The Human Genome Project was completed almost two decades ago, and somebody solved the protein folding problem recently.
Why are we still doing genetics at the machine code level? Shouldn't we have some compilers, assemblers and linkers by now?
Why are we still doing genetics at the machine code level? Shouldn't we have some compilers, assemblers and linkers by now?
Take that piece of RNA. An intuitive mental model is that it's some form of "instruction" or a bunch of instruction, isn't it? It's also wrong, because it just encodes a protein that acts the way it does only because of its shape (that is, one of its potential energy local minimums) and the shape of other proteins around it. That shape is only weakly local, it can be affected by far-away sections of peptide sequence. So it's almost impossible to systematically break it down, you have to consider and model things as a whole , which is insanely complex both computationally and cognitively.
If you want a good mental model of how it works, imagine you assemble a thing from metal balls and springs. You take a few thousands balls and connect most of them with springs of different strengths. You then take this thing and throw it on the floor; it will assume a shape that is implicitly encoded in spring strengths, its environment, and the way you've assembled it. You can even make it change shape if you poke on it the right way. That's how biology works in a nutshell; it's a nightmare to design anything for systems like that. Again, you can't simplify and break down and encapsulate and abstract like you do in programming.