The initial equilibration of a membrane protein system is often the most difficult part of the simulation. Depending on how the system was constructed (there are other methods, aside from those described within this tutorial), problems can arise. The most common problem is the presence of voids around the lipid headgroups. The pre-equilibrated bilayer structures that are available for download online will have water stripped from the structure during the InflateGRO step. Using genbox, as described in this tutorial (see Defining the Unit Cell & Adding Solvent) does an adequate job of replacing these waters, but often times (depending on the van der Waals radius used for carbon in this step) voids remain around the lipid headgroups. The absence of proper solvation may lead to several possible behaviors:

• Collapse of lipid headgroups into themselves due to attractive and repulsive forces in this highly-charged region
• Separation of the bilayer (a void develops in the hydrophobic core), since the lipid headgroups are strongly attracted to the aqueous solvent

These problems will manifest themselves clearly, either through LINCS warnings as molecules are sheared apart, by distortions in the unit cell, or by the development of voids within the lipid core. So how does one overcome these issues? There are several options:

1. Use position restraints on the lipid headgroups in the vertical (z) dimension. Doing so allows the lipids to re-orient within the plane of the bilayer, but suppresses the tendency to pull apart from each other, behavior that would otherwise lead to a void in the hydrophobic core of the bilayer. A position restraint file can easily be created using genrestr, or manually created with a text editor. An example that can be used with DPPC can be found here. The file restrains the lipid P8 atoms in the x-y plane. Obviously, to utilize this file, one must add the following lines to the system topology after the #include "dppc.itp" line.
   

   #ifdef POSRES_LIPID
   #include "lipid_posre.itp"
   #endif
   

   You will need to add -DPOSRES_LIPID to the define keyword in the .mdp file (in addition to -DPOSRES for the protein) for this feature to function properly.

2. Use simulated annealing to slowly warm the system under NPT conditions. Why NPT? In some cases, rapid changes to the positions of the lipids and development or closure of solvent voids severely distort the dimensions of the unit cell. Using an NPT ensemble overcomes some of these distortions. A basic .mdp file suitable for this simulated annealing can be found here. Note the use of position restraints in conjunction with the annealing protocol. Using restraints allows water to slowly soak into the voids around the lipid headgroups.