MyOrigins - MANY Discrepancies - please list yours here

[Germanica]

Yes, my highest percentage is in European Northlands and European Coastal Plain. It's my understanding that the Northlands fed into the European Coastal Plain (Great Britain). I have two lines which are Scottish so I can understand that.

Are you saying the German is actually being represented by the European Coastal Plain also?

What I'm saying is that people with strong British ancestry are getting percentages too low (or not at all) for the British region, while they are getting high numbers for German and Scandinavian. And vice versa - people with strong Scandinavian or German ancestry (and sometimes no British ancestors) are getting percentages too low (or not at all) for those regions and too high for British.

The British do have Scandinavian and Anglo-Saxon (Germanic tribes) in their ancestry and so it's understandable that these three regions are genetically close to one another. But someone with no British ancestors and lot of German ancestors shouldn't be getting a high percentage of British and zero percent German, which is what some people are reporting. It sounds like you're experiencing something similar.

[ISC80]

MyOrigins not only has difficulty differentiating between these two clusters (European Coastal Islands vs. European Coastal Plain), but also between the European Coastal Plain cluster (ECP) and the North Mediterranean Basin cluster (NMB).

Although myOrigins uses 124 samples for Spain (one of the best populations represented), I have noticed that there are many discrepancies in the Spanish results. For example, my father is 77% NMB and 20% ECP. Instead, I'm 60% ECP and 32% NMB. This disparity in results may seem strange because we both have the same roots in the south of the Iberian Peninsula.
Moreover, I've also read that some people from Central Europe have high percentages of NMB cluster, without having ancestors from Southern Europe.

I just think that these general categories overlap each other because are pretty similar, actually.

[Tenn4ever]

MyOrigins not only has difficulty differentiating between these two clusters (European Coastal Islands vs. European Coastal Plain), but also between the European Coastal Plain cluster (ECP) and the North Mediterranean Basin cluster (NMB).

Although myOrigins uses 124 samples for Spain (one of the best populations represented), I have noticed that there are many discrepancies in the Spanish results. For example, my father is 77% NMB and 20% ECP. Instead, I'm 60% ECP and 32% NMB. This disparity in results may seem strange because we both have the same roots in the south of the Iberian Peninsula.
Moreover, I've also read that some people from Central Europe have high percentages of NMB cluster, without having ancestors from Southern Europe.

I just think that these general categories overlap each other because are pretty similar, actually.

Then why can GEDmatch differentiate but FTDNA is unable to. GEDmatch is using my raw data from FTDNA and GEDmatch is giving results which best match to my paper trail which they don't even have access to.???? FTDNA doesn't see to be able to do this with the exact same data.

[Mike McG]

I was born in the London/Middlesex area of southern England. My mother's ancestry is English Protestant also mainly from the London/Middlesex area in the 1800s and earlier with some ancestors from other nearby southern England counties. My father's ancestry is Irish Catholic. He was born near Cashel, County Tipperary, Ireland and all his ancestors are from that area back to the late 1700s and were probably in Tipperary or nearby counties for centuries before that.

FTDNA Population Finder showed my ancestry as 100.00% ±0.01% Orcadian, which I understood to mean from the British Isles in general rather than specifically from Orkney; and essentially matched my expectation.

My recent 23&me results in Speculative Mode also match my expectations. They were as follows:
99.8% European
0.2% Unassigned

99.4% Northern European
82.0% British & Irish
3.9% French & German
0.6% Scandinavian
12.9% Nonspecific Northern European

0.4% Nonspecific European
0.0% for all other 23&me populations

The 3.9% French & German is predominately the whole of my X chromosome and so presumably from my mother side of my family. This is not surprising since it is much more likely for an immigrant into the London area (although I have not found them) than for a Catholic into Tipperary in the past say 500 years.

I was therefore surprised when my FTDNA My Origins showed 13% Trans-Ural Peneplain as follows:
100% European
63% European Coastal Islands
24% European Coastal Plain
13% Trans-Ural Peneplain

The 13% is essentially equivalent to one great-grandparent, and I believe this is incorrect and misleading in a genealogical time frame where FTDNA is linking My Origins to Family Finder. Until I find a ancestral connection to Eastern Europe (which I do not expect) I have decided to keep these My Origins results private.

Mike McG

[greenpea]

Population Finder results

My Population Finder results show 86.4% Middle East Jewish and 12.8% European, specifically Subgroup Basque-Spain-Sardinia-Tuscany. I believe this to be pretty accurate, as I am Ashkenazi Jewish with one (maternal) great-grandmother said to be Sephardic.
MyOrigins shows 98% Jewish Diaspora and 2% European Northlands, which I see as not necessarily a discrepancy so much as a loss of precision, as apparently there is not yet much Sephardic representation in the underlying database.
However, I've been trying to find out whether the presence of the Basque-Spain etc. subgroup in my PF results can still be considered a reliable finding (even if the percentages may be off). I haven't been able to get a definite answer from Bennett, although he has been very responsive. Any insight would be welcome - thanks!

