I found it based on a suggestion here.
My membrane is not a nanodisc. It’s composed of:
Upper leaflet: 128 POPC, 27 Cholesterol, 15 GM3
Lower leaflet: 110 POPC, 27 Cholesterol, 33 POPS
(170 lipids per leaflet, total 340)
I made sure the area per lipid (APL) is balanced between the two leaflets before and after equilibration and that the bilayer plane spans at least 1.5 nm more than the DNA helical axis on each side. Do you think I should still increase the membrane size?
You’re absolutely right to ask — no, DNA does not physiologically partition through a bilayer on its own. In experiments, we use naked DNA as a negative control.
My actual goal is to later simulate DNA + peptide complexes, where different numbers of peptides are bound to the DNA. The current naked DNA simulation serves as a baseline for PMF comparison, to show that the DNA alone has a higher free energy barrier compared to the complexes.
I’ve seen similar PMF studies done using symmetric bilayers. But in those cases, the neutral charge of the membrane tends to favor DNA over complexes. That’s why I’m using an asymmetric bilayer to reflect more realistic surface charge conditions.
I’m intending to pull the DNA along its long (helical) axis.
Regarding your point about applying force only to the center of mass (COM) of the DNA, I understand that this provides little control over the molecule’s orientation. I’ve seen COM-based pulling used in several studies, but I’m open to better alternatives.
Would you recommend a more localized pulling group or a more refined collective variable?
The membrane rotation (from z-normal to y-normal) is concerning.
I’m considering adding the following lines in the .mdp to suppress membrane rotation:
nstcomm = 100
comm_mode = angular
comm_grps = MEMB
Would this be appropriate for a bilayer system, or would it introduce unintended side effects?
Thanks again for all your valuable insights.