Joined: 24 May 2007 {Posts: 129 } Location: Danville, VA
Posted: Wed 14 Oct 2009 19:17 Post subject: 23andme results and comments
Several months back the question was raised about various autosomal tests. I just got results from the fifth such test I've taken, but as with the previous ones they raise as many or more questions than they answer.
The Ancestry Painting result from 23andme.com was 99% European, 1% African, based on a half-million snps. The only percentage-based results I had gotten previously were from DNAPrint; the first version of the test gave an MLE of 100% European based on 75 markers but the second yielded 94% European, 6% East Asian based on 173. Given the vastly greater number of markers involved with 23andme, my inclination is to assume that this is far more reliable than the DNAPrint results, and settles the question. It's quite impressive to see the color-coded scheme showing precisely where the European and African markers are located. When I started out years ago with lots of oral history about Indian blood in the family but finding only "free mulatto" ancestors documented as nonwhite, this was just the result I expected.
But since then, after the 6% East Asian estimate from DNAPrint's AncestrybyDNA, I took their EuroDNA test which reported Most Likely Estimates of 57% Northern European, 35% Mediterranean, 8% South Asian, which made me think that perhaps my "mulatto" ancestors were really Gypsy or East Indian. That suspicion got stronger when I took the DNATribes test, which with only 25 markers yielded matches that underscored the likelihood of Mediterranean and South Asian ancestry and again came up empty on Native American or subSaharan African both of which were much much lower. Four of the top seven Native Population matches were in Italy, with two more in North Africa and one in Romania. When broken down into 36 World Regions matches, over half my match scores were below 1 including every Native American option, every subsaharan African, and most East Asian categories. Highest scores were for:
Arabian 16.59
Northern European 12.52
Mediterranean 12.49
Aegean 10.04
North African 7.96
Levantine 7.28
Mesopatamian 7.09
Eastern European 6.44
Finno-Ugrian 6.17
Mestizo 4.09
North India 2.3
Altaian 1.27
So at this point I was ready to abandon hope of finding any DNA evidence of either Native American or African ancestry. Now from 23andme I get two kinds of results, one of which tends to support the above pattern but the other of which seemingly contradicts it. Ancestry Painting shows no Asian mixture whatsoever (which would also include Native American), and 1% African. Yet the other result from the same 23andme test, Global Similarity, echoes DNATribes fairly closely, with these rankings:
Northern European 67.72
Southern European 67.6
Near Easterners 67.03
Central Asians 66.8
Northern Africans 66.31
North Americans 66.07
South Americans 65.99
Siberians 65.99
Eastern Asians 65.73
Oceanians 65.67
Eastern Africans 63.63
Southern Africans 63.54
Western Africans 63.42
While the raw numbers don't seem to show a wide range, the bar chart shows the 67+ lines to be much longer than the ones around 66, and the ones under 64 to be extremely short. Without background on the range of possible scores (which I've yet to find on their site) this is confusing.
Could it be that these matches say nothing about me as an individual, and run very much the same for anyone with overwhelmingly northern European ancestry, because they are simply measures of distance among these groups? A friend whose roots are in the same county but who had no known mulatto ancestors came up with much the same results.
The only clue I can think of to reconcile these apples and oranges, the percentages vs. the matches, is the high North African results in both Tribes and 23andme. Seems like I might have neither Native American nor subsaharan African ancestry, but North African as part of a Mediterranean blend typical of Spain and Portugal. Which would point to 16th century Iberian arrivals rather than 17th century slaves direct from Africa?
Every few hundred dollars I shell out for DNA testing gets me more confused except for the Y and mitochondrial results which are crystal clear. But with 23andme, the medical and traits information alone is completely worth the price of the test, even though the ancestral aspects of the results are confusing.
Joined: 07 Oct 2007 {Posts: 248 } Location: United States
Posted: Wed 27 Jan 2010 18:37 Post subject:
Interesting. So 23andme is possibly more accurate because more SNPs are used? I've noticed that nearly all of the 23andme tests have given an admixture result that was more European than the other tests. Now I ask, is there anyway that such a test could be slightly biased because of differences in sample sizes? For example, that more Euro populations and samples were used vs. Asian or African? Or is it simply that the large number of SNPs actually does show that these people are in fact very European genetically?
Edit: On another website that I frequent, one member of Eritrean (East African horner) descent has both more Euro and Asian than African admixture on the 23andme test.
Now, I acknowledge that many studies show East Africans to be in between SSA and Europeans, and probably for more than one reason. It could have to do with Eurasians being a subset of ancient East Africans. It could also be that there was recent and ancient back-migration (especially given proximity). Both likely play a role.
