Oct 21 2011
What European DNA Can Say About Ancestry, Disease Risk, and Cultural Traits
All the drama in the European Union right now — the debt crisis, the North-South divide and the kvetching over the wisdom of a common currency — has reignited talk about the cultural divide between people on the continent.
But is the divide more than a political and cultural one? Is there a deeper difference among the people of Europe or those of European ancestry than whether Germans have a firmer handshake or are more punctual, or which nationality prefers a kielbasa to köttbulla?
There may be, and those differences can be seen in people’s DNA.
Although all humans are over 99 percent identical genetically, even in the tight geographic confines of Europe there is enough genetic variation that 23andMe researchers can use it to determine from where in Europe a person, or a person’s ancestors, came.
Using principal component analysis (PCA) and linear regression — statistical tools for processing and visualizing large, complex datasets — researchers at 23andMe analyzed genetic data from 3,000 customers who all had four grandparents from the same country of origin. When the results of the analysis are plotted on a two-dimensional graph, individuals of similar ancestry cluster together, and those clusters correspond closely to the geographic locations of the countries of Europe. Researchers at UCLA and the University of Chicago have found similar results.
In that research and the work at 23andMe the clustering shows that populations within Europe have evolved distinct genetic characteristics. Another team took a similar approach to map the genetic differences in East Asia.
This is more than just a parlor trick for data scientists. These maps illustrate that Europeans and people of European ancestry are not homogenous but are in fact they are genetically diverse. Those differences also manifest themselves in other ways — from physical traits such as eye color, to propensities toward certain diseases, and even social and cultural characteristics.
By comparing the ancestry inferred from their genetics to survey responses, 23andMe scientists have identified a number of physical traits associated with a person’s ancestral origin in Europe. In some cases what researchers found is very intuitive — people with Northern European ancestry are more likely to have blue eyes and blond hair, while the hair and eyes of people with Southern European ancestry are more likely brown.
Other associations were more surprising. For instance, 23andMe researchers found that a number of social and cultural traits were strongly associated with a person’s predicted genetic ancestry of origin in Europe. A self-reported diagnosis of alcoholism was more common than average among people of predicted Irish ancestry for instance, while people with predicted Balkan ancestry were more likely to describe themselves as extraverts.
And then there were more serious associations between predicted genetic ancestry and certain diseases. Using data from thousands of 23andMe customers of European descent, our researchers found that ancestry may be important in determining the risks for Parkinson’s disease and for basal cell carcinoma, the most common type of skin cancer. Among 23andMe customers with European ancestry our researchers found that those diseases were more common among those with ancestry from western and northwestern countries in Europe.
All of these associations and others that we found beg many questions. First, do they reflect actual traits among people of different European ancestry or are they simply a reflection of cultural stereotypes among 23andMe customers? More intriguing is the question of the extent to which these traits are truly influenced by genetics.
While the statistical correlations in the data are strong, we still don’t have a complete picture of why these traits appear to separate according to ancestry. Associations, of course, do not imply causation, so what exactly is going on biologically remains an open question.
See our gallery of Research Findings for PCA plots of ancestry and eye color, socio-cultural traits, and disease risk in Europeans.
I guess it depends on whether you are European or Colonial in your viewpoint about the differences between Europeans. I am European so I see less differences than you as a Colonial. For instance, most Europeans view Colonials pretty much the same way whether the Europeans have blue eyes or black hair or speak German or Greek.
The debt situation in the E.U has more to do with greed, self interest, and wanting people to solve their own problems rather than expecting handouts. Money problems and differences divides families and causes relationship breakups. Europeans are no different there than other humans.
We all know that the centre of blondness and blue eyes is on the east side of the Baltic Sea. It is not news. I am sure Von Linne and Blumenbach could see that in their by-gone days, long before Darwin or Rosalind Franklin’s X-ray of dna given to Crick and Watson.
I am more interested in the fact that non Europeans are in the Europe map, Syrians, Lebanese, Moroccans, Egyptians.. Obviously the claim that Europeans can be placed into their countries of origins based on having grandparents from the same country and ethnic group has flaws.
I think I am the person represented by MT. The placement amongst the Italians is okay but I am not Italian, nor have Italian ancestry for hundreds of years.
It would be cool if you posted the r-code you used to make the graphs publicly so people could put themselves in using data downloaded from your service.
It might be interesting to look at genetic mutations like lactose tolerance in your database. Or is that too early to see differences among Europeans.
Another mutation of interest might be one like Leiden factor 5 that leads to thrombosis. Was it first identified in Leiden in the Netherlands? Or maybe the discovery was coincidental to a particular region?
Hi Sean,
That’s a great idea; unfortunately the code is dependent on a number of software modules so releasing it as a stand-alone package is not really feasible. We are, however, considering potentially developing these and other analyses into new 23andMe features!
Hi Theodora,
We certainly do see interesting associations between genetic ancestry and other traits and conditions — these are just a few of the strongest ones — and may post more of these in the future. Regarding the factor V Leiden mutation, it was first discovered in Leiden in the Netherlands, but the mutation itself is distributed across populations of European descent at a frequency of about 5%. It is less common in Hispanic and African descent populations and very rare in East Asians.
Hi,
Can we find out whether we’re on that map, and if so, our positions?
I’m pretty sure I’m one of the Polish pins on there.
Which scientist at 23andMe should we pester for that info? :p
@Davidski,
We’ve been getting some interest from customers about allowing direct interaction with these analyses and seeing their own data on the maps, and we’d love to provide this functionality at some point in the future! At the moment, it is not feasible to provide individual-level information about the data on these plots (our research data is decoupled from customer-identifying data for security reasons), so we would need to develop a full-fledged feature on the website to provide that level of detail and interaction. We’re definitely considering it, but cannot give an exact timeline.
Caught this in Science. Awesome way to view disease risk around the world. geneworld.stanford.edu.
Thanks for such fascinating material. Although I don’t want any personal information (I would not qualify for the map) I am intrigued by the methodology. How are the coordinates determined? What do the country codes designate: ethnicity or country of birth (often the same if four grandparents are so ‘close’ I grant)?
This is very interesting work. However, while PCA and regression are valuable data-reduction tools, I think you would have gotten richer insights from running a correspondence analysis (the kind developed by Benzecri in the 1960s, and available from SAS and other software). It would have enabled you to find hidden associations within SNPs and populations, as well as correlations linking SNPs and populations. (For a short summary of the method, see http://www.micheloud.com/FXM/COR/e/index.htm).