PDA

View Full Version : New 23andMe Ancestry Timeline is a fraud



Maciamo
25-03-17, 10:25
I was finally able to check the new Ancestry Timeline on 23andMe after my account was migrated to the new 23andMe experience.

The verdict will be brief. It is a fraud. I have a complete paper genealogy for the last 7 generations in every line, and many lines go back over 400 years. All my ancestors were always in the same region. Within the last 8 generations they were all in a 30 km radius and could not possibly have come from anywhere else.

Yet, according to 23andMe I or my close relatives have British/Irish ancestors 1 or 2 generations ago, Italian ones from 4 or 5 generations ago, Balkans, Sardinian and Iberian ones from 5 generations ago, and Finnish/Scandinavian ancestors from 6 generations ago.

One report even claims that a relative is British & Irish in the last 3 generations, but only French & German from the 2nd to 4th generation! Belgium is included in French & German, so that is nonsensical.

23andMe claims on its website that the Ancestry Composition looks at ancestors in the last 500 years. However this is a terrible estimation on their part. I have analysed the Ancestry Composition of over 50 people with all recent ancestors from the Benelux (http://www.eupedia.com/europe/benelux_france_dna_project.shtml#Autosomal_Benelux ) and many others from all over Europe, and it is fairly clear that the Italian, Sardinian, Iberian, Balkans and Scandinavian components are all fairly ancient in Belgium, probably in the range of 1000 to 2000 years before present, and maybe much older for the Sardinian (Neolithic leftover?).

The Scandinavian and Finnish components almost certainly came to the Benelux and the British Isles with the Vikings. If it was older, going back to the Iron Age Germanic expansion out of Scandinavia (Frisians, Saxons, Franks), the percentage would be much higher (over 40%). Since there was no mass migration from Scandinavia to these regions after the Viking Age, nothing else could account for percentages of up to 10% of Scandinavia in some individuals from Britain or Belgium. Besides, the only high Scandinavian percentages in Belgium are found in coastal areas like Bruges that were settled by the Danish Vikings. Many Walloons and East Flemings have 0 to 2%, as is the case for most British people outside the Danelaw.

Likewise, the Italian component can only be explained by ancient Roman ancestry. This is very clear in the Benelux, where the percentages correlate with the density of Roman settlements. In the most Romanised parts of Wallonia, the Italian component reaches up to 4%. In coastal Flanders and the northern half of the Netherlands, it is almost always 0%. Dutch people in the Meuse-Rhine region, that was lined with Roman settlements and forts (Maastricht, Nijmegen and Utrecht were all founded by the Romans) might get up to 2% of Italian, while than other Dutch people typically have 0%.

As for the 'British & Irish' component, it obviously includes a strong Anglo-Saxon part of ancestry in common with Danish and Dutch people. I have seen one Dutch guy from Holland and a Flemish guy from Antwerp with 40% of 'British & Irish'. The average for Danes, Dutch and Flemings is 20-25%. It lower in the eastern and southern parts of the Benelux (15-20%). So 23andMe apparently failed to realise that their "500 year-old British & Irish" component was in fact a blend of ancient Insular Celtic (Gaelic, Brythonic) and ancient Anglo-Saxons.

When you know how to read the Ancestry Composition (what's hidden behind the component names) the percentages can be quite useful to compare populations. But I cannot stand that 23andMe would lie to their customers about how old these components are, and especially make them believe that they have ancestors from such and such region a few generations ago based on ancient admixtures. This is highly misleading, deceitful and even unethical. I have held 23andMe in high regard over the years, but this, combined with the decisions to remove health reports and updates for customers in most of the world (outside North America and parts of northern Europe), even those who paid a high price for the original test 10 years ago, is definitely a shift in policy in the wrong direction.

While we are at it, their Relatives Finder is also a scam, as are most similar services by other companies. Thanks to my detailed paper genealogy and very secluded ancestral region, I could easily verify if any of the dozens of "3rd to 5th cousins" listed in my results were really related in genealogical times. I contacted a few of them and it turns out that none had ancestry in Wallonia, ever. Most didn't even have any known Belgian ancestor. I think that the only reason they launched this tool is to get a share of all the naïve (mostly American) customers from Family Tree DNA and AncestryDNA who really believe that they are connecting with 3rd or 4th cousins based on what the site tells them. I have never understood why people would want to connect with a 3rd or 4th cousin anyway. I barely see my first and second cousins.

Maciamo
25-03-17, 10:43
Roberta Estes shared my opinion in her review two months ago on DNA eXplained: Calling HOGWASH on 23andMe’s Ancestry Timeline (https://dna-explained.com/2017/01/17/calling-hogwash-on-23andmes-ancestry-timeline/).

PaschalisB
25-03-17, 11:19
Very helpful post indeed.
How would you explain the small percentages of northwestern European ancestry in some Greek people? I score French & German while others score Finnish. Most Greeks don't score NW though. And do you think that the Eastern European component is ancient or more recent? I think it's due to the slavic migrations, since it's more common among North Greeks in percentages below 3%, while islanders rarely score it.