However, even considering that he has an italian great grandparent, it still baffles me, that he is more Asian than African according to this test: Europe - 92%; Asia - 7%; Africa - 1%. Any explanation?
Joined: 07 Feb 2007 {Posts: 1829 } Location: Lookin DC Metro, Feelin Geneva
Posted: Fri 29 Jan 2010 18:52 Post subject:
I just ordered the 23&me test, I realize it will be awhile before i get my results but I will post them. I did this before but the technology was not as good as now, that was about 5 years ago.
Interesting. So 23andme is possibly more accurate because more SNPs are used? I've noticed that nearly all of the 23andme tests have given an admixture result that was more European than the other tests. Now I ask, is there anyway that such a test could be slightly biased because of differences in sample sizes? For example, that more Euro populations and samples were used vs. Asian or African? Or is it simply that the large number of SNPs actually does show that these people are in fact very European genetically?
Edit: On another website that I frequent, one member of Eritrean (East African horner) descent has both more Euro and Asian than African admixture on the 23andme test.
Now, I acknowledge that many studies show East Africans to be in between SSA and Europeans, and probably for more than one reason. It could have to do with Eurasians being a subset of ancient East Africans. It could also be that there was recent and ancient back-migration (especially given proximity). Both likely play a role.
However, even considering that he has an italian great grandparent, it still baffles me, that he is more Asian than African according to this test: Europe - 92%; Asia - 7%; Africa - 1%. Any explanation?
I would definitely not be interested in a DNA test that would show more European ancestry(3% is plenty lol), but I have thought about getting a breakdown of the 96% SSA but forgot which test is best for that. I would love to know if the origins are parallel to those populations that occupied the Geechee/Gullah Islands.
Joined: 24 May 2007 {Posts: 129 } Location: Danville, VA
Posted: Sun 07 Feb 2010 12:39 Post subject: decodeme reinterprets 23andme data
Grasshoppa wrote:
Interesting. So 23andme is possibly more accurate because more SNPs are used? I've noticed that nearly all of the 23andme tests have given an admixture result that was more European than the other tests. Now I ask, is there anyway that such a test could be slightly biased because of differences in sample sizes?
Another company, decodeme.com, offered to interpret the data from 23andme for free. Population matches were along the same lines as earlier tests. But autosomal percentages changed to 93% European, 4%Asian and 3% African. So it was the formula whereby scores are calculated, and not the increased DNA data, that made 23andme so different from DNAPrint. Using the same expanded data, decodeme comes up with results closer to the old DNAPrint score of 94 Euro/6 East Asian.
They market themselves largely on the basis of their "African Lineage Database" which supposedly has large samplings of current populations in Africa for DNA comparison.
There is also another such company called Roots for Real that was showcased in a BBC documentary called Motherland (preview HERE). They also offer autosomal testing.
In that documentary a group of British "black Caribbeans" were tested and then supposedly reunited with their modern day African tribes of origin. Of course, there was an enormous degree of cherry picking involved, which led to great criticism of the documentary.
Posted: Tue 23 Feb 2010 00:24 Post subject: Re: 23andme results and comments
kpauljohnson wrote:
Several months back the question was raised about various autosomal tests. I just got results from the fifth such test I've taken, but as with the previous ones they raise as many or more questions than they answer.
The Ancestry Painting result from 23andme.com was 99% European, 1% African, based on a half-million snps. The only percentage-based results I had gotten previously were from DNAPrint; the first version of the test gave an MLE of 100% European based on 75 markers but the second yielded 94% European, 6% East Asian based on 173. Given the vastly greater number of markers involved with 23andme, my inclination is to assume that this is far more reliable than the DNAPrint results, and settles the question. It's quite impressive to see the color-coded scheme showing precisely where the European and African markers are located. When I started out years ago with lots of oral history about Indian blood in the family but finding only "free mulatto" ancestors documented as nonwhite, this was just the result I expected.
