I mean by that a model of computation in the TCS sense. (Some) molecules, be them DNA, RNA or other, are akin to lambda terms (or graphs), beta reduction (or other graph rewrites) are akin to chemical reactions. The algorithm of reduction has to be random, local (i.e. no higher point of view criteria for interaction are used), performed by Nature. All in all, to be able to prove that some part of the biochemistry can be seen, as a concrete asynchronous graph rewrite automaton (capable of universal computation) would be a great advance in biology.