Does it have to be nucleic acids? You can probably get a dataflow-like language using concentrations of various enzymes and substrates. And there are pre-existi

Does it have to be nucleic acids? You can probably get a dataflow-like language using concentrations of various enzymes and substrates. And there are pre-existing models for molecular pathways that may help figure out the "key" for such a code. (In fact, IIRC analysis of metabolite concentration variation tends to be modeled using systems of differential equations, so you could use neural nets to find optimal input states for simple or well characterized networks.) There are some problems with using nucleic acids, chief among which is that the central dogma of molecular biology isn't nearly as true (or as central) as we used to think. Don't know as much about RNA -- there certainly are RNA "machines" where the nucleic acid itself has a function and maybe those could be coded for specific functionality. RNA isn't as stable though (because it's single stranded and RNAses are everywhere) so it sounds hard to deal with experimentally, but I'm no expert.