THE GENOME CHANGES EVERYTHING (p6)
A good example is short sight. In a society where only half the people are literate, the correlation between genes for short sight and short-sightedness is quite poor. In a society where everybody's reading books as a child, the correlation gets much better. So when the environmental factor, which is early reading, becomes stronger, the genetic variability becomes stronger. So actually what these studies are picking up is that the environment is good enough in American society to bring out the genetic variation between people. On the whole you're holding it constant. You're sending them to similar schools, giving them similar curriculums, and giving them similar toys to play with and similar television stations to watch. So you're bound to pick up the genetic differences. However, the twin studies have done a fantastic job of proving heritability of things like personality in particular societies. Going from that to finding out which genes are involved has proved immensely disappointing. There's no question that this is a huge failure. A lot of people ten years ago would have said it's now going to be comparatively easy to start walking down the genome, hunting the actual genes involved in extroversion or neuroticism, and it just doesn't work. Endless results show a small positive effect and then vanish, because it turns out either that the effect is associated only with one population, or it just doesn't replicate. Why is that? Is it because there are so many genes involved in these things that you can't pick out the ones with very small effects? Most of them do have very small effects. I don't think so; it's subtler than that. What's happening is that you're getting gene-environment interactions that are under the radar of the normal gene-hunting techniques.
nice example of this, which is still quite a controversial study,
is Terrie Moffitt's work on antisocial behavior and the mono-amine
oxidase-A gene on the x-chromosome, which is going to set the standard
for how to understand the genes involved in personality and behavior.
I write about it in Nature via Nurture. She's done a study
of a cohort of New Zealanders in Dunedin who've been followed ever
since birth. All the kids in this town were followed every year of
their life to see what happened to them. It's about a thousand kids.
If you take the 400 boys in the sample who have all-white genetic
ancestry up to the grandparent level—boys because we're talking
about a gene on the x-chromosome—and you look at their mono-amine
oxidase-A gene, and you look at whether it's the high-active or low-active
version—there are essentially two versions of this gene according
to how active they are, according to whether the promoter on the
front of the gene has got a certain number of repetitions or a lesser
number—does the less active version of the gene correlate with
ending up a young adult who is antisocial and who's in trouble with
the law? No, it doesn't, in significant correlation. If you then
break the data down, though, into those who were abused in their
childhood and those who weren't, you find a very strong correlation
with this gene. It turns out that if you have the low-active version
of this gene, and you had an abusive childhood, then you're going
to end up with an antisocial adult—not deterministically, but
with a high probability. That seems to me to be a terribly important
study, because it shows that when you parcel out the gene-environment
interaction, you can find genes in here that you wouldn't have found
with the conventional gene-hunting techniques—genes that correlate
with behavior, but that react to the environment.
If you go back to before the first two months of 1953, and ask yourself what people thought life was, you find nobody with anything like the right guess. Absolutely nobody is talking in terms of a linear digital code, until the morning of the 28th of February, 1953, when Jim Watson puts the base pairs together, and suddenly the idea of spelling out an infinitely long, infinitely variable, but completely faithfully reproducible code falls into place. You can say that Schrödinger used the term code script at one point, but he talked much more about quantum mechanical ideas and things like that. There were ideas that the secret of life was going to be some kind of piece of chemistry, a piece of energy, or a piece of quantum mechanics. There were all sorts of ideas out there, but nobody thought it would have anything to do with linear digital information, like we use in books, strings of alphabetical letters. That is why that is such an important moment, not because the thing was shaped like two spirals—that's just aesthetically pleasing—but because the world changed on that day. It took a long time for the world to realize it had changed, and Watson and Crick got invited to give zero seminars in Cambridge during the next three years, which is worth remembering, and there was nothing in the newspapers about it. 1953 was better known for many, many years as the year when Everest was climbed, the Queen was crowned, the first issue of Playboy was printed, and all these other tremendous anniversaries. But in retrospect we can see that it doesn't really click with the population at large until O. J. Simpson and Monica Lewinsky put DNA on the map in the '90s. It's forensic DNA, Alec Jeffries' discovery of DNA fingerprinting, that really brings it home to people what we're talking about here, which is a bar code, a message.
uncanny the way Turing and Shannon and all these people come together
with ideas of computability, digital information theory, and cybernetics
at around the same time as DNA falls into place. Suppose the base
pairing mechanism of the double helix had been discovered in the
1920s, which is not totally impossible. The x-ray diffraction stuff
wouldn't have been possible, but it's conceivable that a chemist
could have worked out what was going on in DNA without x-ray diffraction.
In the '20s, before computing, would we have even understood what
we were looking at? Possibly not. Would we have been able to imagine
one day reading it, and having the storage capacity to decode it?
Or the other way of looking at it then is to suppose that DNA happens
on schedule and we invent machines for sequencing DNA, but we haven't
actually got computers by the '90s. How do we store the data? Do
we have a lot of clerks writing it down instead of computers? It
is wonderful the way the two branches of information technology,
one called life and the other called electronics, fall into place
at the same time. I don't understand how that kind of serendipity
works in history, but it's an intriguing one.
There's no question that the discovery moves in silicon now. In other words, a huge amount of the significant stuff that we do next has to be both understood inside a computer and modeled inside computers. The modeling of gene interactions is something that is beyond the power of a man with a pencil. It's going to require people who are good at systems dynamics. People who come out of business schools are quite good at this kind of thing. It's going to come from some funny directions. The economists are quite good at this kind of thing. The genome is going to turn out to be quite like an economy. When you adjust interest rates you have some effects here and other effects there, and then they have effects and they affect what affects interest rates and so it all feeds back on itself. A lot of genomic phenomena are going to turn out to be like that. So I do think that bioinformatics is the way a lot of this is going. You only have to look inside a molecular biology lab these days and see that they spend half their time comparing sequences on the Web with other sequences, pulling out sequences that are similar, saying, "Oh my goodness, this gene is like that one in fruit flies." But there's still going to be room for a lot of very important wet biology in this, particularly when you get inside the brain, because what's going to turn out is that the gross structure of the brain conceals immense amounts of detail about which nerve cells are talking to which nerve cells, and the genes are going to be the key to finding out what's going on there. These alternatively spliced genes that seem to enable each nerve cell to have almost a unique bar code on it that tells it who it needs to link up with when it gets to its target. There's still room for some heroic biology in there.