Maybe it doesn’t.
I have not worked out the details yet, but I think proteins with Quaternary structure (lock and key type interactions), something like violations of normative expectation could be worked out.
Alternatively, something could be estimated for the violation of expectation of metal-binding proteins since the fold is easier to resolve. The simplest example of metal-binding proteins are Zinc Fingers which, when functional, must have a lock-and-key type interaction with DNA. CTCF is a zinc finger protein that binds to about 55,000 locations on chromatin to create regulatory loops. Haemoglobin has been on my mind too.
I don’t think there is a way to apply the law of large numbers to such situations as protein sequences elegantly, but one can still frame the issue in terms of a violation of some sort of physical/chemical expectation.
EDIT: 13:09 EST for clarity