Modeling Polish ancestry

[rothaer]

^^^
So as you can see Emma has 15% of Polish admixture and you have 1% of Jewish admixture.
So where does your idea come from that they wrongly assigned Polish admixture as Jewish?

If Emma had 100% Jewish results then your interpretation could be correct. But it isn't the case.

But consider, Tomenable, that building population theories based on 1% is very risky. See what Ancestry honestly says at 1%:

"Your ethnicity estimate is 1%, but it can range from 0-2%"

P.S. In my family tree dating back to the 17th and 18th centuries, I do not have a single Jew among my ancestors. I wish I had, but I don't. As Tolkien wrote in a letter: "I regret that I appear to have no ancestors of that gifted people."

All these companies can wrongly assign a segment to Jewish because its common in modern Jews from their reference data base. It doesn't mean this segment was originally Ashkenazi Jewish 300 or 400 years ago. And you can't properly distinguish how you got it, if Jews have it, unless you can create a cluster of matches which leads back to a specific ancestor to which you can assign that segment(s).

Anyway, another issue is that AncestryDNA doesn't decimal numbers. This means that some people with 1 percent have in fact less, others more than 1 percent in the estimate.

It would be a desiderate to see how homogenous or heterogenous the ancestry of what is AJ is. An on/off button for the AJ component in the Ancestry Composition at 23andMe would be ideal. And technically no big deal.

[rothaer]

(...)
Target: Tomenable
Distance: 2.3784% / 0.02378406
56.4 Proto_Slavic
28.2 Germanic
15.4 Celtic-like

And with my Living DNA raw data:

Target: TomenableLD
Distance: 3.0943% / 0.03094295
51.8 Proto_Slavic
34.4 Germanic
13.8 Celtic-like

I will probably merge all of my raw data files and order coordinates for a merged file.
(...)

It's dissatisfactory with so big differences. But I'm pretty sure that the old FTDNA raw data provides more evaluated SNPs when the G25 coords are calulates as essentially all calculators are based on having the Illumina OmniExpress SNP selection. LivingDNA never had that and started with Illumina GSA chip SNP selection data and later switched to an Affymetrix sequencing chip that has a SNP selection that is a compromise between OmniExpress and GSA SNP selections.

This said, a merged file will very likely yield results that are very close to what you got based on your your old FTDNA raw data.

[Tomenable]

AJ admixture among my 23andMe relatives who are ethnically Polish and with at least 3/4 reported grandparents from Poland:

1 - 0.0%
2 - 0.0%
3 - 2.3%
4 - 1.0%
5 - 5.3%
6 - 0.1%
7 - 1.9%
8 - 0.2%
9 - 0.0%
10 - 0.0%
11 - 0.2%
12 - 0.3%
13 - 4.1%
14 - 0.0%
15 - 0.0%
16 - 0.0%
17 - 0.3%
18 - 0.4%
19 - 0.4%
20 - 0.0%
21 - 0.0%
22 - 0.0%
23 - 0.0%
24 - 0.2%
25 - 0.5%
26 - 3.6%
27 - 0.4%
28 - 0.2%

So out of 28 such relatives 17 have this admixture (ranging from 0.1% to 5.3%) and 11 don't have it. And the mean is 0.8% AJ.

^^^
I can now add a 29th Polish relative to this list (as he just accepted my invitation to share ancestry composition results):

29 - 0.8%

So in total out of 29 relatives, 18 have AJ admixture (62% of the sample size) and the mean is still the same, 0.8% AJ.

Note that Galadhorn and Artek (Artmar) have 0.9% and 1.1% so both have around the same amount as a typical Pole.

[Riverman]

It would be a desiderate to see how homogenous or heterogenous the ancestry of what is AJ is. An on/off button for the AJ component in the Ancestry Composition at 23andMe would be ideal. And technically no big deal.

