This is a bit of a long shot (both in terms of expecting something to exist, but also given BioStars usual areas of focus), but:

Is anyone aware of an approach (or ideally an existing tool), that will calculate a pI for a protein *complex*, given that EM density and/or atomic data is available?

To my mind, the process would probably be something along the lines of the usual electrostatic surface generation algorithms (a 'rolling sphere'), over the whole complex, though I imagine this is not going to be simple to implement, if possible at all.

Also, https://en.wikipedia.org/wiki/Protein_pKa_calculations#Using_the_Poisson%E2%80%93Boltzmann_equation