deCODE Admin
|
posted Jan 21, 2010 4:20pm
|
We have received several posts from deCODEme users with questions about our "Ancestral origins" analysis. Here are some details that may answer at least some of the questions raised.
First, however, it is important to point out that there is no algorithm or test that can provide an absolutely certain estimate of ancestry for an individual based on genetic data - much less for a particular chromosome fragment. Absolute certainty in this context is impossible and anyone who claims they can provide such certainty is almost certainly wrong.
We use a Markov-Chain Monte-Carlo (MCMC) algorithm to estimate the likely ancestry of deCODEme users. This algorithm uses samples from three reference populations to define ancestral gene pools - and then estimates the proportion of the genome inherited from these postulated ancestral gene pools.
The algorithm attempts to find the best fitting solution given a particular model by large sample of possible solutions. All possible solutions cannot be evaluated, because there are so many of them. Each time the analysis is performed for a particular individual's set of genotypes, a slightly different combination of possible solutions is evaluated. This is the reason why deCODEme users will see slightly different results when analyses are updated at the deCODEme end or when the same genotypes are uploaded more than once. The difference should be very small for the autosomal genome, but may be larger for the X-chromosome (because it has fewer SNPs than the 22 autosomal chromosomes and there is more uncertainty) and may vary substantially for a handful of chromosomal fragments that do not show major differences between the reference populations used in the analysis.
The reference population samples were obtained from the HapMap project - they are:
1) European Americans from Utah - who most likely have a majority of north European ancestry
2) Yoruban Nigerians
3) Chinese from Beijing and Japanese from Tokyo.
The characteristics of these reference population samples and the clinal nature of human genetic variation (i.e. the fact that people typically become gradually more different as you travel further from your country) have several minor implications for the interpretation of the results. For example, a deCODEme user with a majority of ancestors (during the past >2 generations) from south-east Europe, will typically see higher percentages of African and Asian ancestry than a deCODEme user whose ancestry is mainly from north-west Europe. The difference will be small, but present.
At present, the Ancestral Origins analysis is likely to be most informative for deCODEme users who have recent mixed ancestry (for example African Americans, Hispanic Americans, inhabitants of Central and South America, and others) and and for individuals from populations with an ancient history of mixed ancestry, such as India, Pakistan, populations of Central Asia, Northern Africa and others. |
|
|
|
|
posted Jan 21, 2010 5:31pm
|
Thanks for this very clear explanation Keith. It makes perfect sense.
What does provide some problems in trying to interpret the data is the display mode. There are 22 pairs of autosomes. Females have two X chromosomes and males one. How are we to understand the meaning of the wavy yellow and brown lines considering the fact that only one of each autosome is shown? If someone is European admixed with Native American from one parent is it expected that the squiggle representing an "Asian" block should top out at about half way across the chromosome and visually look like a truncated triangle or an undulating block. Will the percentage be added with whatever is found on the chromosome from the other parent at the same location (say position 35 to 45 Mb from the ptel - telomere, small end of chromosome)? If for example mother was Asian and father African what might we expect to see displayed?
In relation to the X, if there is an Asian block on a primarily European background would the single displayed chromosome show a brown or yellow wave extending across the entire X for a male, and the same sized block shown as extending half way across the chromosome for a female?
If there was a line drawn down the middle of the chromosome, or if there were 44 autosomes it would be far easier to understand what we are looking at - at least from my vantage point.
Thanks again,
David. |
|
|
|
|
posted Jan 21, 2010 7:54pm
|
Hi,
Would it be possible to have some explanation of the common ancestor feature too? Alot of people, myself included are have difficulty understanding what each score (5MB etc) means in relation to finding a distant relative. I know it's hard to say exactly when someone shared an ancestor, but is it possible to have a rough idea together with an idea of how much of a score someone needs (eg is 0.9% significant?)
Many thanks. |
|
|
|
|
posted Jan 24, 2010 7:02am
|
I have used the tool on most of my friends here, and some reference individuals especially the mostly Caucasoid one like the Pathans, Brahui, Sindhi and so on.
What I find is I share at 5Mb with most Europeans and other Caucasoids. The amount varies from one block to four blocks and the sharing is not really tied with ethnicity. I share 1.1% @ 5Mb with a Norwegian who appears near Kari Stefansson on the Europe PCA, Chromosomes 3, 4, 7 and 11, 5 blocks of dna. With my RF cousin from 23andMe, we actually share 6th great grand parents, I share 1.4% @ 5Mb, Chromosomes 3, 6, 8, 11, 12 and 20. I am not related to my Norwegian friend. What is common about the two is Chromosome 3 and 11. I seem to have one or both those chromosome blocks in common with most Europeans, and and usually one with Middle Easterners and SW Asians. I would assume Chromosome 3 and 11 reflect ancient common ancestry probably Paleolithic in age. The other chromosome blocks must reflect more recent genetic contact. How recent must be worked out empirically for each friend or "cousin". 23andMe said my RF cousin was my 4th cousin. That was about right. In years it went back to the 1700s.
I don't consider 0.9% @ 5Mb too significant. I have that level of sharing with the Sardinian reference individual. I have no known Sardinian ancestry, in fact I am wholly Maltese back to the 1500s, actually older as I can go further back in the records but I do not know where my ancestors came from 1000 or more years back. The oldest surname list of surnames used today in Malta is 1419. Mine is on the list. Before that people used place names as surnames, an Italian example, Leonardo da Vinci. Surnames changed with every generation. It is guesswork figuring out relatedness. |
|
|
|
|
posted Jan 24, 2010 4:47pm
|
I am from the North-Eastern corner of Anatolia, so my sharing results are pretty much on the lower side of the spectrum, no surprises here. I have only 5 predicted cousins at 23andme, 2 of which have very close geographic origins to me, one being a native of Caucasus.
