The hypothesis is that genes tend to travel together, and that it's possible to assign a human being living today to a geographically isolated population in which a characteristic collection of genes was very common. That assignment is a "race".
This is a statistical model of race. Imagine a 'gene-space' consisting of (say) 100 or so marker gene values. If we treat this as a 100-dimension space then an individual human should appear as a point in this space. If we then add a dimension for frequency, we may "see" hills and valleys on this "surface". Those are "races". Most of us are somewhere on the flank of a mountain, but there ought to be (how can one resist the word?) "pure" folk at the peaks. Conversely we ought to be able to find folks living in "valleys" who are truly "unique". (Some say I'm "different", but I think they have something else in mind.)
Here's how Leroi puts it:
The New York Times March 14, 2005The full article tries to justify race identification as a way to improve healthcare. I'm skeptical. Maybe as an interim approach, but we'll do better with pharmacogenomics than race as a proxy for individual gene values. I'd call this an interesting hypothesis rather than something that's immediately useful.
A Family Tree in Every Gene
By ARMAND MARIE LEROI
... If modern anthropologists mention the concept of race, it is invariably only to warn against and dismiss it. Likewise many geneticists. "Race is social concept, not a scientific one," according to Dr. Craig Venter - and he should know, since he was first to sequence the human genome. The idea that human races are only social constructs has been the consensus for at least 30 years.
The dominance of the social construct theory can be traced to a 1972 article by Dr. Richard Lewontin, a Harvard geneticist, who wrote that most human genetic variation can be found within any given "race." If one looked at genes rather than faces, he claimed, the difference between an African and a European would be scarcely greater than the difference between any two Europeans...
Three decades later, it seems that Dr. Lewontin's facts were correct, and have been abundantly confirmed by ever better techniques of detecting genetic variety. His reasoning, however, was wrong...
.... The shapes of our eyes, noses and skulls; the color of our eyes and our hair; the heaviness, height and hairiness of our bodies are all, individually, poor guides to ancestry.
But this is not true when the features are taken together. Certain skin colors tend to go with certain kinds of eyes, noses, skulls and bodies... To put it more abstractly, human physical variation is correlated; and correlations contain information.
Genetic variants that aren't written on our faces, but that can be detected only in the genome, show similar correlations. It is these correlations that Dr. Lewontin seems to have ignored. In essence, he looked at one gene at a time and failed to see races. But if many - a few hundred - variable genes are considered simultaneously, then it is very easy to do so...
... Study enough genes in enough people and one could sort the world's population into 10, 100, perhaps 1,000 groups, each located somewhere on the map. This has not yet been done with any precision, but it will be. Soon it may be possible to identify your ancestors not merely as African or European, but Ibo or Yoruba, perhaps even Celt or Castilian, or all of the above.
... Hispanics, for example, are composed of a recent and evolving blend of European, American Indian and African genes, then the Uighurs of Central Asia can be seen as a 3,000-year-old mix of West European and East Asian genes.
... When the Times of India article referred to the Andaman Islanders as being of ancient Negrito racial stock, the terminology was correct. Negrito is the name given by anthropologists to a people who once lived throughout Southeast Asia. They are very small, very dark, and have peppercorn hair. They look like African pygmies who have wandered away from Congo's jungles to take up life on a tropical isle. But they are not.
The latest genetic data suggest that the Negritos are descended from the first modern humans to have invaded Asia, some 100,000 years ago. In time they were overrun or absorbed by waves of Neolithic agriculturalists, and later nearly wiped out by British, Spanish and Indian colonialists. Now they are confined to the Malay Peninsula, a few islands in the Philippines and the Andamans...