1. For the RMSD you better use gmx trjconv to handle periodic boundary condition first. You may need to use -pbc cluster for the two big molecules.
2. For the antigen, if it's a protein just use the protein forcefield.
3. Provide more details on what exactly you have done when you calculate the interaction energy, "不太理想" is not specific enough. There are several ways to calculate, depends on the type of system some maybe more preferable than others.