Maciamo
25-03-17, 12:35
Very helpful post indeed.
How would you explain the small percentages of northwestern European ancestry in some Greek people? I score French & German while others score Finnish. Most Greeks don't score NW though. And do you think that the Eastern European component is ancient or more recent? I think it's due to the slavic migrations, since it's more common among North Greeks in percentages below 3%, while islanders rarely score it.

If, as you say, the Eastern European component in more common among North Greeks, then it is much more likely to be of Slavic origin. Some of it could have been brought by the Goths a few centuries before the actual Slavic migrations, as the Goths were a mix of Germanic and (Proto-)Slavic tribes.

The French & German component is apparently a blend of Germanic (e.g. Frankish) and Gaulish/La Tène Celtic. In Belgium, Walloons (45%) score substantially higher than Flemings (36%) or of people from French Flanders (38%), which can be explained by the higher Gaulish ancestry, but also because the Germanic part of Walloon ancestry is almost exclusively Frankish, as opposed to Frankish, Saxon and Viking for the Flemings. The highest score I have seen on 23andMe was for one of my family members (70%), although South Germans (50-55%) seem to have a bit more in average than Walloons, while Swiss (45%) are similar to Walloons.

So how did the French & German get to Greece? Two possibilities:

1) Some of it came with La Tène Celts, who migrated along the Danube as far as central Anatolia. However they didn't really settle in Greece, so it would be by genetic drift and progressive mixing with neighbours over the last two millennia. Anatolian Greeks who moved to Greece after the independence could potentially have had more of it if the Galatians of Anatolia did carry that admixture.

2) The Goths. Although they were an East Germanic tribe, they would have carried a bit of similar 'French & German' as the Franks, Alemani and other tribes from Late Ancient/Early Medieval Germany. Even modern Danes have about 5% of 'French & German', while Swedes have about 3%. The Goths could also have had some Scandinavian and a bit of Finnish.

In short, I believe that the Goths are responsible for bringing Eastern European, French & German, Scandinavian and Finnish components to the Balkans.

While we are on the topic, I noticed that out of the 20 Italian Ancestry Compositions I saw on 23andMe, none had any Scandinavian, but most had some French & German. North Italians from Lombardy and Veneto typically score between 5 and 12% of French & German, against 0 to 2% for central and southern Italy. So it looks like the Lombards carried predominantly 'French & German' admixture like the Franks - unless of course it is of Celtic origin.

PaschalisB
25-03-17, 13:26
Thank you for the answer.

Here is exactly what I have noticed in a sample of 34 Greeks with mainland greek ancestry:
The percentage of Balkan ranges from 60 to 95%.
The percentage of Italian is between 3 and 20% and is usually related to Balkan (high Balkan means low Italian, low Balkan means high Italian). Italian higher than 20% usually means islander ancestry (the islanders score higher Italian than Balkan - usually around 50-60% Italian)

Eastern European isn't found in all Greeks, but the ones who score it are mainly Northern Greeks (Thessaly, Epirus, Macedonia and Thrace). The highest percentage I've seen is in a Northwestern Greek (areas with large slavic populations) and it's 4.5%.

Northwestern European is very tricky, since it is found in both Mainlanders and Islanders, but not in all of them. It is usually below 1% and falls into the broad northwestern european cluster, but sporadically it can be French&German, Finnish or even British&Irish. There is no clear pattern in it being higher in any area. On the other hand it could also have come due to the Slavs because from what I've seen almost all Balkan Slavs score it in percentages above 1%.

I score 2.5% Eastern European and 1.6% Northwestern (0.2% French&German). 23andme claims I had an Eastern European ancestor who was born between 1720 and 1840 and a French&German ancestor between 1720 and 1810 (which makes no sense).

srdceleva
25-03-17, 13:29
This is also a problem with the Ancestry.dna test, and I think should be explained better by the companies. YouTube is filled with Europeans, and americans who know their ancestry quite far back from their family tree, but then believe they have a recent Spanish ancestor because they show up as 3 % Iberian even though all their ancestry is from the british Isles or Germany. It's extremely misleading

Angela
25-03-17, 16:38
I was finally able to check the new Ancestry Timeline on 23andMe after my account was migrated to the new 23andMe experience.

The verdict will be brief. It is a fraud. I have a complete paper genealogy for the last 7 generations in every line, and many lines go back over 400 years. All my ancestors were always in the same region. Within the last 8 generations they were all in a 30 km radius and could not possibly have come from anywhere else.

Yet, according to 23andMe I or my close relatives have British/Irish ancestors 1 or 2 generations ago, Italian ones from 4 or 5 generations ago, Balkans, Sardinian and Iberian ones from 5 generations ago, and Finnish/Scandinavian ancestors from 6 generations ago.

One report even claims that a relative is British & Irish in the last 3 generations, but only French & German from the 2nd to 4th generation! Belgium is included in French & German, so that is nonsensical.