But since then, after the 6% East Asian estimate from DNAPrint's AncestrybyDNA, I took their EuroDNA test which reported Most Likely Estimates of 57% Northern European, 35% Mediterranean, 8% South Asian, which made me think that perhaps my "mulatto" ancestors were really Gypsy or East Indian. That suspicion got stronger when I took the DNATribes test, which with only 25 markers yielded matches that underscored the likelihood of Mediterranean and South Asian ancestry and again came up empty on Native American or subSaharan African both of which were much much lower. Four of the top seven Native Population matches were in Italy, with two more in North Africa and one in Romania. When broken down into 36 World Regions matches, over half my match scores were below 1 including every Native American option, every subsaharan African, and most East Asian categories. Highest scores were for:
Arabian 16.59
Northern European 12.52
Mediterranean 12.49
Aegean 10.04
North African 7.96
Levantine 7.28
Mesopatamian 7.09
Eastern European 6.44
Finno-Ugrian 6.17
Mestizo 4.09
North India 2.3
Altaian 1.27
So at this point I was ready to abandon hope of finding any DNA evidence of either Native American or African ancestry. Now from 23andme I get two kinds of results, one of which tends to support the above pattern but the other of which seemingly contradicts it. Ancestry Painting shows no Asian mixture whatsoever (which would also include Native American), and 1% African. Yet the other result from the same 23andme test, Global Similarity, echoes DNATribes fairly closely, with these rankings:
Northern European 67.72
Southern European 67.6
Near Easterners 67.03
Central Asians 66.8
Northern Africans 66.31
North Americans 66.07
South Americans 65.99
Siberians 65.99
Eastern Asians 65.73
Oceanians 65.67
Eastern Africans 63.63
Southern Africans 63.54
Western Africans 63.42
While the raw numbers don't seem to show a wide range, the bar chart shows the 67+ lines to be much longer than the ones around 66, and the ones under 64 to be extremely short. Without background on the range of possible scores (which I've yet to find on their site) this is confusing.
Could it be that these matches say nothing about me as an individual, and run very much the same for anyone with overwhelmingly northern European ancestry, because they are simply measures of distance among these groups? A friend whose roots are in the same county but who had no known mulatto ancestors came up with much the same results.
The only clue I can think of to reconcile these apples and oranges, the percentages vs. the matches, is the high North African results in both Tribes and 23andme. Seems like I might have neither Native American nor subsaharan African ancestry, but North African as part of a Mediterranean blend typical of Spain and Portugal. Which would point to 16th century Iberian arrivals rather than 17th century slaves direct from Africa?
Every few hundred dollars I shell out for DNA testing gets me more confused except for the Y and mitochondrial results which are crystal clear. But with 23andme, the medical and traits information alone is completely worth the price of the test, even though the ancestral aspects of the results are confusing.
At this point I would rely on genealogical research versus test results. Our autosomal results only reflect what was passed to us. If you have a sibling they could show up with higher or lower African results or none! Has anyone else in your family taken these test? If your parents are still living I would test whichever one has the mulatto ancestor and compare your genomes on 23andme.
Posted: Tue 23 Feb 2010 05:42 Post subject: Re: decodeme reinterprets 23andme data
kpauljohnson wrote:
Grasshoppa wrote:
Interesting. So 23andme is possibly more accurate because more SNPs are used? I've noticed that nearly all of the 23andme tests have given an admixture result that was more European than the other tests. Now I ask, is there anyway that such a test could be slightly biased because of differences in sample sizes?
Another company, decodeme.com, offered to interpret the data from 23andme for free. Population matches were along the same lines as earlier tests. But autosomal percentages changed to 93% European, 4%Asian and 3% African. So it was the formula whereby scores are calculated, and not the increased DNA data, that made 23andme so different from DNAPrint. Using the same expanded data, decodeme comes up with results closer to the old DNAPrint score of 94 Euro/6 East Asian.
That's interesting! It's so tempting to take several test, but I fear I will run into a similar predictment. You are absolutely correct! The more you dig, the questions are raised that need to be answered. That's what makes genealogy so addictive.
I have my results from 23andme, but I am still waiting on dnatribes. Have you used the relative finder service btw?
Joined: 24 May 2007 {Posts: 129 } Location: Danville, VA
Posted: Sun 28 Feb 2010 18:52 Post subject: Re: decodeme reinterprets 23andme data
That's interesting! It's so tempting to take several test, but I fear I will run into a similar predictment. You are absolutely correct! The more you dig, the questions are raised that need to be answered. That's what makes genealogy so addictive.
I have my results from 23andme, but I am still waiting on dnatribes. Have you used the relative finder service btw?[/quote]
No one I know who takes several tests with autosomal percentages has found much consistency. Y and mitochondrial results show total consistency from company to company, and the population matches results seem roughly parallel. But the Euro/Asian/African percentages seem to vary wildly. If you take many tests your knowledge will expand but so will your confusion!
A friend from the same NC county as my dad has gotten great results with Relative Finder, but so far no such luck for me.
