Talk:Ashkenazi Jews: Difference between revisions

By /

Tags: Mobile edit Mobile web edit
Tags: Mobile edit Mobile web edit

 

Line 159: Line 159:

::::::The issue seems to me that the authors say, in multiple places, that there are uncertainties and caveats that point to 19% being somewhat of a misestimate one way or another.

::::::The issue seems to me that the authors say, in multiple places, that there are uncertainties and caveats that point to 19% being somewhat of a misestimate one way or another.

:::::::So aconcern is that the description of that figure as “most plausible” or “average” in the table could be a bit misleading because, though there is a quote in the ref, this is much less likely to be read, by the general reader, than the table itself or paragraphs. It thus presents a figure that, in the paper is hedged by caveats, in a stand-alone manner.[[User:Skllagyook|Skllagyook]] ([[User talk:Skllagyook|talk]]) 18:48, 27 November 2025 (UTC)

::::::So aconcern is that the description of that figure as “most plausible” or “average” in the table could be a bit misleading because, though there is a quote in the ref, this is much less likely to be read, by the general reader, than the table itself or paragraphs. It thus presents a figure that, in the paper is hedged by caveats, in a stand-alone manner.[[User:Skllagyook|Skllagyook]] ([[User talk:Skllagyook|talk]]) 18:48, 27 November 2025 (UTC)

Hello,
Spain is a political state, that did not existe in the times referred in these 3 sentences. So, it is an absolute anachronism. It is simply wrong.
Thanks for your attention
Henrique Soares Oliveira
« onforming to the custom of designating areas of Jewish settlement with biblical names, Spain was denominated Sefarad (Obadiah 20),»

« while some Ashkenazi Jews joined Sephardic Jewry in Spain.»

«  the centers of Jewish religious authority were in the Islamic world, at Baghdad and in Islamic Spain. » 78.137.210.80 (talk) 01:12, 3 May 2025 (UTC)[reply]

Hello,

Spain is a political state, that did not existe in the times referred in these 3 sentences. So, it is an absolute anachronism. It is simply wrong.

Thanks for your attention
Henrique Soares Oliveira

« onforming to the custom of designating areas of Jewish settlement with biblical names, Spain was denominated Sefarad (Obadiah 20),»

« while some Ashkenazi Jews joined Sephardic Jewry in Spain.»

«  the centers of Jewish religious authority were in the Islamic world, at Baghdad and in Islamic Spain. » Henrique Soares Oliveira (talk) 01:13, 3 May 2025 (UTC)[reply]

While this is correct, it is common on Wikipedia articles to use present day place or region names when it is less confusing, and to follow what sources do. Many sources refer to Sephardic Jews coming from Spain or Portugal even though at the time, in Moorish Spain the places were more likely to be called al-Andalus by the inhabitants or Sepharad, and when they were Christian they were more likely to refer to specific locations like Castile or Aragon since the present-day political units didn’t exist. Still, this probably exists throughout historical sources and Wikipedia. There was no Germany or France but they still refer to the Rhineland or Provence or Narbonne as being part of Germany or France, or parts of then-Russia as Ukraine or Poland or Romania etc. So anyway, you are right but I am not sure we really need to change it because it is just understood that these modern-day place names can be used regionally anachronistically. Andre🚐 03:08, 3 August 2025 (UTC)[reply]

Regarding the Lerga-Jaso et al. source recently included in a table of admixture estimates, it seems perhaps a bit UNDUE, at least in its current position for a few reasons.
It’s very new, and doesn’t seem to largely agree with previous studies as the authors state, including the studies they reference in that quote. In the quote they reference Costa et al. Waldman et al., and Xue et al. Costa et al. does not concern autosomal DNA at all and gives no estimate for it, but studies only Ashkenazi maternal (not paternal) lineages, which it argues are mostly European.(Some more recent studies have agreed and some have challenged that.)
Waldman gives a range for Ashkenazi Middle Eastern Autosomal admixture of 19-43% (stating that the exact ammount is not certain). The former is estimated if Southern Italian references are used, and the latter if Northern Italian ones are used. But they suggest that (particularly in the model with Southern Italian references), Levantine/Middle Eastern admixture in Ashkenazi could be underestimated (due to Middle Eastern admixture in Italians, this being higher in Southern Italy) and genetic admixture in modern Middle Easterners within theast 2,000 years.

