GENE GENIE

Dick Pountain/17 February 1999/Idealog 55

I promised myself I wasn't going to talk about genetic engineering in this column: this is a computer magazine and I must try to stay 'on message' as they say nowadays. However everything conspires to break my resolve: the shifty behaviour of Monsanto and HM government, my reflex interest as an ex-biochemist, and the fact that I just thought up  a way to drag computers into it!

First I should clarify one point - as a general rule I'm in favour of progress. Even though I may gripe a lot, I embrace the new information technologies and have no time for techno-luddites. I have less time still for 'deep' greens who prefers plants and animals to their own species. I'm sceptical about many environmental scare stories (and all nutritional and dietary fads). I'm not a vegetarian, and I approve of experiments on animals to advance the biological sciences. And yet I can't swallow GM foodstuffs, and I'm going to tell you why.

My reasoning only partly depends on Monsanto Chemical's despicable behaviour - namely its reacting to an understandable desire among some members of the public not to eat GM soy products by mixing them with ordinary beans so no-one can tell which is which. It's not just that I dislike and distrust the chemical industry's long-term marketing goal (though I do) to create GM plant species resistant to certain manufactured pesticides and herbicides so they can sell farmers both the seed and the chemicals. No, my reasons for opposing the distribution of GM foodstuffs, indeed of any GM plants, are far more fundamental than that, and to explain them I'm going to use a computer-based metaphor (phew, back on message!)

There is more than a passing resemblance between a chromosome and a computer program. Both consist of very long strings of a few basic information-encoding units: nucleotide molecules in one case and binary opcodes in the other. In both cases the information encoded by these strings can be read by a 'machine', which provokes certain actions. In the case of DNA that action is to create proteins, which serve as components of muscle and other tissues and as enzymes that organise and catalyse the building the rest of whichever living organism out of raw inorganic materials. In the case of a computer program the actions might be to draw pixels on a screen, magnetize bits of a rotating metal disk, or to switch on a motor that shuts a door or opens a valve. The 'machines' that read DNA (to be more precise, RNA copies of the DNA) are cell components called ribosomes, themselves built of proteins, which act like millions of tiny Turing Machines running in parallel, translating RNA chains into chains of amino-acids (ie. proteins). The machine that reads a computer program is a metal-on-silicon machine called a microprocessor. Each DNA strand contains many different "programs" called genes, each of which generates a single type of protein. Computer programs are similarly divided up into many subroutines and data structures, each of which performs a specific action.

To be pedantic, I'm pushing things by calling ribosomes Turing Machines. Back in 1992 I corresponded with an American biochemist (also trained as a linguist) who works on the Human Genome Project using Prolog programs to parse DNA sequences into their constituent genes: he concluded that the grammar of DNA sequences lies somewhere between context-free and context-sensitive, either of which can be parsed by a Linear Bounded Automaton (less general than a Turing Machine). But, hey, it's only a metaphor.

Genetic engineering works by cutting out a single gene from one species - for example the gene that codes for the toxic protein of scorpion venom - and splicing it into the DNA of another, where hopefully it will be expressed (ie. executed) and produce its protein. Huge amounts of research are devoted to sequencing genes and proteins, using wonderful computerised machines that weren't invented in my day, and they also use lots of enzymes to help with the splicing (including that cauliflower mosaic virus promoter that appears not to be Sainsbury's Own-Brand). I'd be reasonably confident about eating a single protein, like chymotrypsin (ie. Rennet) that has been produced by a Genetically Modified bacterium grown inside an isolated fermentation vessel. They know the sequence and there are unlikely to be any nasty surprises.

We have similarly powerful tools in the programming world; code editors, compilers, symbolic debuggers, profilers, bounds checkers etc. So here's the cruncher for me. Take one copy of Windows 95 and one copy of Linux. Compile some routine that belongs in Linux (one whose purpose you understand) and using your debugger splice it somewhere into Windows 95 (if you are really smart at assembler programming, you might even fix up its parameter passing). If it was only a small routine, the chances are reasonable that Windows 95 might come back up, and may even appear to work properly, but one day something is going to go bang. Moreover there is no way that you could stand up in public and say that you believe otherwise; that you guarantee this code is workable; that you understand all the possible side effects. Hell, Microsoft can't guarantee that for its own patches.

That seems to me a metaphor for what Monsanto and others are doing whenever they release a whole GM organism (as opposed to a single purified protein) like a soy plant, or even worse a bacterium or virus, outside the confines of a laboratory. And that is why I want no truck with it; I prefer to keep the gene genie in a bottle.