Joined: 02 May 2006 {Posts: 443 } Location: Īle-de-France
Posted: Wed 03 Mar 2010 15:01 Post subject:
Grasshoppa wrote:
Edit: On another website that I frequent, one member of Eritrean (East African horner) descent has both more Euro and Asian than African admixture on the 23andme test.
Now, I acknowledge that many studies show East Africans to be in between SSA and Europeans, and probably for more than one reason. It could have to do with Eurasians being a subset of ancient East Africans. It could also be that there was recent and ancient back-migration (especially given proximity). Both likely play a role.
However, even considering that he has an italian great grandparent, it still baffles me, that he is more Asian than African according to this test: Europe - 92%; Asia - 7%; Africa - 1%. Any explanation?
Here's a guess. Dragon Horse just posted this in another thread from 28&me's website
Quote:
23andMe takes advantage of publicly available data for four populations studied extensively via the International HapMap project (hapmap.org). That project obtained the genotypes for 60 individuals of western European descent from Utah, 60 western African individuals from Nigeria, and 90 eastern Asian individuals, 45 from each of Japan and China. Because the two eastern Asian populations are geographically near one another and relatively similar at the genetic level, 23andMe combines these to form a single eastern Asian reference population.
Meaning their whole comparison for Africa is 60 individuals from Nigeria. As the source population for all humanity, genetic diversity is higher in Africa than anywhere else in the world, and I recall reading recently that on average, there is more genetic distance between two individuals from Khoi-San tribes that speak different languages than there is between a western european and an east asian. If all they are using for a reference for Africa is 60 people from Nigeria, they are probably missing a lot of potentially ancestry informative markers common elsewhere in Subsaharan Africa.
Joined: 07 Feb 2007 {Posts: 1829 } Location: Lookin DC Metro, Feelin Geneva
Posted: Wed 03 Mar 2010 16:34 Post subject: Re: decodeme reinterprets 23andme data
kpauljohnson wrote:
Grasshoppa wrote:
Interesting. So 23andme is possibly more accurate because more SNPs are used? I've noticed that nearly all of the 23andme tests have given an admixture result that was more European than the other tests. Now I ask, is there anyway that such a test could be slightly biased because of differences in sample sizes?
Another company, decodeme.com, offered to interpret the data from 23andme for free. Population matches were along the same lines as earlier tests. But autosomal percentages changed to 93% European, 4%Asian and 3% African. So it was the formula whereby scores are calculated, and not the increased DNA data, that made 23andme so different from DNAPrint. Using the same expanded data, decodeme comes up with results closer to the old DNAPrint score of 94 Euro/6 East Asian.
Joined: 07 Feb 2007 {Posts: 1829 } Location: Lookin DC Metro, Feelin Geneva
Posted: Wed 03 Mar 2010 17:12 Post subject:
MisterLawyer wrote:
Grasshoppa wrote:
Edit: On another website that I frequent, one member of Eritrean (East African horner) descent has both more Euro and Asian than African admixture on the 23andme test.
Now, I acknowledge that many studies show East Africans to be in between SSA and Europeans, and probably for more than one reason. It could have to do with Eurasians being a subset of ancient East Africans. It could also be that there was recent and ancient back-migration (especially given proximity). Both likely play a role.
However, even considering that he has an italian great grandparent, it still baffles me, that he is more Asian than African according to this test: Europe - 92%; Asia - 7%; Africa - 1%. Any explanation?
Here's a guess. Dragon Horse just posted this in another thread from 28&me's website
Quote:
23andMe takes advantage of publicly available data for four populations studied extensively via the International HapMap project (hapmap.org). That project obtained the genotypes for 60 individuals of western European descent from Utah, 60 western African individuals from Nigeria, and 90 eastern Asian individuals, 45 from each of Japan and China. Because the two eastern Asian populations are geographically near one another and relatively similar at the genetic level, 23andMe combines these to form a single eastern Asian reference population.
Meaning their whole comparison for Africa is 60 individuals from Nigeria. As the source population for all humanity, genetic diversity is higher in Africa than anywhere else in the world, and I recall reading recently that on average, there is more genetic distance between two individuals from Khoi-San tribes that speak different languages than there is between a western european and an east asian. If all they are using for a reference for Africa is 60 people from Nigeria, they are probably missing a lot of potentially ancestry informative markers common elsewhere in Subsaharan Africa.