I have made a similar experiment inspired by Ponto,
and observed a similar pattern with Chr 3 and 11 @5Mb. Those bits are consistently shared with my European contacts, if shared at all¿plus one Turk, Armenian and the token Pathan.
Tiny bits on Chromosome 5 are shared with Italian, Cauc. Brazilian, Bedouin, Assyrian, Turkish, Russian and Basque.
Although their %s are somewhat insignificant, my top 3 %s at 5Mb are with
a Frenchman (0,7)
Brazilian and Ashkenazi (0.6)
Maltese (0.5)
all other matches are invariably 0.4 or 0.2 %.
And an interesting asymmetry I have observed:
I am sharing zero segments @5Mb with people whom I share moderately high @250Kb (ie 2 Turks, Adygei, Iranian and Middle Eastern) whereas I share 0.4% with some people whom I have a low overall (250Kb) similarity. |
|
|
|
|
posted Jan 24, 2010 6:45pm
|
Hi Bonal,
How do you know that 0.7% is insignificant? I ask because I'm trying to establish what is significant or not..my highest scores at 5Mb are 0.9% 0.8% 0.8% I would love to know if this actually means I have a shared ancestor with these people. |
|
|
|
|
posted Jan 25, 2010 5:49am
|
300 km in ancient times is a long way. My connection with Sardinians, really their Paleolithic ancestors, is very ancient, far older than some people from the Eastern Mediterranean whose presence was felt in Europe only 3,000 years ago and who were exterminated as an ethnic group and language in the historic era.
Personally I think the Phoenicians are overrated and their effect in Europe, over exaggerated. In North Africa, Iberia, Cyprus, western Sicily, Sardinia and Cyprus the Phoenicians did colonize and left some legacy. Malta they did little but use the island as a convenient toilet stop, provide the locals with some goods they not normally get hold of but they did not colonized the place. Malta had nothing for those traders to exploit. Sardinia, Cyprus, Sicily are large islands with civilizations with which to trade with and exploit. Malta was in the Magna Graecia sphere of influence and the closest landfall in Sicily in Greek territory. It was the Greeks that called Malta, Melita because of the sweetness of the honey produced by the Maltese variety of honey bee. Malta was colonized by Siculians from Sicily. Archaeologically in Malta, there is practically nothing left by the Phoenicians but some simple tombs, some votive offerings and grave goods which were mainly Greek in manufacture. No sarcophagi as found in Spain. The Romans hated the Punic kin of the Phoenicians so much they utterly destroyed Carthage, spread salt all over its soils so that it would never be habitable again. What did the Romans do to the Maltese. Nothing except incorporate it into the Roman empire. There is quite a bit of Roman archaeology in Malta.
Sorry, if you are looking for Phoenicians, Spain and western Sicily, especially Motya Island, is the place to go. |
|
|
|
|
posted Jan 25, 2010 6:22am
|
These are some of my 5Mb results for friends and referenced individuals.
RF cousin from 23andMe 1.4% We are related, goes back to the 1700s.
Norwegian 1 1.1%
Sardinian Ref 0.9%
USA North Euro1 0.7%
Balochi Ref 0.7%
Druze Ref 0.7%
USA North Euro2 0.6%
Spaniard 0.6%
Basque Ref 0.5%
Puerto Rican 0.5%
Lithuanian Jew 0.5%
Italian 0.4%
USA Calabrian Italian 0.4%
Palestinian Ref 0.4%
Croatian 0.4%
Assyrian 0.2%
Pole 0.2%
That is just a precis of those I am friends with and some reference individuals. I don't think it says much about my relatedness except it was a long time ago particularly for the Pole and the Assyrian. |
|
|
|
|
posted Jan 25, 2010 10:59am
|
Glenbrenbur said:Hi Bonal,
How do you know that 0.7% is insignificant? I ask because I'm trying to establish what is significant or not..my highest scores at 5Mb are 0.9% 0.8% 0.8% I would love to know if this actually means I have a shared ancestor with these people.
I wish I knew, I just referred to Ponto's assumption.
0.9% is hard to ignore, I guess. |
|
|
|
|
posted Jan 26, 2010 1:51am
|
I have just posted a few of my 5Mb results.
I have a 1.1% sharing of dna with a Norwegian friend, and 0.9% with the Sardinian reference Individual. Going by the logic of pasulem. Norway is not that far from Malta, and I am sure the Phoenicians or Aztecs or Atlanteans sailed to the Northlands often for their herrings and whale meat, and while there took up with a few Norwegian lasses. Those dark Caucasoids and non Caucasoids just love fair skin, blond hair and blue eyes.
That site posted by pasulem is a lot of horse pooky. Utter anti Arabian racism and the lack of acceptance by some Lebanese that their genetic ties to Arabians is fact and permeate every sector of Lebanese society Christian, Muslim, and Druze.
What applies to Glenbrenbur does not apply to me. For one thing, North Europeans come from a smaller founder population, which makes their genetic affiliations much closer to otherwise complete and utter strangers who happen to be North Europeans. Southern Europeans come from a larger and more diverse genetic population. It is very unlikely Southern European would match as highly to each other than Northern Europeans do as a rule. Southern Europeans have higher sharing with Asians and Africans than Northern Europeans do, even than Northern Europeans who have slight admixture as shown on 23andMe. Northern Europeans highest sharing outside their group is with Southern Europeans, and often at a higher rate than Southern Europeans share with each other.
For me my sharing of 1.1% with a Norwegian friend is of no genealogical consequence, just random chance and my particular combination of dna. Nothing more can be read it in.
P.S I also share 0.9% with another Norwegian friend. Gee those Phoenicians got around! |
