PDA

View Full Version : 23andme guesses ! False results!



dhunter
10-18-2015, 03:03 PM
Hi all,

Thought some of you might find this interesting:

I had 23andme do my dna and they said my mtdna was N1b2 and that this group was made up European Jews. Being N1b2 made me half Jewish In "Speculative" based on 23andme's model.

I had been tested prior by Oxford ancestors and so I compared the results. I discovered that I do not have the transversion at 16176G to A and therefor not in N1b2.

I contacted 23andme and was told that they based their assumption on the fact I am 98% European and have N1b mtdna, their model assumes that I must then have the transversion from at 16176 G to A putting me in N1b2 group even though I do not.

They advised me that they "do not do a complete test", 23andme said that there broad testing model is "correct for the most part" and that although in my case they are mistaken they can not correct it.

The bottom line is 23andme does not provide accurate information and you should pay the extra and have your DNA done by a company that tests for all the makers.

AnnieD
10-18-2015, 07:25 PM
My MtDNA results over two years ago with 23andMe were very major-clade oriented, e.g. H haplogroup plus 1 subclade = "H1." In Europe, haplogroup H1 is reputedly the largest subclade of the very common "H," so this is a fairly generic result.

Unfortunately, I believe that they are on record for stating a lack of intent to offer further subclade testing. (My own perception, and possibly of many others, is that they are a bio-health oriented firm who used the ancestry tests as a marketing tool to further their real interests. Then they were side-lined by the USA Food-and-Drug Administration, so don't hold your breath that will start offering a Big Y-like test like FTDNA any time soon :\

In contrast, FTDNA and TribeCode provided my subclade out 3 more levels. In my case, this does not result in any ethnic 'confusion' as it may with your N subclades as even my full subclade level has matches from Morocco, Lithuania, Poland, Sweden, all over British Isles and down to Spain. You may wish to join the 23andMe thread that poses questions or issues with 23andMe services and results. However, my own 'issue' with this thread is that they tend to 'preach to the choir (forum)', e.g. no one seems to actually reach out to management directly with their concerns. ;)

"Questions for 23andMe"
https://www.23andme.com/you/community/thread/39366/

Petr
10-18-2015, 08:00 PM
According to http://www.isogg.org/wiki/Mitochondrial_DNA_tests 23andMe uses mtDNA tree Build 7, in Build 7 N1b2 is similar to current N1b1b.
Build 7 (10 Nov 2009) N1b2 defining mutations: 4735A 4917 11928 12092 13129 13710 14581 16176A
Build 16 (19 Feb 2014) N1b1b defining mutations: 4735A 4917 11928 12092 13129 13710 16176

To obtain haplogroup according to the current phylogenetic tree using your 23andMe data, I'd recommend to use this tool: http://dna.jameslick.com/mthap-new/

dhunter
10-19-2015, 12:10 AM
According to http://www.isogg.org/wiki/Mitochondrial_DNA_tests 23andMe uses mtDNA tree Build 7, in Build 7 N1b2 is similar to current N1b1b.
Build 7 (10 Nov 2009) N1b2 defining mutations: 4735A 4917 11928 12092 13129 13710 14581 16176A
Build 16 (19 Feb 2014) N1b1b defining mutations: 4735A 4917 11928 12092 13129 13710 16176

To obtain haplogroup according to the current phylogenetic tree using your 23andMe data, I'd recommend to use this tool: http://dna.jameslick.com/mthap-new/



Hi, I am not sure you understood what I posted.

23andme does not do a complete test. They assume that based on certain broder data that you will possess other mutations. In my case they put me in the wrong mtdna subcategory.

So suggesting I transfer their data to another service will not alter the mistake 23andme made in my coding. The other sight will read the false 23andme mtdna data that has me having the transversion 16176A putting me in N1b2. I do not have this transversion, I am 16176G. 23andme will not correct their data so it is useless.

Garbage in garbage out.

AJL
10-19-2015, 12:30 AM
I had 23andme do my dna and they said my mtdna was N1b2 and that this group was made up European Jews. Being N1b2 made me half Jewish In "Speculative" based on 23andme's model.
If I understand you right you are talking about two different unconnected things here.

Your mtDNA assessment of N1b2 may very well be incorrect. Shotgun/chip technology doesn't deal with mtDNA transversions very well.