To add, I found this information:
Quote:
Part I of difference between 23andMe and Decode
To understand why 23andMe and Decode come up with different overall ancestry proportions, we need a brief digression into human evolutionary genetics. All human beings around the world are pretty darned similar at the genetic level, and a given stretch of chromosome from any of us is quite likely to be found in all world
populations. However, the frequency with which this stretch is
observed will differ across populations. It's these differences in
frequency that underlie both companies' methods. To illustrate, let's suppose that we're looking at a stretch of chromosome from a person whose ancestry we don't (but wish to) know. When you consult the reference populations, if that stretch were found at 85% in Europeans, at 3% in Asians, and at 31% in Africans, your "educated guess" would be that the stretch came from Europe, although it is possible that it came from Africa, and still less likely, Asia.
snip
How does what was discussed in Part I matter to the overall ancestry proportions reported? Let's look at an average Northern European's genome. Both companies will walk along each of this person's chromosomes, and will tend to find that each stretch is found in all three reference populations, but is most likely from Europe. This would be expressed by Decode as a high chance, maybe 80-90%, that the stretch comes from Europe, and a smaller chance, maybe more like 5-10% apiece, that the stretch comes from Africa or Asia. Adding up all the stretches, you'll tend to get 80-90% European ancestry, and 5-10% African and Asian ancestry, each, for the Northern European -- this is consistent with the Decode chromosomal ancestry analyses of Northern Europeans that I've seen. This is a reasonable way to show the data.
report abuse
snip
he reasoning behind 23and Me's Ancestry Painting is, while it's true that each stretch is found all over the world, we know or are willing to assume that the stretch can only have come from /one/ population, and it chooses the most likely population. For our example Northern European, for just about every stretch along their genome, the single most likely origin will almost always be European, and would be expressed in final ancestry proportion estimates of about 100% European, and about 0% African and Asian.
This is how Ancestry Painting and Decode can end up with different total ancestry estimates for the same person
Posted: Thu 04 Mar 2010 00:39 Post subject: Re: decodeme reinterprets 23andme data
kpauljohnson wrote:
That's interesting! It's so tempting to take several test, but I fear I will run into a similar predictment. You are absolutely correct! The more you dig, the questions are raised that need to be answered. That's what makes genealogy so addictive.
How accurate are these tests?
I know someone who had it done and he told me the results were not what he thought.
He was being funny. The test showed 100 European.
He thought it would have showed 75 & European, something like that and some Nat. American, some SSA. He is Creole Black American.
He looks White American. Lives as White American.
His kids look and go as WA. His wife is from Filipino and her test showed Chinese, European.
Posted: Mon 08 Mar 2010 14:48 Post subject: Re: decodeme reinterprets 23andme data
kpauljohnson wrote:
That's interesting! It's so tempting to take several test, but I fear I will run into a similar predictment. You are absolutely correct! The more you dig, the questions are raised that need to be answered. That's what makes genealogy so addictive.
I have my results from 23andme, but I am still waiting on dnatribes. Have you used the relative finder service btw?
No one I know who takes several tests with autosomal percentages has found much consistency. Y and mitochondrial results show total consistency from company to company, and the population matches results seem roughly parallel. But the Euro/Asian/African percentages seem to vary wildly. If you take many tests your knowledge will expand but so will your confusion!
A friend from the same NC county as my dad has gotten great results with Relative Finder, but so far no such luck for me.[/quote]
Hello kpauljohnson,
Sorry I am responded so late. Just finished exams so I haven't had much time for recreational activities. Relative finder started off slow for me, but since they were featured on Prof Gates PBS series 'Faces of America', I noticed an increase in responses from RF cousins. So far 6 have accepted my invitation within the last 3 days. Not bad considering my request were practically being ignored for the past month.
I still have not received my dnatribes results. Apparently their results are confusing for people that have mixed heritage. But at least they provide raw date, which I have been told you can compare with other published studies.
I have heard 23andme is pretty reliable with their admix results, compared to everything else on offer out there.
There global similarity tool, however, makes no sense to me. My admix results are 66% AFrican, 23% European, and 11% Asian, which I assumed was Amerindian or at least part of it. Yet I have low matched in North and South America compared to people with 100% European ancestry that show higher matches.
If you would like share genomes my username is ladyla
DragonHorse, I'm so glad you cleared that up for me. Interesting. I think that shows that these results shouldn't be taken as absolute truth.
I agree! If your entire genome is tested then you can be certain the results are spot on accurate. 23andme test the largest amount of genomes I have seen so I feel confident their predictions are close.
But I still believe dna test should be used in conjunction with genealogial research. That's when it is the most helpful.
'
FAQ
Search
Register
Log in
Relative finder started off slow for me, but since they were featured on Prof Gates PBS series 'Faces of America', I noticed an increase in responses from RF cousins. So far 6 have accepted my invitation within the last 3 days. Not bad considering my request were practically being ignored for the past month.