From Waldman:
“We caution that the specific identity of the source populations that we inferred, as well as the admixture proportions, should not be considered precise. This is due to the multiple Southern European populations that fit the EAJ data, as well as our reliance on modern populations as a proxy of the true ancestral sources. The levels of Middle Eastern ancestry in Italy were historically variable (Aneli et al., 2021; Antonio et al., 2019; De Angelis et al., 2021; Posth et al., 2021; Raveane et al., 2019), and Middle Eastern populations have also experienced demographic changes in the past two millennia, particularly African admixture (Moorjani et al., 2011)…Under the extensive set of models we studied, the ME ancestry in EAJ is estimated in the range 19%–43% and the Mediterranean European ancestry in the range 37%–65%. However, the true ancestry proportions could be higher or lower than implied by these ranges.”

The third study referenced, Xue et al., estimates around 40% Middle Eastern ancestry in Ashkenazi Jews.
None of these sources quite agree with the estimate of Lerga-Jaso.

Also, the admixture tool Lerga-Jaso et al. use is unconventional, and apparently, according to them, as mentioned by USER:Teflawn is accurate for more recent admixture, while the majority of admixture in the Ashkenazi is much older.
I propose removing Lerga-Jaso from the table, which seems to give it UNDUE prominence and, if included, perhaps instead incorporating it into the autosomal section as follows:

“Lerga-Jaso et al. (2025) mapped Ashenazi Jews as 68% Italian, 23.4% Middle Eastern, 7.2% Balkan/Greek, and 1.7% Eastern European. Authors acknowledge that their admixture tool, Orchestra is best suited for admixture within ~12 generations. Where Ashkenazi admixture is older (at 30-40 generations). [1]|

Some might argue that including this final detail about the admixture tool used is SYNTH. I am uncertain. But it does seem to be relevant and useful information that helps a general reader more accurately understand Skllagyook (talk) 22:43, 19 November 2025 (UTC)[reply]

I agree that if Lerga-Jaso is to be retained in the article, the method/tool used should be mentioned along with the authors’ own stated limitations. The “Where Ashkenazi admixture is older (at 30–40 generations)” could be reworked to instead reference previous studies that provide admixture time-frame estimates for Ashkenazim, so as to avoid potential WP:SYNTH issues.
Ultimately, I do not think this paper should be featured on this page, which is intended to be a holistic overview of all things Ashkenazim. If included at all on Wikipedia, it would be better located on Genetic studies of Jews, where discussion of methodological limitations would be more appropriate, given that the current article is intended as a general overview rather than a detailed genetics-focused discussion. The genetics section is beginning to grow unruly, especially in comparison to the pages on Sephardic and Mizrahi Jews and their respective genetics sections.Teflawn (talk) 23:58, 19 November 2025 (UTC)[reply]
I’ll try to make my reply as organized and as informative as possible
Xue, 2017
Xue study gave two ancestry proportions estimates using two different tools, one using GLOBETROTTER and the other using RFMix, the former’s result fully align with lerga-jaso result estimating their ancestry as 30% middle eastern and 70% European. Whereas while RFMix do significantly differs (47% middle eastern and 53% european), we should highlight two major things about its result:

  • Xue writes: “Our simulations suggested that the accuracy of LAI for an EU-ME admixed population is only around ≈70%, much lower than the near-perfect accuracy observed for cross-continental admixture”
  • in lerga-jaso’s study, orchestra largely outperformed RFMix in every metric on benchmark.
Waldman, 2022
even though waldman cautious specifically about the interpretation of his qpAdm modeling results and highlights some reservations. His qpAdm modeling results nonetheless still agree with both Xue 2017 GLOBETROTTER results and Lerga-Jaso 2025 orchestra results. Nonetheless i do not disagree that his reservations and comments on his qpAdm modeling results should be added in a note tag
Lerga-Jaso, 2025
I’ll start with saying that i completely reject the premise of using one part of the study to refute another part of the same study, thats the exact definition of WP:SYNTH. So this whole talk about this point whether your criticism or my response to below is redundant.
regarding this specific quote if you expand it further, the authors say: “Orchestra is relatively accurate at least up to around 12 generations of admixture and can detect trace ancestries from further back in time. This means it is suited to reconstruct events of relatively recent admixture, within Modern and potentially Medieval history. However, for reconstruction of Ancient history, other non-window based LAI models would have a clear advantage”.
recalling from Xue, 2017. AJs inferred admixture dates was also in the middle ages. Also it is worthy noting that RFMix used by Xue is also a window based LAI model just like orchestra.
summary
three different autosomal ancestry proportions genetic tools from three different studies all agree over roughly the same proportions estimate (20-30% middle eastern, 70-80% European), this is actually the largest consensus among all tools that estimated autosomal ancestry proportions of AJs. So i largely disagree with the WP:UNDUE claim of the Nature Communications study in every aspect, and even if we assume it was undue that means it shouldnt be in the lede or given a large portion of the genetics section, not that it should be entirely removed, which is not the case either ways as the study neither has content in the lede nor given a significant portion from the genetics section —- literally just a single row in a table. Chafique (talk) 12:52, 20 November 2025 (UTC)[reply]