How your Ancestry Composition is assessed in speculative mode, however, is unconnected to your mtDNA or yDNA. Someone could have yDNA J1 and mtDNA K1a9 and still show 0% Ashkenazi. It is nigh on impossible that any person could get a result of 50% Ashkenazi on their autosomal test at 23andme unless they were in fact 50% Ashkenazi, within an insignificant margin of error.

dhunter
10-19-2015, 02:49 AM
If I understand you right you are talking about two different unconnected things here.

Your mtDNA assessment of N1b2 may very well be incorrect. Shotgun/chip technology doesn't deal with mtDNA transversions very well.

How your Ancestry Composition is assessed in speculative mode, however, is unconnected to your mtDNA or yDNA. Someone could have yDNA J1 and mtDNA K1a9 and still show 0% Ashkenazi. It is nigh on impossible that any person could get a result of 50% Ashkenazi on their autosomal test at 23andme unless they were in fact 50% Ashkenazi, within an insignificant margin of error.



:) So you think that 23andme does full autosomal test without making guesses based on their model, well they don't.

Also, you appear quite sure that when 23andme says I'm 50% Askenazi in Speculative then they must be correct? Well according to them it's speculative has a maximum 50% probability :) You have way more faith in them then I.



P.S. look at the over sample of a certain population.


Population Source Sample Size
Ashkenazi 23andMe 1305
United Kingdom 23andMe 870
Italy 23andMe 556
Germany 23andMe 367
Ireland 23andMe 292
Russia 23andMe 253
Finland 23andMe 245
Poland 23andMe 240
Spain 23andMe 234
Norway 23andMe 220
Netherlands 23andMe 207
France 23andMe 200
Sweden 23andMe 170
Greece 23andMe 147
Portugal 23andMe 132
Romania 23andMe 129
Denmark 23andMe 125
Italy 1000 Genomes 98
Finland 1000 Genomes 93
United Kingdom 1000 Genomes 89
Switzerland 23andMe 87
Hungary 23andMe 86
Ukraine 23andMe 82
Belgium 23andMe 80
Bulgaria 23andMe 75
Austria 23andMe 54
Croatia 23andMe 51
Slovakia 23andMe 45
Czech Republic 23andMe 44
Belarus 23andMe 31
France HGDP 29
Sardinian HGDP 28
Bosnia and Herzegovina 23andMe 25
Russian HGDP 25
France Basque HGDP 24
Slovenia 23andMe 20
Serbia 23andMe 17
Spain 1000 Genomes 14
North Italian HGDP 13
Macedonia 23andMe 12
Albania 23andMe 8
Tuscan HGDP 8
Montenegro 23andMe 6
Malta 23andMe 6

Táltos
10-19-2015, 03:29 AM
Hi all,

Thought some of you might find this interesting:

I had 23andme do my dna and they said my mtdna was N1b2 and that this group was made up European Jews. Being N1b2 made me half Jewish In "Speculative" based on 23andme's model.

I had been tested prior by Oxford ancestors and so I compared the results. I discovered that I do not have the transversion at 16176G to A and therefor not in N1b2.

I contacted 23andme and was told that they based their assumption on the fact I am 98% European and have N1b mtdna, their model assumes that I must then have the transversion from at 16176 G to A putting me in N1b2 group even though I do not.

Please explain what you mean a little more. First 23andme does not do a full mitochondria sequencing test. Only Family Tree DNA offers that. Are you on their v4 chip? That uses imputing, but there is supposed to be more mtDNA SNPs on that chip compared to the v3, and v2. http://snpedia.com/index.php/23andMe_v4_differences

Second just because you *might* have a Jewish haplogroup on one of your lines does not automatically mean you will be half Jewish. Now you might be at least 50% Ashkenazi if you have recent Jewish ancestors. You probably won't be if don't have any recent Jewish ancestors. Haplogroups are a small part of our ancestry, and are very old.

Mbantua
10-19-2015, 03:49 AM
The problem here is not dhunter's mtDNA haplogroup or the reference groups 23andMe uses in Ancestry Composition or the number of haplogroup specific SNPs contained in the v2, v3 or v4 chips. The problem is that dhunter is rejecting his AC because it says his ancestry is 50% Ashkenazi Jewish and he is doing everything to blame the messenger, 23andMe and not doing the mature thing and finding his Ashkenazi Jewish parent or seeking counseling or just have a talk to his on paper parents.

