Thursday, October 25, 2007

Design brainy babies an easier way?   posted by Razib @ 10/25/2007 12:08:00 AM

Back when this blog was young and its was age measured in months, not years, "godless capitalist" (gc) would debate Paul Orwin and Charles Murtaugh, especially on the issue of QTLs which affect normal variation in IQ. When gc made the case for possible genetic engineering of one's offspring to have higher IQs Murtaugh was aghast. His basic logic seemed to be that the QTLs which contribute to normal variation in IQ are of small effect, so there are many, and who knows what sorts of problems might be caused by "switching" dozens of genes from state A to B? Genes are often of course pleiotropic, and Murtaugh's assumption seemed to be that changing the genetic architecture in such a profound way might not be a good thing in the genetic background of the typical human. We're a species with a very high rate of spontaneous abortion, perhaps as much as 3/4 of fertilizations do not come to term. Much of this is likely due to chromosomal abnormalities, but there are likely other factors as well, so let's take Murtaugh's objection seriously for a minute.

What sort of superior child do most parents which frank eugenic inclinations want? Wouldn't you want your son to be both tall and extremely intelligent? For example, a male who is 1 standard deviation taller than the norm and 2 standard deviations more intelligent would be at a relatively advantage in life. But the chances of having a tall Mensa level IQ son is not high for most humans, and even if you assume some dependence of the deviation of one trait conditioned upon other, that dependence is likely still relatively weak. Finally, both of these are quantitative traits where the average effect from a given gene seems small in their contribution to normal human variation. The right flavor of HMGA2 gives you 1 extra centimeter, but that really isn't that much, and this is likely a QTL of very large effect for this trait (height). For intelligence the prospects may even be weaker. To engineer a very tall and intelligent son if you are of normal intelligence & height (the typical future consumers) would require alterations on many loci, and this is where Murtaugh might pipe up with cautionary tales.

But then I thought of something: there are other traits where most of the variation seems associated with a few loci of large effect. Europeans' light skin is in large part due to SLC24A5, SLC45A2, TYR and OCA2. 4-6 loci probably account for around 90% of the variation. What about eye color? OCA2 is responsible for 3/4 of the variation, with TYR rounding out much of the balance. Skin color, eye color, even hair color; these are gross outward phenotypes controlled by a few loci of large effect! The loci are of such large effect I think that South Asian couples of middle complexion who want lighter skinned offspring can now feasibly engage in selective abortions to "load the die" so that their offspring are the "optimal" combination of their genes. We have the information and the technology. Then I began thinking, do people really care about total genome content in relation to their offspring? We've floated the possibility of switching a few hundred loci to shift the expected phenotypic value in the offspring, but the Murtaugh objection looms in the background. But we already have genetic backgrounds which have been "field tested" for viability and health in highly intelligent people. Why not just use them and fiddle around with the loci which control superficial physical appearance!

What I'm saying here is that instead of taking the genetic material from one's own biological offspring and fiddling with hundreds of loci to shift the quantitative value of traits of interest such as height and IQ, why not create a clone of a tall and very intelligent person, and switch a few dozen loci so as to sculpt that individual so that it can pass as your natural offspring? Imagine that a Japanese couple hosts a tall Swedish exchange student who is both a stellar scholar and athlete, and is moderately tall to boot. Additionally, this individual has a very agreeable personality. The son they always wanted! Perhaps they can get that son. They could clone the student, and then make changes to complexion, hair form, eye color, nasal form and include in Asian traits such as the epicanthic fold. Eye color, hair color and skin color are known to some degree now, perhaps a dozen genes could do most of the trick. But what about nasal form? I don't honestly know. Epicanthic fold? Again, I don't know. I suspect that some of these traits are subject to QTLs of larger effect than height or intelligence. One would have to do the cost vs. benefit.

Now, some of you might ask, "but why would people want to have offspring who are predominantly not descended from them?" Perhaps that is an issue for many or most people. Honestly though, I think if you could make a child resemble the midpoint of both parental phenotypes in terms of complexion and facial features the intellectual (conscious) knowledge would quickly be overruled by the reflexive (proximate) cognitive processes which would identify the physical resemblance and induce the normal emotional response (the main objection is that I do think that personality ticks are highly heritable, so perhaps some parents would start to treat their offspring as incongruous impostures who exhibit the right look, but with strange mannerisms). This goes to human psychology, it is a complicated area and there is some evidence that humans exhibit essentialisms which may transcend morphology. That being said, in this case I believe not all loci are created equal. If the "important" loci (those which contribute to visual parent-child resemblance) are identical by descent & state from the putative parents I think that many would enter into the tradeoff of alien genome content for the sake of building a "better baby."

Anyway, just a thought I had on the way to the Post Office (I'm not shitting you!).

Note: Even if the Murtaugh objection does not hold, it might be cheaper to do what I'm talking about. I don't really take the objection that much to heart, humans mix & match genes across genetic backgrounds all the time when we mate. The key would be to get the exogenous material in early during development so any problems would lead to a spontaneous abortion. Obviously playing with the genetic architecture after the child had been delivered might be more problematic. Also, though purchasing the rights to someone's genome for your offspring might be expense, I don't see how it would be that much more expensive than eggs purchased for fertility clinics are today. Of course, the types of parents I'm talking about are probably going to be in line for androids too. Let your imagination fly.