This is an excellent tutorial on MM-GB(PB)SA:

http://ambermd.org/tutorials/advanced/tutorial3/

Assuming you performed MD in AMBER, it will be very easy to do MM-GB(PB)SA straight away. If you've used GROMACS, there are plugins for these types of calculations, see

Article g_mmpbsa--a GROMACS tool for high-throughput MM-PBSA calculations

and

http://rashmikumari.github.io/g_mmpbsa/

Also, you can convert GROMACS topologies and trajectories to AMBER (and other) format(s), and then do MM-PBSA in AMBER.

There's an excellent tool by Rashmi Kumari et al. from Jawaharlal Nehru University, New Delhi, India. It's called g_mmpbsa. Although, there are certain drawbacks of the package, you can use it smoothly on your trajectories (xtc files).

Try it. http://rashmikumari.github.io/g_mmpbsa/

Good luck...

I also had the same question. Thanks for the answers. HOw much computationally exhaustive are these calculations? On a CPU which gives you 7ns simulation for 250 amino acids protein in a solvated system, how long could the MMgbsa calculation take?

On a typical protein - protein structure, MM-GBSA takes around 20 seconds per frame.