Lots of people have found out one of their parents is not their biological parent via dna testing, dhunter is not the first and won't be the last

dhunter
10-19-2015, 04:23 AM
The problem here is not dhunter's mtDNA haplogroup or the reference groups 23andMe uses in Ancestry Composition or the number of haplogroup specific SNPs contained in the v2, v3 or v4 chips. The problem is that dhunter is rejecting his AC because it says his ancestry is 50% Ashkenazi Jewish and he is doing everything to blame the messenger, 23andMe and not doing the mature thing and finding his Ashkenazi Jewish parent or seeking counseling or just have a talk to his on paper parents.

Lots of people have found out one of their parents is not their biological parent via dna testing, dhunter is not the first and won't be the last



Funny that you had to create a new profile to post, coward :)


I have Ashkenazi family history you rube, the issue is the percentage that 23andme achieves from an over sample of a self identified group. Either you don't comprehend statistics and sample populations or you are just being ignorant. The bottom line is that 23andme is inaccurate, I challenge you to do what I did, have a creditable DNA test done and compare the results to 23andme.

Táltos
10-19-2015, 04:39 AM
Funny that you had to create a new profile to post, coward :)


I have Ashkenazi family history you rube, the issue is the percentage that 23andme achieves from an over sample of a self identified group. Either you don't comprehend statistics and sample populations or you are just being ignorant. The bottom line is that 23andme is inaccurate, I challenge you to do what I did, have a creditable DNA test done and compare the results to 23andme.

Thank you for pointing out that you have known Ashkenazi ancestry. How far back? Do you have lots of autosomal matches to other people who are Jewish?

I have to say I like 23andme for ancestry composition. But I'm not a fan of their conservative mode as my mom and myself get so much Unassigned in that mode. 20.7% for my mom, and 11.3% for me.

I have no known Jewish ancestry to my knowledge. DNA testing revealed I have several Jewish matches along one of my chromosomes. 23andme puts my Ashkenazi at 0.9% in all three modes. The likely source of this ancestry is my direct paternal line. So far records I have reveal no Jewish ancestry in this line back to the 1700s. So it is further back. I hope that helps you for comparison. FTDNA puts my Ashkenazi at 5%.

Táltos
10-19-2015, 04:42 AM
Funny that you had to create a new profile to post, coward :)

Who do you think this is?

dhunter
10-19-2015, 01:28 PM
Thank you for pointing out that you have known Ashkenazi ancestry. How far back? Do you have lots of autosomal matches to other people who are Jewish?

I have to say I like 23andme for ancestry composition. But I'm not a fan of their conservative mode as my mom and myself get so much Unassigned in that mode. 20.7% for my mom, and 11.3% for me.

I have no known Jewish ancestry to my knowledge. DNA testing revealed I have several Jewish matches along one of my chromosomes. 23andme puts my Ashkenazi at 0.9% in all three modes. The likely source of this ancestry is my direct paternal line. So far records I have reveal no Jewish ancestry in this line back to the 1700s. So it is further back. I hope that helps you for comparison. FTDNA puts my Ashkenazi at 5%.



Hi,

I am not sure how far back, although that's a tough question because Ashkenazi Jews have European DNA.

The fact that Ashkenazi Jews are European and are allowed to self identify creating an issue where genetic composition gets conflated with religion rather than a region.

As far as your Jewish matches, it becomes a Chicken or egg argument. Because the DNA of the person is European, and they entered the Jewish faith, does that make you part Jewish?

Táltos
10-19-2015, 01:47 PM
Hi,

I am not sure how far back, although that's a tough question because Ashkenazi Jews have European DNA.

The fact that Ashkenazi Jews are European and are allowed to self identify creating an issue where genetic composition gets conflated with religion rather than a region.

As far as your Jewish matches, it becomes a Chicken or egg argument. Because the DNA of the person is European, and they entered the Jewish faith, does that make you part Jewish?

Okay you say you're not sure how far back. Not even a guess of great great grandparent? Well you are really vague in your information and seem to want to make a fuss about Ashkenazi ancestry on this forum.

They are a distinct group that have ties to the Middle East and Europe. You can start by looking for links to studies here as I will not spoon feed them. This is a topic that is beaten to death already. https://en.wikipedia.org/wiki/Genetic_studies_of_Jewish_origins
Best wishes on your search.

dhunter
10-19-2015, 02:13 PM
Okay you say you're not sure how far back. Not even a guess of great great grandparent? Well you are really vague in your information and seem to want to make a fuss about Ashkenazi ancestry on this forum.