23andMe claims on its website that the Ancestry Composition looks at ancestors in the last 500 years. However this is a terrible estimation on their part. I have analysed the Ancestry Composition of over 50 people with all recent ancestors from the Benelux (http://www.eupedia.com/europe/benelux_france_dna_project.shtml#Autosomal_Benelux ) and many others from all over Europe, and it is fairly clear that the Italian, Sardinian, Iberian, Balkans and Scandinavian components are all fairly ancient in Belgium, probably in the range of 1000 to 2000 years before present, and maybe much older for the Sardinian (Neolithic leftover?).

The Scandinavian and Finnish components almost certainly came to the Benelux and the British Isles with the Vikings. If it was older, going back to the Iron Age Germanic expansion out of Scandinavia (Frisians, Saxons, Franks), the percentage would be much higher (over 40%). Since there was no mass migration from Scandinavia to these regions after the Viking Age, nothing else could account for percentages of up to 10% of Scandinavia in some individuals from Britain or Belgium. Besides, the only high Scandinavian percentages in Belgium are found in coastal areas like Bruges that were settled by the Danish Vikings. Many Walloons and East Flemings have 0 to 2%, as is the case for most British people outside the Danelaw.

Likewise, the Italian component can only be explained by ancient Roman ancestry. This is very clear in the Benelux, where the percentages correlate with the density of Roman settlements. In the most Romanised parts of Wallonia, the Italian component reaches up to 4%. In coastal Flanders and the northern half of the Netherlands, it is almost always 0%. Dutch people in the Meuse-Rhine region, that was lined with Roman settlements and forts (Maastricht, Nijmegen and Utrecht were all founded by the Romans) might get up to 2% of Italian, while than other Dutch people typically have 0%.

As for the 'British & Irish' component, it obviously includes a strong Anglo-Saxon part of ancestry in common with Danish and Dutch people. I have seen one Dutch guy from Holland and a Flemish guy from Antwerp with 40% of 'British & Irish'. The average for Danes, Dutch and Flemings is 20-25%. It lower in the eastern and southern parts of the Benelux (15-20%). So 23andMe apparently failed to realise that their "500 year-old British & Irish" component was in fact a blend of ancient Insular Celtic (Gaelic, Brythonic) and ancient Anglo-Saxons.

When you know how to read the Ancestry Composition (what's hidden behind the component names) the percentages can be quite useful to compare populations. But I cannot stand that 23andMe would lie to their customers about how old these components are, and especially make them believe that they have ancestors from such and such region a few generations ago based on ancient admixtures. This is highly misleading, deceitful and even unethical. I have held 23andMe in high regard over the years, but this, combined with the decisions to remove health reports and updates for customers in most of the world (outside North America and parts of northern Europe), even those who paid a high price for the original test 10 years ago, is definitely a shift in policy in the wrong direction.

While we are at it, their Relatives Finder is also a scam, as are most similar services by other companies. Thanks to my detailed paper genealogy and very secluded ancestral region, I could easily verify if any of the dozens of "3rd to 5th cousins" listed in my results were really related in genealogical times. I contacted a few of them and it turns out that none had ancestry in Wallonia, ever. Most didn't even have any known Belgian ancestor. I think that the only reason they launched this tool is to get a share of all the naïve (mostly American) customers from Family Tree DNA and AncestryDNA who really believe that they are connecting with 3rd or 4th cousins based on what the site tells them. I have never understood why people would want to connect with a 3rd or 4th cousin anyway. I barely see my first and second cousins.

Your analysis is spot on. This newest feature is even more ridiculous than their other results, like their previous claim that they are only reflecting ancestry within the last 500 years. Like you I can trace most of my family lines relatively very far back, usually to the mid-1500s and everybody was in place, half of them stuck up on the northern Apennines, and half of them in the Lunigiana.

Yet, supposedly I got my Iberian, my French and a bunch of other things, including my .2 Korean three to four hundred years ago. It's absurd.

@Paschalis,
Their "Balkan" cluster is a mess. The reference population includes all their "Balkan" samples, including Greeks, and they even throw the Maltese in, for some unknown reason.

The "Italian" that's showing up is just "extra" Italian like affinity, other than what is included in the Balkan sample itself.

PaschalisB
25-03-17, 18:52
.

@Paschalis,
Their "Balkan" cluster is a mess. The reference population includes all their "Balkan" samples, including Greeks, and they even throw the Maltese in, for some unknown reason.

The "Italian" that's showing up is just "extra" Italian like affinity, other than what is included in the Balkan sample itself.
I had noticed that they include Maltese samples in the Balkan cluster and I also wondered why.

Sile
25-03-17, 19:09
If, as you say, the Eastern European component in more common among North Greeks, then it is much more likely to be of Slavic origin. Some of it could have been brought by the Goths a few centuries before the actual Slavic migrations, as the Goths were a mix of Germanic and (Proto-)Slavic tribes.