Well, that's an issue not just for Ashkenazi Jewish, but what I suggested in the past, a proper ancient and prehistorical DNA assignment, like it was tried by FTDNA in a rather unsuccessful and outdated way, since they don't even have a proper steppe component in their calculator.
But in the end all modern populations could split up that way and for that we need an Antiquity, Bronze Age, Eneolithic and Mesolithic calculator - basically. I'm pretty sure if the companies exist and develop long enough, sooner or later they will offer such tools.

All modern people descend from various older ethnicities, not just Ashkenazi Jews.

[Pylsteen]

There is also the following; we know from the medieval jewish samples from Erfurt that one subgroup had significant (up to 40%) Slavic-like admixture (in today's Ashkenazim ca. 10-15%), IMO most likely West-Slavic. That means that there must also be some medieval Slavic segments in the modern Ashkenazi reference samples.

[rothaer]

It would be a desiderate to see how homogenous or heterogenous the ancestry of what is AJ is. An on/off button for the AJ component in the Ancestry Composition at 23andMe would be ideal. And technically no big deal.

Well, that's an issue not just for Ashkenazi Jewish, but what I suggested in the past, a proper ancient and prehistorical DNA assignment, like it was tried by FTDNA in a rather unsuccessful and outdated way, since they don't even have a proper steppe component in their calculator.
But in the end all modern populations could split up that way and for that we need an Antiquity, Bronze Age, Eneolithic and Mesolithic calculator - basically. I'm pretty sure if the companies exist and develop long enough, sooner or later they will offer such tools.

All modern people descend from various older ethnicities, not just Ashkenazi Jews.

I fully agree on your take that ideally a time-related ancestry composition for a number of different times could and should be provided.

I just want to point out that nevertheless the current AJ component to my perception is not in line with the system of display, where is displayed NW European, E European etc., i. e. AJ is on a more specific level then the other components. If you have AJ you can also display Mennonites, Arcadians and maybe even traditional European nobility due to their actual separation in procreation and thus specific genetic signatures. But if you assign someone to Mennonite, you more fog his ancestry in terms of biogeographic  ancestry than you do enlight it. I don't oppose displaying AJ ancestry. This is a doubtless interesting topic. But I consider it not to be on the level of the other displayed biogeographic components.

[Riverman]

It would be a desiderate to see how homogenous or heterogenous the ancestry of what is AJ is. An on/off button for the AJ component in the Ancestry Composition at 23andMe would be ideal. And technically no big deal.

Well, that's an issue not just for Ashkenazi Jewish, but what I suggested in the past, a proper ancient and prehistorical DNA assignment, like it was tried by FTDNA in a rather unsuccessful and outdated way, since they don't even have a proper steppe component in their calculator.
But in the end all modern populations could split up that way and for that we need an Antiquity, Bronze Age, Eneolithic and Mesolithic calculator - basically. I'm pretty sure if the companies exist and develop long enough, sooner or later they will offer such tools.

All modern people descend from various older ethnicities, not just Ashkenazi Jews.

I fully agree on your take that ideally a time-related ancestry composition for a number of different times could and should be provided.

I just want to point out that nevertheless the current AJ component to my perception is not in line with the system of display, where is displayed NW European, E European etc., i. e. AJ is on a more specific level then the other components. If you have AJ you can also display Mennonites, Arcadians and maybe even traditional European nobility due to their actual separation in procreation and thus specific genetic signatures. But if you assign someone to Mennonite, you more fog his ancestry in terms of biogeographic  ancestry than you do enlight it. I'm don't oppose displaying AJ ancestry. This is a doubtless interesting topic. But I consider it not to be on the level of the other displayed biogeographic components.

I'm not sure I agree, because clearly they try to do it whenever possible. Think about Finns, which being recognised as well and could be split in a similar way. Also the German component includes already a bit of Southern and Eastern ancestry IMHO.
But I get what you mean.

Looking at LivingDNA, which doesn't use AJ, it becomes more confusing for people with AJ ancestry I'd say and kind of misses the point for recent ethnicities.

