August 12, 2011

Armenian population structure with autosomal STRs

It is unfortunate that what seems to be a good sized sample of regional Armenian sample was only analyzed with a small set of autosomal STRs. Hopefully, the sample may still be available in the future for more comprehensive tests.

Population codes can be found in a freely available supplementary table.

I had previously used a similar set of STRs with structure to estimate inter-continental admixture in a set of "pure", and widely differentiated individuals (Germans, Africans, Chinese). That suggested to me, that while the continent origin can indeed be guessed for most individuals, admixture proportions are extremely noisy. For much closer-related populations (Armenian groups and various other Caucasoid neighbors), I fear that we should be extremely cautious in over-interpreting the limited evidence that a small panel of STRs provides.

Here are also Dodecad v3 population portraits for the Behar et al. (2010) Armenian sample and the Armenian_D sample from the Dodecad project:

I do not have enough Armenian participants yet, to speak about possible substructure in that population, but, from the samples available so far, which are from several different regions, a picture of relative homogeneity emerges. I encourage individuals with 4 Armenian grandparents who've tested with either 23andMe or Family Finder to join the Project (send e-mail to

American Journal of Physical Anthropology
DOI: 10.1002/ajpa.21558

Regionalized autosomal STR profiles among Armenian groups suggest disparate genetic influences

Robert K. Lowery

The archeology and ethnology of Armenia suggest that this region has acted as a crossroads for human migrations from Europe and the Middle East since at least the Neolithic. Near continual foreign influx has, in turn, led to the supposition that the gene pools of geographically separated Armenian populations may have diverged as differing historical influences potentially left distinct genetic traces in the various regions of the Armenian plateau. In this study, we seek to address whether any evidence for such genetic regional partitioning in Armenians exists by analyzing, for the first time, 15 autosomal short tandem repeat (STR) loci in 404 Armenians from four geographically well-characterized collections (Ararat Valley, Gardman, Sasun, and Lake Van) that represent distinct communities from across Historical Armenia. In addition, to determine whether genetic differences among these four Armenian populations are the result of differential affinities to populations of known historical influence in Armenia, we utilize 27 biogeographically targeted reference populations for phylogenetic and admixture analyses. From these examinations, we find that while close genetic affiliations exist between the two easternmost Armenian groups analyzed, Ararat Valley and Gardman, the remaining two populations display substantial distinctions. In particular, Sasun is distinguished by evidence for genetic contributions from Turkey, while a stronger Balkan component is detected in Lake Van, potentially suggestive of remnant genetic influences from ancient Greek and Phrygian populations in this region.



Maju said...

I just discussed a similar paper on Basques (Young 2011). They share the methodology (a few STR sites as material, focus on one population within the wider Eurasian context). However the results are very different: with Armenians as focus, these are central, while with Basques as focus, these are extreme and make other European populations (Scots, Catalans) appear extreme as well, while the various polarities shown here with Armenians (Georgians, Moroccans, Egyptians) are all dumped in the same zone in comparison with Basques and specially Catalans.

True that the Young paper has a very limited non-European sample, while this one has a very extensive one instead (what makes sense because Armenia is in Asia, even if borderland).

Unknown said...

The reliance on such a small number of autosomal STRs permeates the duration of the paper. The majority of the deductions simply cannot be substantiated based on such a limited scope, regardless of the corrective methods employed . Such ambitious conclusions as the 'Turkic' affinity of Sasun or the North African connection are misleading at best and wholly irresponsible at worst. The authors also takes the liberty of making wide historical assumptions such as the Phrygian influx, all while disregarding alternative explanations and simultaneously protesting the "lack of anthropological and archaeological evidence". A confounding work ripe for potential abuse.

TwoYaks said...

The idea that 14 µsats is limited is amusing to me, in the wildlife field. For many species, you don't have more than 5-7 µsats that have been developed for them. We're wrapping up a project with 21 µsats, which people have told me is incredibly respectable. And yet we still manage to leverage good information about population structure from our smaller numbers! I'll concede more is better, though, especially when talking about as fine scale patterns as they're suggesting.