The French & German component is apparently a blend of Germanic (e.g. Frankish) and Gaulish/La Tène Celtic. In Belgium, Walloons (45%) score substantially higher than Flemings (36%) or of people from French Flanders (38%), which can be explained by the higher Gaulish ancestry, but also because the Germanic part of Walloon ancestry is almost exclusively Frankish, as opposed to Frankish, Saxon and Viking for the Flemings. The highest score I have seen on 23andMe was for one of my family members (70%), although South Germans (50-55%) seem to have a bit more in average than Walloons, while Swiss (45%) are similar to Walloons.

So how did the French & German get to Greece? Two possibilities:

1) Some of it came with La Tène Celts, who migrated along the Danube as far as central Anatolia. However they didn't really settle in Greece, so it would be by genetic drift and progressive mixing with neighbours over the last two millennia. Anatolian Greeks who moved to Greece after the independence could potentially have had more of it if the Galatians of Anatolia did carry that admixture.

2) The Goths. Although they were an East Germanic tribe, they would have carried a bit of similar 'French & German' as the Franks, Alemani and other tribes from Late Ancient/Early Medieval Germany. Even modern Danes have about 5% of 'French & German', while Swedes have about 3%. The Goths could also have had some Scandinavian and a bit of Finnish.

In short, I believe that the Goths are responsible for bringing Eastern European, French & German, Scandinavian and Finnish components to the Balkans.

While we are on the topic, I noticed that out of the 20 Italian Ancestry Compositions I saw on 23andMe, none had any Scandinavian, but most had some French & German. North Italians from Lombardy and Veneto typically score between 5 and 12% of French & German, against 0 to 2% for central and southern Italy. So it looks like the Lombards carried predominantly 'French & German' admixture like the Franks - unless of course it is of Celtic origin.

me a veneto ..............my family are all similar in numbers

Southern European 51.3%

Italian 27.7%

Iberian 5.9%

Broadly Southern European 17.8%


Northwestern European 39.1%

French & German 15.7%


Broadly Northwestern European 23.4%


Ashkenazi Jewish 0.1%


my phased results ...........father and then mother
Southern European 28.3%

Italian 17.6%
Balkan 0.6%
Iberian 2.4%
Sardinian 0.0%
Broadly Southern European 7.7%
Northwestern European 15.4%

French & German 11.8%

British & Irish 0.4%
Broadly Northwestern European 3.3%




Southern European 29.3%

Italian 18.2%
Balkan 2.8%
Iberian 0.0%
Sardinian 0.6%
Broadly Southern European 7.6%
Northwestern European 18.3%

French & German 11.6%
British & Irish 1.8%
Broadly Northwestern European 4.8%




maybe the 23andme french-german is the same as per this article

http://nofe.me/razib/WordPress/2017/03/your-ancestry-inference-is-precise-and-accurateish/

Maciamo
25-03-17, 20:11
me a veneto ..............my family are all similar in numbers

Southern European 51.3%

Italian 27.7%

Iberian 5.9%

Broadly Southern European 17.8%


Northwestern European 39.1%

French & German 15.7%


Broadly Northwestern European 23.4%


Ashkenazi Jewish 0.1%


That must be the least Italian results I have seen from Italy. So far anybody south of Emilia-Romagna scored between 70 and 90% Italian. Those from Piedmont have about 60-65%, Lombardy 40-60%, and Friuli-VG 30%. It's odd that Veneto is so low. Your Iberian score is also the highest I've seen in Italy.

Sile
25-03-17, 20:59
That must be the least Italian results I have seen from Italy. So far anybody south of Emilia-Romagna scored between 70 and 90% Italian. Those from Piedmont have about 60-65%, Lombardy 40-60%, and Friuli-VG 30%. It's odd that Veneto is so low. Your Iberian score is also the highest I've seen in Italy.

The iberian does not make sense when you add my parents ...........


add my parents french-german and its 23% ..........if you half for me it would be 11.5% , yet I get 15.7%


maybe my results reflect
paternally........ from present to 1590 in Veneto
1590-1430 Trentino ( south tyrol )

maternally ........from present to 1700 in Veneto
1700 - 1600 .........Carnian alps Friuli


clearly 23andme do not programme any paternal phasing into ones results and rely solely on each individual alignment with the samples they have...........IIRC they have only 12 north-italian samples ( someone can correct me on this )

even my timeline is odd

http://i103.photobucket.com/albums/m153/vicpret/timeline_zpsg8yeceo2.jpg (http://s103.photobucket.com/user/vicpret/media/timeline_zpsg8yeceo2.jpg.html)

Regio X
25-03-17, 21:51
That must be the least Italian results I have seen from Italy. So far anybody south of Emilia-Romagna scored between 70 and 90% Italian. Those from Piedmont have about 60-65%, Lombardy 40-60%, and Friuli-VG 30%. It's odd that Veneto is so low. Your Iberian score is also the highest I've seen in Italy.Not so odd. Likely the Italian % in Veneto would be higher if there were more reference samples from the region.
I have access to twelve 23andMe results of people with Venetian* ancestry, including my own and my parents', but not Sile's. The Italian % vary from ~13% (parents from TV) to ~47% (parents from TV as well), but both are outliers in this specific group of matches. Three have less than Sile (~13, ~26, ~26%), four have less than 30% and ten have less than 33%. The average is 31,0%.
As for Iberian in 23andMe, I myself have 4,9%, but indeed the average seems to be much lower than that. My guess is that trio-phasing have increased the Iberian percentage, mine and Sile's.
*However, one has a grandmother from N. MN (and ~33% of Italian) - very close to VR -, and other has a grandfather from UD (~30%).