[rothaer]

(...)
Looking at LivingDNA, which doesn't use AJ, it becomes more confusing for people with AJ ancestry I'd say and kind of misses the point for recent ethnicities.

Yes, LivingDNA does not have an AJ component so that is exactly the suitable comparison. Unfortunatley, I've no no full AJ result from there in mind. I'd even like to see a number of them.
And it's true that the "recent" origin is then not displayed but kind of a pre-AJ one which will pre-date the referred to time that is intended to be displayd for others. Iirc it's aimed at just abt. 10 generations ago, i. e. at abt. 1700 AD.

Do you have any overview on full AJ LivingDNA results? Do they notably differ among each other?

[Riverman]

(...)
Looking at LivingDNA, which doesn't use AJ, it becomes more confusing for people with AJ ancestry I'd say and kind of misses the point for recent ethnicities.

Yes, LivingDNA does not have an AJ component so that is exactly the suitable comparison. Unfortunatley, I've no no full AJ result from there in mind. I'd even like to see a number of them.
And it's true that the "recent" origin is then not displayed but kind of a pre-AJ one which will pre-date the referred to time that is intended to be displayd for others. Iirc it's aimed at just abt. 10 generations ago, i. e. at abt. 1700 AD.

Do you have any overview on full AJ LivingDNA results? Do they notably differ among each other?

You can't even tell with certainty whether a match for my kits is fully Ashkenazi or not on LDNA, it remains a bit of guessing because of this kind of estimate and no trees. Also, most matches don't share their ethnicity results.

But one who might be at least mostly AJ has these results:

South Italy
53.6%
Cyprus
16.4%
Arabia
10.1%
Northeast Europe
8.2%
Pannonia
5.4%
East Balkans
2.6%
Levant
2.1%
East Iberia
1.7%

Another one:

South Caucasus
34.5%
West Iberia
21.9%
Northeast Europe
18%
South Italy
14.1%
Arabia
7.8%
Levant
2.4%
North Africa
1.2%

The most consistent pattern is primarily South Italy and Arabia-Levant, secondarily North East Europe (Baltoslavic). The rest behaves more stochastic I guess, like usual on LDNA or MH etc.

[rothaer]

a sample from Viminacium (Serbia) from abt. 146 AD. I consider that assignment arbitrary and not legit. First, these samples are from notably before Slavs at all showed up in history and second, when Slavs showed up in history, they were genetically not like that. And third, you have no clue where these samples from Sicily and Serbia hail from.

Some Viminacium samples (including R9673) share IBD segments with later Medieval Slavs according to upcoming Vyazov et al. study, so maybe they represent some early migrants (dead ends) from the Slavic homeland (i.e. from post-Zarubintsy horizon).

-------------

EAA 2024 abstract...

Geography and chronology of Slavic dispersal in Central and Eastern Europe according to archaeogenetic data

Limited suitability of PCA, ADMIXTURE, and qpAdm methods of genetic analysis relying on allele frequencies for distinguishing many populations that were important historical actors in the Roman and Migration Period Europe can be overcome by the implementation of an approach based on autosomal haplotypes, “identity-by-descent” (IBD). Imputation of missing genotypes, phasing and IBD inference were performed for ancient Eurasians using the ancIBD method (Ringbauer et al. 2023). IBD links for subsets of these individuals are represented as graphs, visualized with a force-directed layout algorithm, and most clusters that were inferred in these graphs with the Leiden algorithm are in remarkable agreement with archaeological evidence, reflecting communities of distantly related individuals. Based on this approach, we identified a distinct population group that existed in Central and East Europe from the late 2nd century CE to the Medieval period. This group played a significant role in population processes and, from the 6th century onwards, was identified in written sources as the Slavs. Considering recent linguistic and archaeological advancements, it is plausible that this group included Slavic speakers from its early formation. Considering the results of other analyses based on allele frequencies (PCA, ADMIXTURE), a hypothesis for the origin and history of this population can be proposed: it was formed in close vicinity of the Eastern Baltic region and dispersed over Central, Eastern, and Central-South Europe in several waves of migrations. The earliest representatives of this population group were found along the wide arc from the Middle Danube to the Middle Volga that limits from the south the poorly sequenced area with the domination of the cremation ritual, in the context dated to the late 2nd-early 3rd century CE. The latest individuals of this IBD cluster were buried in the Medieval Slavic-associated context.

