How to prevent impact of bilayer periodic picture on peptides?

System is made of membrane, small peptides, ions and water. Membrane is assymetric (lower and upper leaflet of membrane contains different lipids, and consequently different charge).

Peptides are usually placed above upper leaflet of bilayer to study their interaction (See picture 1). However, often attraction between upper leaflet and peptides are not so strong, and periodic picture of lower leaflet dominates and attract peptides outside of the box (See picture 2).

I tried to increase z-size of the box up to ~5 nm, but effect is still observable.

How to deal with situations like this, ie, how to prevent impact of membrane periodic picture on peptides?

Should I use pbc only in xy with 2 walls (I never tired that, just read about) or take different approach?

This is not the drift due to pbc? what about membrane only system?

Thanks, but simulations of membrane only is just normal.

Peptides are losely attracted by upper leaflet, few of them attaches to it, rest of them move “randomly” accros upper part of the box, so from time to time few peptides comes to the top of the box, and then crosses out of the box - so they appear in the lower part of the box (as in picture 2). I suppose in that position they are attracted by lower leaflet. So maybe it is not pbc “artifact”, and everything is normal?
I’d like to see some experienced researcher how they deal with or interpret this…

I have seen peptide on the other side of the membrane but all the time it was due to pbc issue. Is it Amyloid beta peptide? if yes, i have done it and didnt see such problems other than the PBC.
or may be the water thickness is small or you keeping the peptide at the top of the water box.

To circumvent this problem, you can create a flat bottomed potential for your lipids and your peptides. This would add a conditional weak repulsive force (only acting when lipids/peptides get close to the z-top/bottom). Flat bottomed potentials are described in the GROMACS manual under restraints.

I would add a z-height inverted biasing potential and for the reference file (just another gro or pdb) I would place all my lipid and peptide particle coordinates to the desired z height of separation. In other words, use vim to set all the (z) coordinates in my file to 0. Then feed this file as the -ref file in grompp. Also you need to define the flat bottomed potential in the ITPs of your lipids and peptides.

I am not sure if there is an easy example file to illustrate this principle, I do have something around, but it is for coarse grained systems, which might confuse you.