The iberian does not make sense when you add my parents ...........why is mine doubled?Theoretically, because some of their Iberian % would be "hidden" in the Broadly Southern European.

Angela
25-03-17, 22:44
That must be the least Italian results I have seen from Italy. So far anybody south of Emilia-Romagna scored between 70 and 90% Italian. Those from Piedmont have about 60-65%, Lombardy 40-60%, and Friuli-VG 30%. It's odd that Veneto is so low. Your Iberian score is also the highest I've seen in Italy.

That's not quite what my shares show. I'm half northwest Tuscan/Ligurian and half Emilian, and I score 58% "Italian". Someone from LaSpezia with whom I share scores 54% "Italian". Someone from Piemonte also scores that. I share with a few 100% Tuscans, and they only score around 64% "Italian" to the best of my recollection*.

Those scores above are odd. People from Udine, Friuli, and the Swiss cantons with whom I share get more than that, some 34-38%, and some even more than that. Maybe those are the "conservative" numbers, which shouldn't be used. The "conservative" number is so conservative that it's meaningless.

Plus, for your real "Italian" score you should really add "Southern European" to "Italian". The only reason that the algorithm doesn't recognize some of those snps from the Veneto as "Italian" is because they have only a few samples from there. At least for Tuscans and adjoining areas they have the over 100 samples from 1000 genomes. That's one reason why northwest Italians are scoring higher "Italian"; there's a genetic link running all through northwest Italy and reaching into Toscana. Most of 23andme's Italian samples are from southern Italy, courtesy of all the Italian-Americans. That's the dense part of the "cluster" against which other Italians are measured.

*Why did they get rid of the feature where you could see the detailed Ancestry Comp results not only of your DNA relatives but of all your shares? Those people gave their consent for you to see it, but you can't? Now I would have to go back to my print outs. It's so annoying. Does anyone have the link to the archive of the forum? I used to be able to sign in to it. Knowledgeable people were still posting there occasionally. The new forum is useless.

Sile
25-03-17, 23:38
Plus, for your real "Italian" score you should really add "Southern European" to "Italian".

How does this work out ? .............the Italian score is already counted in southern European

More logical for me is that southern European also include the Italian Alps and that my 27% Italian is the non Alpine Italy and other 17.8% ( 51.3 minus 33.6% ( Italian +Iberian ) = 17.8% ) alpine italian

Sile
25-03-17, 23:47
Just checked my sons Phased ...........with me and his mother

me - Southern European 26.5%

Italian 14.6%
Iberian 2.7%
Broadly Southern European 9.2%
Northwestern European 16.5%

French & German 7.9%
Broadly Northwestern European 9.6%



his mother - Southern European 24.8%

Italian 15.1%
Iberian 3.2%
Broadly Southern European 6.5%
Northwestern European 22.6%

French & German 8.9%
Broadly Northwestern European 13.8%


I cannot work out why the Phasing results makes no sense to the parental individual results

Regio X
25-03-17, 23:51
Mine below. Comic. :)
http://i38.photobucket.com/albums/e102/paleoven/2017-03-25%2017.01.55_zps9roeyv20.png (http://s38.photobucket.com/user/paleoven/media/2017-03-25%2017.01.55_zps9roeyv20.png.html)
(nonsense)

@Angela
Try https://www.23andme.com/you/community/threads

I posted here before, but the text doesn't show up. It was said it would be moderated (or something like that). (?)

Angela
26-03-17, 00:13
Mine is below. Just comic. :)
http://i38.photobucket.com/albums/e102/paleoven/2017-03-25%2017.01.55_zps9roeyv20.png (http://s38.photobucket.com/user/paleoven/media/2017-03-25%2017.01.55_zps9roeyv20.png.html)

@Angela
Try https://www.23andme.com/you/community/threads

I posted here before, but the test doesn't show up. It was said it would be moderated (or something like that). (?)

That's a random feature of the program. I just approved it so it should show up.

Thanks for the link.

@Sile,
The way it works to get a slightly more realistic number is to add Broadly Southern European to the "Italian" speculative number.

If there were more samples from your area, and so you weren't only being compared to southern Italians, and to a lesser extent to Tuscans, the number would be different.

Maciamo
26-03-17, 09:01
That's not quite what my shares show. I'm half northwest Tuscan/Ligurian and half Emilian, and I score 58% "Italian". Someone from LaSpezia with whom I share scores 54% "Italian". Someone from Piemonte also scores that. I share with a few 100% Tuscans, and they only score around 64% "Italian" to the best of my recollection*.

