There must be many more essential details to your study left unsaid. At the least there must be significant bench sources of data about similar interactions you're curious about. So armed with only a vague understanding of what you're trying to do i'd vaguely suggest constructing an argument based upon those prior studies as a gradient against your particular situation. If in doubt, join the crowd and set up a Bayesian network containing what is known and how it surrounds what you'd like to know.
Heey, if your API is not complexed with the macromolecule in that case you need to do docking in order to put your API in the cage of macromolecule. So in this process docking only helps in generating initial starting structure. Then you can do MD simulation either in implicit or explicit solvent models using any programme , I typically use Amber 14. in the post simulation analysis you can run free energy calculations such as MM/PBSA or MM/GBSA to see the overall binding free energy divided into electrostatic, vdw, solvation terms.
In case you have a complex then you dont need to do docking and you go straight to MD simulation. The interaction you see using Chimera, VMD, PyMOl or even with LigPlot softwares.