@Skllagyook, i believe your recent edits contain some errors

In your edit here, you removed the estimated average of all accepted qpadm models saying “Adding additional quote. 19% is average only if Southern Italian is used. Waldman and Xue both find Levantine admixture.”

thats not true if you check the table that the author is referring to in the quote (Table S3) you’ll find models with north italian sources as well along side dozens of other models, the 19% is the average portion of all the plausible/accepted models, not just northern or southern italians.

In Xue’s estimate you also added 47% (Levantine) whereas the author writes that this 47% is Middle Eastern not just Levantine: “Running RFMix on the AJ genomes with our EU and ME reference panels and summing up the lengths of all tracts assigned to each ancestry, the genome-wide ancestry was ≈53% EU and ≈47% ME”, the ME source is composed of “The Middle-Eastern genomes were divided into three regions: Levant, Southern Middle-East, and Druze.”, for accuracy i’ll replace levantine with mostly levantine as the authors note Chafique (talk) 16:29, 20 November 2025 (UTC)[reply]

You wrote:
“19% is the average portion of all the plausible/accepted models, not just northern or southern italians.”
This is a bit confusing as they explicitly state that 43% ME is the result they get with Northern Italian (and about 35% if I recall using Greek) and Eastern Southern European reference.
You wrote:
“this 47% is Middle Eastern not just Levantine..”
The authors state that the Levantine references (Palestinian, Jordanian, and Syrian) were the best fit than the non-Levantine (Southern Middle East: Egyptian, Bedouin, Saudi). (The Druze are also a largely Levantine population, but are for some reason categorized separately.) They say:
“The best match to the AJ data was obtained (in both cases) when the Levant ancestry was almost entirely exclusive (45% out of the total 50% ME ancestry; the magnitude of the minor components was close to zero also when we simulated 50% Southern EU ancestry). This result supports a predominantly Levantine origin for the ME ancestry in AJ, and justifies using the Levantine genomes for the ME ancestry in our simulations.”
So you are technically right that the estimate or the ME is mostly Levantine, but it is apparently nearly exclusively so.

Waldman also specifies Levantine admixture, and uses Levantine references. You seem to have removed that.Skllagyook (talk) 17:45, 20 November 2025 (UTC)[reply]

This is a bit confusing as they explicitly state that 43% ME is the result they get with Northern Italian (and about 35% if I recall using Greek) and Eastern Southern European reference.
Yes because thats just a single model and the only model where northern italian source was accepted/plausible, the authors also write (if you check the full pdf):“Overall, a South-Italian source was more plausible than a North-Italian or a Greek source, and Lebanese, Saudi, Syrian, Jordanian, and Bedouin B were the more likely Middle Eastern sources.”
so not just the 19% ME figure is the mean over all the plausible models, the southern italian models is also more plausible than northern one per the study.
so can you please restore the average (and most plausible) figure you removed back while also keeping the range ?
So you are technically right that the estimate or the ME is mostly Levantine, but it is apparently nearly exclusively so
yeah i agree thats why i just added “mostly levantine” on my own anyway Chafique (talk) 17:55, 20 November 2025 (UTC)[reply]
Hello. My apologies for taking so long to respond. I was busy with some things.
Regarding Waldman, you wrote:
“Yes because thats just a single model and the only model where northern italian source was accepted/plausible…so can you please restore the average (and most plausible) figure you removed back while also keeping the range ?”
Yes. But, as mentioned, the models that find 19% are the ones that use Southern Italian as a reference rather than the one that uses Northern Italian (or anything else). I think that should be clarified in the table, as it is currently. Describing 19% simply as the average, or more plausible, in the table, seems likely to be a bit misleading, as it would look to suggest to the general reader that it is the average across models using different references and reference combinations when that is not the case.
This seems particularly so since the authors, despite perhaps leaning toward Southern Italian admixture as more likely, make a point, more than once in the study, to express caveats and reservations about the 19% figure in that model, about the possibility of it being misleading/unreflective of the true amount of ME ancestry in Ashenazi, given that Middle Eastern admixture exists in modern Southern Italians (and other admixture exists in some modern Middle Easterners). The second being the following, where the authors explicitly caution against interpreting the models as “quantifying direct contributions of specific populations”:
“We caution against interpreting the qpAdm models for the ancestral sources of EAJ as quantifying direct contributions of specific populations to the early AJ gene pool. This is because (1) the wide range of inferred ancestry proportions across models; (2) the historical fluctuations in ME ancestry in Italy…. the large space of models not explored here. Instead, the results reflect genetic links between the ancestral sources of AJ and modern populations, who may or may not have genetic continuity with their ancient antecedents.”
If more about this is added, it might be better to include it (re the plausible models, their context, etc.) instead in the Waldman paragraph (including the range and the authors’ caveats) while keeping the table as it is.
It appears that one problem is, as pointed out by Waldman et al., the use of modern populations as references (also an issue in Xue, Lerga-Jaso, and others), when modern populations may not be the same as their ancient antecedents (despite some substantial continuity with them). Hopefully more studies with (historically appropriate) ancient references will be done in the future.Skllagyook (talk) 16:07, 23 November 2025 (UTC)[reply]