Those scores above are odd. People from Udine, Friuli, and the Swiss cantons with whom I share get more than that, some 34-38%, and some even more than that. Maybe those are the "conservative" numbers, which shouldn't be used. The "conservative" number is so conservative that it's meaningless.

Good to know. I am only sharing with one full Tuscan from Massa and he scores 76% Italian.



*Why did they get rid of the feature where you could see the detailed Ancestry Comp results not only of your DNA relatives but of all your shares? Those people gave their consent for you to see it, but you can't? Now I would have to go back to my print outs. It's so annoying. Does anyone have the link to the archive of the forum? I used to be able to sign in to it. Knowledgeable people were still posting there occasionally. The new forum is useless.

The new 23andMe experience is a real let down, IMO. They removed the profile descriptions where people could list their location, birth country, ancestral surnames and interests. To make it worse, they also got rid of the search engine that allowed to browse profiles by country, so it has become virtually impossible to find people with whom to share now, unless they show up in your DNA Relatives results. For people like us interested in comparing haplogroups and ancestry compositions for research purposes, the new 23andMe has become useless. I think we should start a section for people to share their results. But unfortunately that will only include a small minority of 23andMe customers who are interested in ancient ancestry and population genetics research like us.

Maciamo
26-03-17, 09:04
The iberian does not make sense when you add my parents ...........


add my parents french-german and its 23% ..........if you half for me it would be 11.5% , yet I get 15.7%

...

How does this work out ? .............the Italian score is already counted in southern European

More logical for me is that southern European also include the Italian Alps and that my 27% Italian is the non Alpine Italy and other 17.8% ( 51.3 minus 33.6% ( Italian +Iberian ) = 17.8% ) alpine italian

You do understand that "phased results" means that you have to add up the results from each parent, right? So to infer each parent's composition, you need to multiply their scores by 2. Otherwise the score is only 50%.

Anyway, 23andMe's Ancestry composition is not very reliable. I have seen many times individuals with a few percents of some admixtures that wasn't present in their parents, which is normally impossible.

Sile
26-03-17, 19:15
You do understand that "phased results" means that you have to add up the results from each parent, right? So to infer each parent's composition, you need to multiply their scores by 2. Otherwise the score is only 50%.

Anyway, 23andMe's Ancestry composition is not very reliable. I have seen many times individuals with a few percents of some admixtures that wasn't present in their parents, which is normally impossible.

Ok

then from the post above for the phased numbers of my son ( post #15), ...........my numbers doubled are slightly out for my individual ones and his mother

whose paternal line and maternal line are recorded in registries for at least 300 years

Paternal line from Gairaine ( eastern Treviso province , Veneto )
Maternal line from Torre di Mosto (eastern Venice province , Veneto )

his mother - Southern European 49.6%

Italian 30.2%
Iberian 6.4%
Broadly Southern European 13%
Northwestern European 45.2%

French & German 17.8%
Broadly Northwestern European 27.6%


you can use these numbers in your data if you like


Again , odd that she has 0.4% Mongolian in 23andme ..........and 4% finnic in myorigins Ftdna

Regio X
27-03-17, 19:52
You do understand that "phased results" means that you have to add up the results from each parent, right? So to infer each parent's composition, you need to multiply their scores by 2. Otherwise the score is only 50%.I get what you mean. However, this inference won't be always realistic, of course. It depends on what one inherited from each parent. For example, if I multiply by 2 what I inherited from my father (or my mother) according to 23andMe, the resultant scores are very different from his (or her) actual scores.


Anyway, 23andMe's Ancestry composition is not very reliable. I have seen many times individuals with a few percents of some admixtures that wasn't present in their parents, which is normally impossible.
I questioned 23andMe on that a bit more than a year ago. They answered that trio-phasing increases the resolution of the set of results - parent(s)' and child's - but that a child's results increase in resolution even more. That's why, still in their words, a child may have more detailed assignments than their parents. They said discrepancies can arise when comparing what can be seen in a child's results versus what can be seen in their parent(s)' results. They said, additionally, that viewing the results in the speculative threshold - the least confident prediction of ancestry - magnifies any discrepancies when comparing results. This was the explanation, convincing or not.
So, for example, if each of my parents have 1% of Iberian and I have, say, 5%, then my parents supposedly have some Iberian % "hidden" in the Broadly Southern European - or Broadly European (?).

@Angela
It happened again. :) Please see.

Vallicanus
18-07-17, 17:25
That must be the least Italian results I have seen from Italy. So far anybody south of Emilia-Romagna scored between 70 and 90% Italian. Those from Piedmont have about 60-65%, Lombardy 40-60%, and Friuli-VG 30%. It's odd that Veneto is so low. Your Iberian score is also the highest I've seen in Italy.


In the recent update I received 57.8pc Italian and my ancestors are from the province of Lucca in NW Tuscany.

