DNA computers in vitro and vivo

| We show how DNA molecules and standard lab techniques may be used to create a nonde-terministic Turing machine. This is the rst scheme that shows how to make a universal computer with DNA. We claim that both our scheme and previous ones will work, but they probably cannot be scaled up to be of practical computational importance. In vivo, many of the limitations on our and previous computers are much less severe or do not apply. Hence, lifeforms ought, at least in principle, to be capable of large Turing universal computations. The second part of our paper is a loose collection of biological phenomena that look computational and mathematical models of computation that look biological. We observe that cells face some daunting computational problems, e.g. gene regulation, assembly of complex structures, and antibody synthesis. We then make simpliied mathematical models of certain biochemical processes and investigate the computational power of these models. The view of \biology as a computer programming problem" that we espouse, may be useful for biologists. Thus our particular Turing machine construction bears a remarkable resemblance to recently discovered \RNA editing" processes. In fact it may be that the RNA editing machines in T.Brucei and other lifeforms are clonable, extractible and runnable in vitro, in which case one might get a far better performing Turing machine than all constructions so far, including our own. The fact that RNA editing is a Turing machine may in turn have a lot to do with the origins of life. We also have a computer science explanation for \junk DNA."