My feeling is that any truly successful Giant Immunity Engine is going to end up with in vitro processing, rather than in silico. The other way around is a bit like trying to simulate a modern CPU by using mechanical gears.

Every human thyroid-gland is doing a massive amount of information processing for pennies a day, generating and validating antigen combinations in a way which is almost cryptographic.