Friday, October 05, 2007

Racial DNA Profiling?   posted by Razib @ 10/05/2007 01:13:00 PM

Wired has an article up titled The Inconvenient Science of Racial DNA Profiling. The core of the story is Tony Frudakis, who has written papers such as Eye colour: portals into pigmentation genes and ancestry. The science is pretty easy to understand. You don't need that many genes to fix in on someone's likely racial identity. Take two genes, SLC24A5 and DARC. There are genetic variants in African Americans on these loci which are basically non-existent in those of European descent. Because African Americans are genetically admixed with about 20% of their ancestry being European they have a less unequivocal profile on these loci; but all you need to do is increase the number of genes to increase your confidence that you've made a positive identification.

The science is banal. The politics, less so. For example:

Besides the expense, many people who might benefit from DNAWitness either don't know it exists or are extremely skeptical that it works. William C. Thompson, Chair of the Department of Criminology, Law & Society at UC Irvine is a prominent expert on the use of DNA in criminal trials but was only marginally familiar with this technology. When I tried to describe how it works, he literally screamed at me, calling Frudakis a hack and a charlatan who obviously did not understand statistics.
But even the people one might think should be his biggest allies aren't supporting that, including Tony Clayton, the special prosecutor who tried one of the Baton Rouge murder cases. Clayton, who is black, admits that he initially dismissed Frudakis as some white guy trying to substantiate his racist views. He no longer believes that and says "had it not been for Frudakis, we would still be looking for the white guy in the white pick-up truck." But then he adds, "We've been taught that we're all the same, that we bleed the same blood. If you subscribe to the (Frudakis) theory, you're saying we are inherently unequal."

William C. Thompson has written about forensic DNA evidence in his research, so his anger must not have been driven by ignorance. And as for the idea that because two populations have different genetic profiles that means that they are "inherently unequal," that's depends on what metric you're using as a measure. A lawyer should know that all Americans have the same rights and equal worth before the law whether they are good at digesting lactose or tall, both by the grace of genes.

But think about what Frudakis is doing, he's taking genes and mapping them onto phenotypes. For example, if someone has a particular allele of OCA2 there is a strong conditional probability that they will, or won't, have blue eyes. Amazing! It changes everything! Who knew that blue eyes were genetic! Now we're all unequal...or are we? The fact is that everyone knows that skin color, hair form, eye color and a host of other traits are due to your inheritance, your genes. It just goes to show how magical most people view science in that they are terrified that now we know the genes which are responsible for the variation which we intuitively would assume was due to inheritance of different genetic variants in the first place. This goes to some extent to confirm A.W.F. Edwards' criticism of Lewontin's Fallacy, if we derive our normative prescriptions from science's understanding of nature our basis is fundamentally shaky because science does not deal in ultimate truths. Lewontin's obscurantism convinced everyone that genetic differences were trivial, even though they all noted the distinctive physical types which obviously derive from genetic differences. Now that genetics, or specifically genomics, is advanced enough to fill in the gaps and elucidate the genetic architecture underpinning distinctive physical types people are terrified because they rested in their presuppositions on Lewontin's Fallacy.

Finally, there is the issue of the practical use of these tests. As noted in the article genomic profiles aren't necessarily cheap, and they aren't necessarily value added in some cases when cost is taken into account. But, if we are likely going to see the era of the $1,000 genome within the next 15 years, how expensive do you think a kit which checks a few dozen markers will be? For most people science is magical, but when the wizardry becomes a banal part of the consumer life through technology people accept it as a having real truth value. All the expositions of genetic correlation structure are going to do far less in debunking Lewontin's Fallacy than the coming ubiquity of personal genomics.