They are a distinct group that have ties to the Middle East and Europe. You can start by looking for links to studies here as I will not spoon feed them. This is a topic that is beaten to death already. https://en.wikipedia.org/wiki/Genetic_studies_of_Jewish_origins
Best wishes on your search.

I'm not making a fuss ? Vague information?

The topic of this post was so people won't get ripped off from 23andme. The information they provide is not correct and the public interested in DNA should be made aware.

Also please explain the Ashkenazi ties to the middle east? The DNA does not support this now.

Petr
10-19-2015, 02:18 PM
23andMe tests 3 markers for position 16176, i4001354, i4001355, i4001356, all three are incorrect?

In build 7 that is used by 23andMe the difference between N1b and N1b2 consists of 14 markers, see https://www.23andme.com/you/labs/haplogroup_tree_mut_mapper/
N1b2 defining mutations
variant call rCRS anc -> der
i4000871 4735 C -> A
rs28357980 4917 A -> G
i3002607 4917 A -> G
i5050504 4917 A -> G
i4001017 11928 A -> G
i5050477 11928 A -> G
i3001076 12092 C -> T
i4000927 13129 C -> T
i3001213 13710 A -> G
i5050293 13710 A -> G
i4000581 14581 T -> C
i4001354 16176 C -> A
i4001355 16176 C -> A
i4001356 16176 C -> A

All of them are incorrect?

23andMe should test all of them. Not guess. The test may contain errors.

I see no relation between mtDNA and ancestry report.

Petr
10-19-2015, 02:22 PM
The information they provide is not correct and the public interested in DNA should be made aware.
I wouldn't say that. Yes, the number of markers is limited, but I made 23andMe tests for 24 persons and all markers were correct - in comparison to FTDNA.

AJL
10-19-2015, 02:24 PM
Funny that you had to create a new profile to post, coward :)

Whom are you addressing?

dhunter
10-19-2015, 02:44 PM
23andMe tests 3 markers for position 16176, i4001354, i4001355, i4001356, all three are incorrect?

In build 7 that is used by 23andMe the difference between N1b and N1b2 consists of 14 markers, see https://www.23andme.com/you/labs/haplogroup_tree_mut_mapper/
N1b2 defining mutations
variant call rCRS anc -> der
i4000871 4735 C -> A
rs28357980 4917 A -> G
i3002607 4917 A -> G
i5050504 4917 A -> G
i4001017 11928 A -> G
i5050477 11928 A -> G
i3001076 12092 C -> T
i4000927 13129 C -> T
i3001213 13710 A -> G
i5050293 13710 A -> G
i4000581 14581 T -> C
i4001354 16176 C -> A
i4001355 16176 C -> A
i4001356 16176 C -> A

All of them are incorrect?

23andMe should test all of them. Not guess. The test may contain errors.

I see no relation between mtDNA and ancestry report.



Okay last attempt, then you guys win.

In my case 23andme guessed and they guessed wrong. The only way I know is because I was tested elsewhere.

When I contacted 23andme they advised me that they only test some strands of DNA and apply their Algorithm to assume the rest. They do not do a complete test.

Lastly, oversampling and allowing people who are Ashkenazi to self identify makes the data suspect.

Guessing, oversampling, self identification, plus a low rate of autosomal success makes 23andme a waist of money.

Accuracy of Autosomal tests:

50% (parents and siblings)
25% (grandparents, aunts/uncles, half-siblings, double first-cousins)
12.5% (first cousins)
6.25% (first cousins, once removed)
3.125 (second cousins, first cousins twice removed)
0.781% (third cousins)
0.195% (fourth cousins)

Táltos
10-19-2015, 03:02 PM
Okay last attempt, then you guys win.

In my case 23andme guessed and they guessed wrong. The only way I know is because I was tested elsewhere.

When I contacted 23andme they advised me that they only test some strands of DNA and apply their Algorithm to assume the rest. They do not do a complete test.

Lastly, oversampling and allowing people who are Ashkenazi to self identify makes the data suspect.

Guessing, oversampling, self identification, plus a low rate of autosomal success makes 23andme a waist of money.

