You could define the stability in terms of binding energy of the complex. First you can do md simulations of each protein in water, then of the complex. A cheap way to calculate the binding energy is to use mm-pbsa or mmgbsa method. A good thing of these methods is the you can decompose the binding energy in contributions from each residue.

You can do md simulations using coarse grained models, namd and vmd have a residue based coarse graining model build in. You can find tutorials how to set up the system. It is also possible to map back to atomistic details after you finish with md simulations of coarse grained models.

Have you tried using the MARTINI CG forcefield with Gromacs? It's relatively easy to build the models and simulate.