A Better Answer To “Where Ya From?”

Geneticists Eran Elhaik and Tatiana Tatarinova have developed a fascinating new tool, which they call the Geographic Population Structure (GPS), that allows anyone to identify where their ancestors came from as far back as 1,000 years ago. The technology has a much greater degree of accuracy than previous methods:

Previously, scientists have only been able to locate where your DNA was formed to within 700km, which in Europe could be two countries away; however this pioneering technique has been 98 per cent successful in locating worldwide populations to their right geographic regions, and down to their village and island of origin. The breakthrough of knowing where the gene pools that created your DNA were last mixed has massive implications for life-saving personalised medicine, advancing forensic science and for the study of populations whose ancestral origins are under debate, such as African Americans, Roma gypsies and European Jews.

Jordan Pearson explains why the GPS is so precise:

The increased accuracy of the new model is based on a simple, if controversial, assumption made by the study authors: that race doesn’t exist.

“The model of races is incorrect and should be dismissed,” Elhaik told me in an email. Up until now, tracing genetic origins assumed that people could be typified as a mix of two to three defined races, presupposing a homogenous “European” identity, Elhaik said. “By contrast, GPS represents a paradigm shift in population genetics whereby all populations are considered admixed to various degrees.”

Admixing occurs when one gene pool mixes with another to create a whole new one. You can think of it like how primary colours mix to create new palettes and shades—“red” people from region A breed with “blue” people from region B, creating a new group of “purple” people, genetically speaking. What the study assumed, if you’ll forgive the analogy a moment longer, is that there aren’t purely “red,” “yellow,” or “blue” people in terms of genetic makeup; we’re all somewhere in between, and every population worldwide displays a certain amount of admixing.

Update from a reader:

I’m always keen to see new ideas and methods in human evolution and genetics, so checked out the clip from Elhaik.  It pinged my bullshit sensors pretty hard, so I asked a friend and PhD student in the Graham Coop lab (much more an expert on this sort of thing than I) his take.  He says “it’s plausibly a decent genetic clustering algorithm, and they present evidence that they do better than SPA, which is probably the best existing method for this kind of thing, but the hype is way over the top.”

He also sent me a blog review from an editor on the paper, a summary of some questionable reviews of the analysis service, and an allegation of intellectual theft including a response from Elhaik himself.

Briefly, there are a few big issues with this paper.  First, the concept of a single ancestral origin for an individual is obviously ridiculous.  We all have thousands of ancestors who lived 1000 years ago, geographically spread to varying degrees; this method at best finds an average between all of these, so their prediction for a lot of folks ends up in the middle of an ocean.  Second, the authors claim no competing financial interests yet started a for-profit company offering this analysis on the same day the paper was released.  Third, the paper uses methods first suggested several years ago on a popular genomics blog, but makes no effort to credit the source.

Finally, the idea that race has been considered a useful concept by any serious human population geneticist in the last decade is absurd.  The notion that this method is superior because it eschews racial divisions is a straw man.