Vallicanus
18-07-17, 17:29
In my case there are a few medieval links to Bergamo/Brescia in Lombardy though because of the immigration of iron workers into the Serchio valley.

Angela
18-07-17, 21:12
The only place in Toscana where people seem to get over 70% "Italian" is right on the coast near Massa. The northwestern area (Lunigiana and Garfagnana) seems to get around 58%, and the rest of Toscana in the 60s. What it means I'm not entirely sure, except that 23andme's "centrum" seems to be southern Italy and to a lesser degree Toscana.

I absolutely know that their "timeline" is inaccurate. For half of my ancestors they didn't leave their Emilian Apennines perch from the 1400's until the time of my grandparents. For most of my mother's family it's the same. Their results make no sense. As with the Alder program, all these admixtures are dated way too late.

Jovialis
11-11-17, 23:16
https://i.imgur.com/V8joq3N.png

This is how v5 groups the populations.

I agree that the timeline is totally useless.

Promenade
11-11-17, 23:34
https://i.imgur.com/V8joq3N.png
This is how v5 groups the populations.

Maltese in the balkans, lol...

Angela
11-11-17, 23:41
Maltese in the balkans, lol...

As I've been saying for years to the ******, that's why the Near Eastern proportions look lower in the Balkans in 23andme than they do through ADMIXTURE and formal stats. It only picks up "farmer" type ancestry which is in excess of what is already present in Greeks and the Maltese.

Likewise, Middle East is, or used to be heavily weighted toward the Caucasus, Iran, Turkey, etc. Palestinians and Saudis are in with North Africans. Perhaps Jordanians as well, which is what people usually think of when they think of Middle Eastern or Levantine populations.

When people are pushing an agenda, however, they have no intention of letting facts get in the way.

davef
11-11-17, 23:46
^^ i caught that too, i expected Maltese to be more Siciilian than Balkan since it's located close to Sicily. I could be wrong, though

davef
11-11-17, 23:52
Whoops, never should've put those arrows up in my previous post, they were pointing at promenade's post.

Sile
12-11-17, 00:05
me a veneto ..............my family are all similar in numbers
Southern European 51.3%
Italian 27.7%
Iberian 5.9%
Broadly Southern European 17.8%
Northwestern European 39.1%
French & German 15.7%
Broadly Northwestern European 23.4%
Ashkenazi Jewish 0.1%
my phased results ...........father and then mother
Southern European 28.3%
Italian 17.6%
Balkan 0.6%
Iberian 2.4%
Sardinian 0.0%
Broadly Southern European 7.7%
Northwestern European 15.4%
French & German 11.8%

British & Irish 0.4%
Broadly Northwestern European 3.3%

Southern European 29.3%
Italian 18.2%
Balkan 2.8%
Iberian 0.0%
Sardinian 0.6%
Broadly Southern European 7.6%
Northwestern European 18.3%
French & German 11.6%
British & Irish 1.8%
Broadly Northwestern European 4.8%
maybe the 23andme french-german is the same as per this article
http://nofe.me/razib/WordPress/2017/03/your-ancestry-inference-is-precise-and-accurateish/
mine changed recently ....still the numbers are from v3
....
European 99.2% ............................up 0.7
Southern European 57.6%................up 6.0
Italian 35.8% ..................................up 8.1
Balkan 3.4% ......................................up 3.4
Iberian 2.4% .....................................down 3.5
Sardinian 0.6%
Broadly Southern European 15.3% ...................down 2.5
Northwestern European 33.7% ....................down 5.8
French & German 13.4% ..........................down 2.3
British & Irish 2.2% ....................................up 2.2
Broadly Northwestern European 18.1% ..................down 5.3
.................................................. .........
father phased
Southern European 28.3%
Italian 17.6%
Balkan 0.6%
Iberian 2.4%
Sardinian 0.0%
Broadly Southern European 7.7%
Northwestern European 15.4%
French & German 1.8%
British & Irish 0.4%
Broadly Northwestern European 13.3%
.................................................. .........
mother phased
Southern European 29.3%
Italian 18.2%
Balkan 2.8%
Iberian 0.0%
Sardinian 0.6%
Broadly Southern European 7.6%
Northwestern European 18.3%
French & German 11.6%
British & Irish 1.8%
Broadly Northwestern European 4.8%


They have also fixed the chromosome bars ...for me , maternal is top and paternal is bottom

Jovialis
12-11-17, 00:20
As I've been saying for years to the ******, that's why the Near Eastern proportions look lower in the Balkans in 23andme than they do through ADMIXTURE and formal stats. It only picks up "farmer" type ancestry which is in excess of what is already present in Greeks and the Maltese.
Likewise, Middle East is, or used to be heavily weighted toward the Caucasus, Iran, Turkey, etc. Palestinians and Saudis are in with North Africans. Perhaps Jordanians as well, which is what people usually think of when they think of Middle Eastern or Levantine populations.
When people are pushing an agenda, however, they have no intention of letting facts get in the way.

https://i.imgur.com/pNYPQzc.png

This is how they're grouping the middle east and North Africa for V5.

davef
12-11-17, 00:30
North African includes Bedouins, Arabians, Palestinians, Jordanians, and Yemenis.....

davef
12-11-17, 08:41
Maltese in the balkans, lol...

