DNA Test - Your ethnic mix

[animalszoos]

If you took a DNA test to find out your ethnic mix, what do you think your results be?

For example, it could be (this is probably not mine):
- 20% West African
- 10% Pacific
- 30% Native American
- 5% Scandinavian
- 5% North African
- 20% Turkish
- 10% Iberian

 

[Shorts]

I mean this in the nicest possible way -I don't really care, it doesn't really have any relevance to who I am or how I live. (I do realise that certain ethnicities have higher risks of some diseases, etc. but I'll leave that to my doctors to worry about if it's relevant to me).

 

[Nikola Chavkosk]

If I can chose, I will chose to be a Brit or a German, white, with yellow hair

Otherwise, I believe I am closely related to Albanians, Serbians, Italians, Ukrainians and Syrians.

 

[CGSwans]

Biologically, race is almost meaningless. For example, if you have Western or Central European heritage I guarantee you are descended from Charlemagne. You have royal blood.

Don't believe me? Assume there's roughly 25 years between generations. Charlemagne lived in the late 8th century, so call it 50 generations ago. 2 to the power of 50 - the number of hypothetical ancestors you have from that time - is over one quadrillion. The odds that there was anybody alive in Western Europe at that time who has any descendants at all, but who is not your ancestor are so remote they can be safely ignored.

The same is true, albeit potentially on a slightly different time scale, for everybody with any Afro-Eurasian heritage. The movements of populations within those two continents have been so great that it's simply not credible to believe otherwise. Hitler will have had Jewish and African ancestry. The only exceptions are where populations have become isolated for a long period of time. Think the Americas before 1492. But I'd be shocked if there's anybody left in the Americas who has no genetic heritage from post-1492 migrants.

 

[FunkyGibbon]

Don't believe me? Assume there's roughly 25 years between generations. Charlemagne lived in the late 8th century, so call it 50 generations ago. 2 to the power of 50 - the number of hypothetical ancestors you have from that time - is over one quadrillion. The odds that there was anybody alive in Western Europe at that time who has any descendants at all, but who is not your ancestor are so remote they can be safely ignored.

I'm not so sure. I'd want to see some evidence either way because I think until recently a lot of people didn't really move at all. There would be genetic exchanges between adjacent populations, and occasional large scale movements of people, but I really think there's scope for a lack of common ancestors across 1200 years.

To make a specific example, I think there is some evidence that the Basque community was highly isolated for a long time, although this may be linguistic rather than genetic.

 

[FunkyGibbon]

Most Europeans share recent ancestors

It looks like I will have to retreat to the position that there may be some people who aren't related to Charlemagne (or we can choose a different ancestor because Charlemagne's DNA will have received preferential treatment).

The argument @CGSwans uses about 1 quadrillion ancestors for 50 generations makes me uncomfortable. For 25 generations that number drops to 30 million. That's conveniently the population of Europe at the time of Charlemagne. To put that another way, if we assume the population stays constant, and we might get away with that, even if Europe was just one large body of people freely interacting it would take 600 years for one individual's DNA to propagate throughout the continent. But if we model Europe as a series of smaller population fragments, with periodic exchanges of random DNA, then the higher growth rates 'up' the curve of exponential growth become less and less available.

I'm going to try this out, and see.

 

[Coelacanth18]

To put that another way, if we assume the population stays constant, and we might get away with that, even if Europe was just one large body of people freely interacting it would take 600 years for one individual's DNA to propagate throughout the continent

For example, if you have Western or Central European heritage I guarantee you are descended from Charlemagne. You have royal blood

I'm not sure if this rebuts either argument or just adds scientific clarity, but no individual's DNA actually persists that long. Each parent contributes ~50% of their genome to a child, which means each grandparent contributes ~25%, and so on. By the time you're about 8 generations back, each individual ancestor is contributing a miniscule, insignificant amount of their genome to that eventual descendant. Therefore, from a genetic standpoint, nobody alive today is actually "related" to Charlemagne more than they are to any other random person.

 

[Chlidonias]

I'm going to try this out, and see.

um... you're going to try to propagate your genes across a continent...?

 

[CGSwans]

Most Europeans share recent ancestors

It looks like I will have to retreat to the position that there may be some people who aren't related to Charlemagne (or we can choose a different ancestor because Charlemagne's DNA will have received preferential treatment).

The argument @CGSwans uses about 1 quadrillion ancestors for 50 generations makes me uncomfortable. For 25 generations that number drops to 30 million. That's conveniently the population of Europe at the time of Charlemagne. To put that another way, if we assume the population stays constant, and we might get away with that, even if Europe was just one large body of people freely interacting it would take 600 years for one individual's DNA to propagate throughout the continent. But if we model Europe as a series of smaller population fragments, with periodic exchanges of random DNA, then the higher growth rates 'up' the curve of exponential growth become less and less available.

I'm going to try this out, and see.

Here's the article. Bear in mind I'm not a scientist and have relied on reporting of its findings - I can't claim to have the skills to critically evaluate its claims. But I find them persuasive.

The Geography of Recent Genetic Ancestry across Europe

 

[FunkyGibbon]

Here's the article. Bear in mind I'm not a scientist and have relied on reporting of its findings - I can't claim to have the skills to critically evaluate its claims. But I find them persuasive.

The Geography of Recent Genetic Ancestry across Europe

The discussion in that paper seems very good, even though the science is beyond me. Basically, there has been far more movement in Europe than I imagined. My model, which assumed small towns with infrequent but regular contact with each other, would take about 5000 years to produce a Europe with universal common ancestors

um... you're going to try to propagate your genes across a continent...?

xkcd: G-Spot

 

[CGSwans]

The discussion in that paper seems very good, even though the science is beyond me. Basically, there has been far more movement in Europe than I imagined. My model, which assumed small towns with infrequent but regular contact with each other, would take about 5000 years to produce a Europe with universal common ancestors



xkcd: G-Spot

Itinerant tradespeople, soldiers, merchants, slaves, even clergymen with weak will power. It only takes one randy Crusader from Britain who stopped off in Italy on his way to the Holy Land and the ripple effects will be profound.

