The simplest self-replicating system we know of that exists in some natural environment (not, say, in a carefully purified solution of energetic monomers) has billions of atoms. And self-reproducing proteins have never, to my knowledge, ever been demonstrated in any experiment.