I think it will prove computationally very expensive, both to solve protein folding and to subsequently design a bootstrapping automaton. It might be difficult enough for another method of assembly to come out ahead cost-wise.