GROMACS

By long convention, GROMACS expects membranes to be oriented so that the Z-axis is the surface normal. We have some other places of more-or-less surprising anisotropy (DD decomposition works best if the x and y dimensions are not much longer than the z dimension, some analysis tools partition the cell into boxes along axes).

At https://gerrit.gromacs.org/#/c/8314/3/src/gromacs/mdlib/mdebin.cpp we thought perhaps an .mdp option that declares this could be useful, although for now we would only implement e.g. the z option. At least then it is easier for the user to see that their choices are not free.