Accuracy of Autosomal tests:

50% (parents and siblings)
25% (grandparents, aunts/uncles, half-siblings, double first-cousins)
12.5% (first cousins)
6.25% (first cousins, once removed)
3.125 (second cousins, first cousins twice removed)
0.781% (third cousins)
0.195% (fourth cousins)

No one is trying to win. Your statements seem to show disdain for Ashkenazi. The part of your statement in bold. Well if you want to say that then all of Western Europe also self identifies. Does that also make these tests suspect for you? I used them as an example as populations from that region of the world are also over sampled.

You do not inherit autosomal DNA perfectly, and may receive more from one grandparent vs another. You have been vauge. If you are serious you would have offered a guesstimate of how far back your Jewish ancestor is, instead of posing more questions to me. You also would have indicated which chip you are on at 23andme, furthermore you could have posted all of your results in all three modes from that company.

Admixture tests are not perfect, one company may do a better job at picking up your KNOWN ancestry than another.

We have already established that 23andme does not test all regions of the mtDNA. http://www.isogg.org/wiki/MtDNA_testing_comparison_chart

dhunter
10-19-2015, 03:15 PM
No one is trying to win. Your statements seem to show disdain for Ashkenazi. The part of your statement in bold. Well if you want to say that then all of Western Europe also self identifies. Does that also make these tests suspect for you? I used them as an example as populations from that region of the world are also over sampled.

You do not inherit autosomal DNA perfectly, and may receive more from one grandparent vs another. You have been vauge. If you are serious you would have offered a guesstimate of how far back your Jewish ancestor is, instead of posing more questions to me. You also would have indicated which chip you are on at 23andme, furthermore you could have posted all of your results in all three modes from that company.

Admixture tests are not perfect, one company may do a better job at picking up your KNOWN ancestry than another.

We have already established that 23andme does not test all regions of the mtDNA. http://www.isogg.org/wiki/MtDNA_testing_comparison_chart



I do not have disdain for Ashkenazi Jews, I have disdain for pseudoscience! (if you don't know the meaning of pseudoscience please look it up)

We can let those who read this post decide who is posting facts vs opinion.

Good luck, I give up.

AJL
10-19-2015, 04:10 PM
You can have a one-week vacation for your generally vitriolic tone and personal attacks.

During the one week you may wish to acquaint yourself with the forum rules (http://www.anthrogenica.com/faq.php), more particularly:

3.11 Invectives and posts devoid of substance (e.g. threads or replies consisting solely of inflammatory content or triviality) will be considered junk postings and deleted. Anthrogenica encourages its members to engage in debate with each other, but discussions that degenerate into little more than flaming and inanity will be considered spam (see Rule 3.9).

vettor
10-19-2015, 05:30 PM
No one is trying to win. Your statements seem to show disdain for Ashkenazi. The part of your statement in bold. Well if you want to say that then all of Western Europe also self identifies. Does that also make these tests suspect for you? I used them as an example as populations from that region of the world are also over sampled.

You do not inherit autosomal DNA perfectly, and may receive more from one grandparent vs another. You have been vauge. If you are serious you would have offered a guesstimate of how far back your Jewish ancestor is, instead of posing more questions to me. You also would have indicated which chip you are on at 23andme, furthermore you could have posted all of your results in all three modes from that company.

Admixture tests are not perfect, one company may do a better job at picking up your KNOWN ancestry than another.

We have already established that 23andme does not test all regions of the mtDNA. http://www.isogg.org/wiki/MtDNA_testing_comparison_chart

maybe he was confused by 23andme use of the ashkenazi marker. I once asked 23andme about it and they stated they use Sephardic markers as named ashkenazi too in their admixture.

Anyway, as for me, my ashkenazi started out as 0.6%, as I added my son to 23andme my askenazi dropped to 0.4% and this month after adding my father to 23andme it dropped for me to 0.2% ...............So I do not know if adding more family members gives one more accuracy . But I also posed another question to 23andme and that is ........my ashkenazi sits in the first quarter of chromosome 6 which is the medical part , is it accurate for ashkenazi.......I never received a reply.

In the end , I use conservative and it shows 0.1% ashkenazi and for me the only disappointing part is to see if it was sephardic or not.

to conclude, my split view for my father states my ashkenazi as well as my middle-east came via my mother side as my father does not have either.

AJL
10-19-2015, 09:12 PM
^ I think we can all understand doubting the relevance and directionality of something in the area of 0.4-0.6%. 50% is a completely different story.