 

[gentle lemur]

I blame the Vikings! Raping and pillaging from Iceland to Muscovy and to Sicily. I expect that I have inherited some of their genes

 

[SealPup]

Race is just population. And if you look at the racial maps from the 20th or 19th C they match genetic clusters when mapped out.

"Race does not exist" started as a taxonomic issue then acquired moral and political baggage. It feels odd to try and justify the classic races of typology, when subspecies are hard to justify, really. But with that baggage came a wierd desire on the part of some people, to try and convince people what we can all see, is somehow untrue and "just skin deep".

 

[Jurek7]

For example, it could be (this is probably not mine):
- 20% West African
- 10% Pacific
- 30% Native American
- 5% Scandinavian
- 5% North African
- 20% Turkish
- 10% Iberian

Sorry but this is pseudo-scientific nonsense. Human populations differ in certain DNA alleles on average, but it is not possible to distinguish origin of individual humans this way. Please read some serious stuff if you want to know exactly why.

An easy to understand parallel: suppose that Africans have almost 100% black curly hair, and Europeans have 30% blond, 55% black straight, 5% black curly and 5% ginger (the numbers are examples). It is not possible to say about an individual human, that your hair are say '50% African and 50% European'.

There are some companies offering such tests but they are pulling people's leg.

 

[Giant Panda]

Sorry but this is pseudo-scientific nonsense. Human populations differ in certain DNA alleles on average, but it is not possible to distinguish origin of individual humans this way. Please read some serious stuff if you want to know exactly why.

An easy to understand parallel: suppose that Africans have almost 100% black curly hair, and Europeans have 30% blond, 55% black straight, 5% black curly and 5% ginger (the numbers are examples). It is not possible to say about an individual human, that your hair are say '50% African and 50% European'.

There are some companies offering such tests but they are pulling people's leg.

Sorry Jurek, but this is false. The genome is far larger than a single imaginary locus for hair colour. As the number of loci sampled increases, we start to generate a clear picture of where even the most ethnically diverse individual's ancestor's originated.

This links to the discussion above. Whilst there has been extensive admixture between different races and regions, genetic substructuring hasn't been eliminated (which you can see in CGSwans' paper). I think language serves as a good analogy here; the European languages have invaded, stolen, and overlapped with one another for millennia, but they're still recognisably distinct. A linguist could make historical inferences about the influx of French words into English post-1066, for instance, but they wouldn't confuse the two. A geneticist can do essentially the same thing (and the geneticist only has four letters).

 

[Jurek7]

Sorry Giant Panda, but you are mistaken. The original idea wrongly moves from population average to individual, and from probability to certainty, and fakely throws up strict numbers. The paper you quote confirms it (bold is mine):

'patterns of IBD provide ample but noisy geographic and temporal signals, which can then be connected to historical events. Rigorously making such connections is difficult'

 

[SealPup]

Does Dienekes still do Dodecad? It was supposedly the best of them. The commercial tests came in for flak for unreliability and privacy issues. But Dodecad was with sincere intent.

 

[Giant Panda]

@Jurek7 I don't believe so (and I think we can dispense with the apologies ). Your example was overly simplistic because it only took into account a single allele, which would be a generally poor way to measure what you claimed can't be measured (as you pointed out). As I wrote, however, sequencing many alleles would start to give a much clearer picture. This is also essentially what the passage you're quoting (slightly out of context) says, so I fail to see how you think I'm mistaken. Or possibly you intended to rebut my language analogy, in which case I'll clarify that was an analogy rather than a strict comparison.

Regardless, the probabilities involved can be so high as to be all but certain for practical purposes, particularly when many loci are sampled (say, 90% likelihood, on top of another 90%, on top of another, and another, etc.). Sometimes we're much less certain, but dealing with population averages is still a far cry from "pseudo-scientific nonsense". And, often strict numbers are not only appropriate but necessary (eg. 50% from each parent). I realise the latter wasn't the case in the post you originally quoted, however.

 

[gentle lemur]

I'm not sure if this rebuts either argument or just adds scientific clarity, but no individual's DNA actually persists that long. Each parent contributes ~50% of their genome to a child, which means each grandparent contributes ~25%, and so on. By the time you're about 8 generations back, each individual ancestor is contributing a miniscule, insignificant amount of their genome to that eventual descendant. Therefore, from a genetic standpoint, nobody alive today is actually "related" to Charlemagne more than they are to any other random person.

There are two exceptions to this rule which I find particularly interesting, if only because I have a very limited understanding of statistical methods. The first is mitochondrial DNA, which we inherit only from our mothers, and the second is the Y chromosome which males inherit only from their fathers. If your mitochrondrial DNA were exactly the same as Charlemagne's, it would be very probable that you shared a common female ancestor at some indeterminate time in the past. Likewise if you are a man with an identical Y chromosome to Charlemagne's you would most probably have a common male ancestor. I don't think anyone has sampled Charlemagne's DNA, but the same argument applies to Julius Caesar, Harald Hardrada, Suleiman the Magnificent, Francis Crick or any other possible relative.
As I understand it, both these DNA strands are short, so actually only a limited number of different sequences are known (in other words, relatively few mutations have taken place during human history) - but the frequencies of these variations do differ between populations, providing some evidence for migration, interbreeding etc. I will not use the 'r-word' in this context, it carries too many inappropriate shades of meaning.