[Barreldriver]

I think Dr. McDonald explained it quite well on the first page of this thread. When I first got my results I was quite pissed until I looked more into it and they do make sense.

I am inclined to think that the Blind Men and An Elephant parable is relevant here. In a given space there is a group of blind men, and there is an elephant. The men touch the elephant so to learn something about it. Each man feels a different part, but only one part, such as a leg or the trunk. One man who is feeling the leg says “this thing must be a sturdy tree”. One man who is feeling the trunk says “this thing must be a mighty worm”. Each other man then goes on to share the content of their subjective experience, and the account of one is not always consistent with the account of another even though they are experiencing the same object.

In the case of a deficient account of my genetic admixture, I would expect that such an account would detrimentally skew my “ancestral average”. I expected my location to be somewhere in between the British, Irish, Germans and French and voila it was.

Final Result: 51.326753, 3.471273. Oostburg, Sluis, Zeeland, The Netherlands. This final spot is in between the British, Irish, Germans, and French so my expectation has been satisfied. Actual ancestral locations are well within 600 miles of here.

The result was obtained from considering myOrigins, Ancestry Composition, Dr. McDonald's BGA, UCLA SPA, Dodecad V3 Gedmatch Oracle-4, and Eurogenes K13 & K15v2 Gedmatch Oracle-4.

The results from each utility were given their own section, and I substituted a given admixture component with a geographic coordinate. The coordinate was a population centroid. The percentage of a given component was recycled and was used as the weight on a line between centroids. With that I could find a weighted geographic midpoint. Finally I found a geographic midpoint between the midpoints found for each section, non-weighted since the points are from sets of results that are “equally blind” (just as the men who were trying to experience the elephant were).

[Barreldriver]

Trying to find a better way to word reasons for why results would vary

Differences in Population Selection can cause different results between utilities -

Most country-level populations overlap to some degree (23andMe 2014), e.g. within Europe, the populations of Western & Central European countries would overlap to some degree. One utility might be dependent upon a method that has slightly different lines drawn between overlapping populations. So, results would may vary depending upon where those lines are drawn.

Differences in Testing & Validation can cause different results between utilities -

Utilities might vary in terms of recall and precision values. A case of high precision and low recall precedes a conservative report of ancestry; A case of low precision and high recall precedes a liberal report of ancestry (23andMe 2014). E.g. In the 23andMe reference set the British & Irish category has a precision value of 0.90 but a recall value of 0.39 (23andMe 2014) as such Ancestry Composition is conservative with reporting British & Irish ancestry; but another utility could have a lower precision value and a higher recall value for their version of a British & Irish category and as such they would over report ancestry for that category.

References

23andMe Inc. 2014. Ancestry Composition: 23andMe’s State-of-the-Art Geographic Ancestry Analysis. Updated 29 May, 2014. URL: https://www.23andme.com/ancestry_composition_guide/


I would also like to reemphasize Dr. McDonald's point from post #2 in this thread:

"Now if the average spot on the map of a purely
European/Jewish/Armenian(etc.)/Turkish person differs by more than 1000 miles between programs, then you have a really bad and exceptional case. But 600 mile error is common and to be expected."

I have not received a set of results yet where there's an "error" of more than 600 miles. So I think that I've had a really good experience with regard to atDNA admixture testing.

It's also a fancy coincidence that all of the spots on the map that I have gotten, and the average spot between those average spots, have actual ancestral locations within 600 miles of them.

[esev974]

Hello, i am an extremely admixed person. I am about 60% euro on MO with about 80% of it french. Yet MO does not show anything close to France. It mainly shows british isles and north mediteranean with strong spot on Spain an Portugal. How is that possible? Yet it matches every gedmatch result I get for europe as well as 23andme.
Also i show 7% jewish diaspora and 4% mid east. Again it matches gedmatch but not 23andme (could be due to the time span which is 500 years for 23andme). Now with 7% on MO and gedmatch how come i do not have 1 single declared or undeclared jewish cousin?
Strange.

[suttonwho]

I thought I would update anyone interested in my dad's results which are just in.

On paper he is half Irish and half Belgian (mostly Flemish). MyOrigins gave these results:

56% Scandinavian
15% Southern Europe
14% British Isles
11% Eastern Europe
5% Asia Minor

My results were:

67% Western and Central Europe
32% Scandinavian
1% Eastern European