For people doing MD it is a common question to ask the effect of solvent on there simulation. MD packages allow you to do Gas Phase/Vacuum simulation. Here is a short tutorial to run NVE simulation in vacuum using Gromacs.
What is key for user to understand is the fact that almost all force-fields are parameterised for solvent phase simulations. Hence the validity of gas-phase simulation done is in question.
In vacuum simulation, integration should be reduced compared to solvent phase simulation. Periodic boundary conditions, non-bonded interaction cutoffs, temperature coupling and pressure coupling were all turned off. Center of mass translation and rotation around the center of mass are to be removed to avoid the fast spin of the protein. One should also constrain the hydrogen bond using algorithms like SHAKE/LINCS.
In case domain decomposition does not work, for large systems use Point Decomposition method with ‘mdrun’
mdrun -pd -s *.tpr -v
Here is the mdp file ‘nvevacc.mdp‘ that can be used to run gas-phase simulations. However the user must remember that the with non-bonded interaction cutoffs turned off the simulation results can not be directly compared with solvent phase reactions even on a qualitative level.