As I happen to deal a lot with DNA genealogy and in this context with common IBD segments ("matches") and have encountered incredibly much nonsense (false matches) I'd like to know who shares how much cM whith whom and how many tested - NOT imputed!!! - SNPs are shared in that segment. Is there phased DNA data used, onde-sidedly or both-sidedly?

The next thing is if there is a very small real segment determined between individuals that are separetd by centuries, what does it imply? Does it imply any relationship exceeding banality for people coming from Central and East Central Europe? That people are all very distantly related and also Slavs do have forebears that are inteconnected this way is selfexplanatory.

If an individual from 200 CE shares 3 cM with a later medieval Slav it doesn't even indicate that this is any kind of main predecessor of later Slavs. There are findings thinkable that are indicating that but this is totally dependent in the actual figures.

[Tomenable]

Interesting research based on mtDNA haplogroups:

https://www.researchgate.net/publication...sh_origins

[Tomenable]

(...) I'm not sure why you include the two depcited expressed Wielbark outliers in the Wielbark mean. Although these individuals doubtlessly existed and thus were present, they have obviously a notably different ancestry. (...)

I forgot to exclude the two Wielbark outliers from the average. Here is new average with two outliers excluded:

Code:

Wielbark(n=49),0.127807,0.129698,0.070221,0.060902,0.042369,0.022129,0.006719,0.008467,0.002175,-0.006029,-0.004341,0.002857,-0.00604,-0.005415,0.016633,0.007888,-0.003874,0.0025,0.004815,0.005942,0.007792,0.003636,-0.000473,0.011979,-0.001481

[Tomenable]

Looks like we have 3 new Early Slavic samples which can be used in modelling:

Code:

RKF188,0.141141,0.126941,0.085984,0.083334,0.045855,0.034025,0.011516,0.016384,0.003068,-0.033349,0.000162,-0.008842,0.029137,0.040599,-0.014522,-0.004641,-0.01356,-0.007981,-0.003771,-0.005628,-0.00574,0.001731,0.000493,-0.009037,-0.002275
RKO013,0.12862,0.117801,0.080327,0.076228,0.050471,0.031794,0.011045,0.015692,-0.004909,-0.032074,-0.006983,-0.012889,0.022299,0.027937,-0.01045,-0.003713,0.007171,-0.003167,0.004022,-0.005127,-0.003494,-0.012365,0.010476,-0.017352,-0.000599
RKF106,0.12862,0.128972,0.07731,0.076551,0.040315,0.026774,0.012926,0.008307,0.001432,-0.030798,-0.003735,-0.01154,0.01665,0.022432,-0.016829,0.011005,0.025555,0.000887,0.008925,0.008754,-0.008984,-0.014344,0.000123,-0.005181,0.003592

Average of these three:

Code:

Early_Slavic(n=3),0.132794,0.124571,0.081207,0.078704,0.045547,0.030864,0.011829,0.013461,-0.000136,-0.032074,-0.003519,-0.01109,0.022695,0.030323,-0.013934,0.000884,0.006389,-0.00342,0.003059,-0.000667,-0.006073,-0.008326,0.003697,-0.010523,0.000239

The distance is MUCH better now:

Target: Polish_Medieval(n=62)
Distance: 1.1869% / 0.01186853
76.6 Early_Slavic(n=3)
13.4 Wielbark(n=49)
10.0 Przeworsk_PCA0012