Allow me to contribute to 23andme's new version. I'll create a new category, let's call it "French".

French
(Spanish, Portuguese, French, North Italian, British, Irish, German, Swiss, Dutch, Belgian).

The 23andme support team has its work cut out, big time. It had better be well prepared for an onslaught of questions from confused customers.

AdeoF
12-11-17, 12:35
https://i.imgur.com/pNYPQzc.png

This is how they're grouping the middle east and North Africa for V5.

I think 23andme team needs to go back to school and re-do there geography, I don't think they know anything about it nor the genetics in the regions(like even the simple one that berbers and arabs are different). I don't think I would get a test from them anytime soon.

Jovialis
12-11-17, 13:40
I think 23andme team needs to go back to school and re-do there geography, I don't think they know anything about it nor the genetics in the regions(like even the simple one that berbers and arabs are different). I don't think I would get a test from them anytime soon.

I don't think it's bad, but like with all of these tests it requires the person to do a lot of research into genetics. They have to know the history of their background, and what these terms imply. I think the geneplaza K12 calculator helped to clarify a lot for me. For example, my Balkan admixture (https://www.eupedia.com/forum/threads/34450-Iapygians?p=523996&viewfull=1#post523996) comes from ancient times (Greeks & Messapii), not the more recent populations; based on the ancient samples and the history of that area. I'm only from one area, so its easier for me to get an idea of what the results mean.

I think 23andme is one of the better DNA testing kits; much better than National Geographic was.

@Davef, it's not just the new version that groups them like this. If you read further back in the thread, the previous versions did as well. There's nothing inherently wrong with V5, and it's actually an improvement from V4. Nevertheless, V3 tested the most SNPs.

https://i.imgur.com/mfxZers.png

Jovialis
12-11-17, 15:37
mine changed recently ....still the numbers are from v3
....
European 99.2% ............................up 0.7
Southern European 57.6%................up 6.0
Italian 35.8% ..................................up 8.1
Balkan 3.4% ......................................up 3.4
Iberian 2.4% .....................................down 3.5
Sardinian 0.6%
Broadly Southern European 15.3% ...................down 2.5
Northwestern European 33.7% ....................down 5.8
French & German 13.4% ..........................down 2.3
British & Irish 2.2% ....................................up 2.2
Broadly Northwestern European 18.1% ..................down 5.3
.................................................. .........
father phased
Southern European 28.3%
Italian 17.6%
Balkan 0.6%
Iberian 2.4%
Sardinian 0.0%
Broadly Southern European 7.7%
Northwestern European 15.4%
French & German 1.8%
British & Irish 0.4%
Broadly Northwestern European 13.3%
.................................................. .........
mother phased
Southern European 29.3%
Italian 18.2%
Balkan 2.8%
Iberian 0.0%
Sardinian 0.6%
Broadly Southern European 7.6%
Northwestern European 18.3%
French & German 11.6%
British & Irish 1.8%
Broadly Northwestern European 4.8%
They have also fixed the chromosome bars ...for me , maternal is top and paternal is bottom

Interesting, they must be applying the updated algorithm (https://www.eupedia.com/forum/threads/33825-Share-your-23andMe-Ancestry-Composition/page8?p=523195&viewfull=1#post523195) the new users got.

Angela
12-11-17, 17:24
I think Jovialis is right. It's still the best test out there.

Geographical borders really don't necessarily overlap with genetic borders, which is why genome testing for consumers is very difficult. In terms of the Balkans, including Malta was just wrong. It should definitely be with Italy. On the other hand, I think I do understand what is going on with the Middle East and North Africa. North Africa was settled by farmers from the Levant. There was, I'm sure, constant gene flow from the Middle East into the region, up to and including the Arabic invasions of the early Medieval period. It's true, of course, that they have very low levels of Iran like ancestry, so one could make the argument that they should split the larger area into an Iran like group or northern West Asian, and a Levant like/North African grouping. However, the Levant has been receiving gene flow from Iran since right after the beginning of the Neolithic. There's also been some gene flow in the opposite direction, as in Armenia. So, where to draw the lines? I think what they did is to divide the heavily East African and SSA admixed groups from the rest, so Yemen, Saudi, Palestinians and Jordanians were grouped with North Africans. It's imperfect, but I don't know the ideal division.

The problem is really with the naming. When people see Middle East as a result in 23andme, they think Palestinian, Jordanian, and Saudi, maybe even Egypt, but it's not correct. That would be represented by the North African score.

As someone pointed out, this has always been the division in 23andme.

The timeline, however, is best just ignored. No one but Ralph and Coop has gotten the timing of admixture basically right just from looking at modern genomes. All the rest make it too recent, imo.