Gene Therapy: What's wrong with the software metaphor?
Brian Hanley
2015-10-22 00:00:00

The hackers look at the earth and decide to do something virtuous – feed the homeless to improve life here. The hackers have worked out the instruction set for making Ketchup. It's an old instruction set, stable, and well proven. So, they make a trillion copies of this instruction set, put them in little seed-like pods, and shoot them at the roof of the factory. Maybe one pod in a hundred thousand makes it through intact to the factory floor, gets picked up by robots and taken to the ancient CNC machines, but that's still millions of copies.

The factory obediently sets much of its machinery to the task of churning out thousands of tons of Ketchup. Thousands of people die in Ketchup floods. Firemen break in and shut down the worst of the over-production, but for a year or two afterward, streamlets of Ketchup run out into the countryside.

Gene therapy is quite like that. First of all, our understanding of the genome is quite limited. Second, our best intentions are crude hacks. Third, the cell responds to an assault by DNA that has hyperactive genes. What is a hyperactive gene that is not part of the chromosomes the most like? It's like a virus. We don't understand exactly how it is brought under control, or how it is destroyed, but this definitely happens.

A piece of DNA can code for instructions to make a protein. In that very limited sense, it “programs” the cell for production of that protein. But it is the interactions of the proteins that are the “execution” of the “program,” not the simple translation of DNA into proteins.

As soon as that protein is produced, it becomes part of the extremely complex Rube-Goldberg machine that is the cell, and the much larger organism that the cell is part of. Cells are a strange combination of evolved elegance and bizarre dependencies. Each cell is filled with a gel containing a chemical soup surrounding large, continuously flexing proteins that interact.

We have primitive beginning glimmerings of understanding how various control genes work – about 30 years behind our understanding of protein translation. For instance, U6 can be used to program some suppression of a gene. But that also has problems with dose control, just like protein expression, and we have little experience with it.

For the larger organism, each cell is part of some organ (or system) that serves the whole body. A long time ago, it was found that the less healthy a heart was, the more ordered the firing of its cells was – the opposite of what the researchers expected. Chaotic dynamics is healthy. Chaotic dynamics also describes the thermodynamics of enzymes and other macromolecules.

Biology is messy, complicated, and wet. Computers are electronic devices that rigidly execute instructions with precision, and always do it the same way. Cells are not like that. Cells are massively complicated factories that have evolved over more than a billion years into these astonishing heaps of cooperating cells that think of themselves as individuals. There are somewhere around 37 trillion cells in an average human body, each one an exceedingly complex factory. Once in a while, one of those 37 trillion cells decides that it is the one cell that should get everything and live forever, reproducing its kind without limit, and cancer ensues, killing the entire heap.

All of this is why biologists tend not to be impressed much with ideas based on first principles. Since there is a great deal we don't know, and predicting everything that will happen is functionally impossible, biologists want to see the experimental results. Even then, animal models are not humans, and people have died because of that.