Thermodynamic Constraints on DNA-based Computing

Computing with biological macromolecules, such as DNA, is fundamentally a physical/chemical process. The DNA chemistry introduces a level of complexity that makes reliable, e cient, and scalable computations a challenge. All the chemical and thermodynamic factors have to be analyzed and controlled in order for the molecular algorithm to produce the intended result. For instance, a computation based on DNA requires that the problem instance be encoded in single strands of DNA and that these strands react as planned, that molecular biology protocols, such as PCR or a nity separation, correctly extract the result, and that su cient exibility remains so that worthwhile computations can be done. In this paper, various thermodynamic and chemical constraints on DNA computing are enumerated. A similarity measure, based on Gibb's free energy of formation, is de ned to judge the goodness of DNA encodings. Finally, the DNA computation problem for implementing molecular algorithms is de ned, and it is likely that it is as di cult as the combinatorial optimization problems they are intended to solve.