@Skllagyook i already said that i dont mind keeping the range and adding a note tag regarding author’s reservations over the interpretation of the results, note that we are talking here about the interpretation of the result, not about the qpAdm result itself —- which is the table’s point to show—-, i find no merits in hiding or omitting the fact that qpAdm most plausible models / results is that of the south italian models, which is acknowledged by the author himself: “Overall, a South-Italian source was more plausible than a North-Italian or a Greek source, and Lebanese, Saudi, Syrian, Jordanian, and Bedouin B were the more likely Middle Eastern sources.”, what is misleading is portraying a single passing north italian model as being as plausible as 10+ passing south italian models which qpAdm (and the author) recognize as being more plausible.
Regarding the interpretation —-which i want to reaffirm that it is another story from the result itself which is our concern —- i also want to add that while the author expresses cautious over the interpretation of the qpAdm results and offers multiple possible interpretations for this result, he also explicitly states (again in his full PDF paper) that “This may also be interpreted to imply that present-day AJ derive only a small proportion of their ancestry from ancient Judaeans; and if so, most AJ ancestry would owe its origin to European converts … we believe all types of scenarios are plausible.“. You cant pick one and throw the other.
And for the third time i want to affirm, there is a difference between the genetic analysis method / tool result (which is our main concern here), and their interpretations. For example, two geneticists may view the same PCA result but interpret it very differently (Elhaik, 2022). Our main concern here is mainly about portraying the ancestry proportions analysis and tools results, just like PCAs.the issue we are having right now is you wanting to omit part of a result, that is south italian modeling is more genetically plausible, i find this as meritless as omitting / removing the Sicilian and south Italian clusters from (lazaridis, 2014) PCA because ashkenazi jews cluster with them Chafique (talk) 10:45, 24 November 2025 (UTC)[reply]

Where can I find the full pdf? Are you referring to the supplimentary materials.?Skllagyook (talk) 13:11, 24 November 2025 (UTC)[reply]

Yes Chafique (talk) 14:59, 24 November 2025 (UTC)[reply]
Chafique I found it. This is the full quote (from the first pdf, pages 46-47):
“Models with a South-Italian source were more frequently favored by qpAdm (Table S3) and have a plausible historical basis (above). However, these models suggested that only about 20% of EAJ ancestry derived from Middle Eastern sources. This is less than previous estimates based on modern SNP and sequencing data… This may also be interpreted to imply that present-day AJ derive only a small proportion of their ancestry from ancient Judaeans; and if so, most AJ ancestry would owe its origin to European converts. While this is one possible explanation, modern Italians themselves have had much higher proportions of ME admixture since at least European Imperial Roman times [43] and this is especially the case in modern Southern Italy [108]. Thus, an alternative explanation for these observations is that the true ME proportion in AJ is higher than in our fitting model, and that the actual contribution of Italians is not as large as suggested by this analysis. Under this scenario, good qpAdm fits are obtained using South-Italians as sources simply because they are a modern population that harbors a relatively high proportion of ME ancestry. If this alternative explanation is right, the true ME proportion could be higher than in our models, e.g., close to the 30-50% estimates from previous studies [40, 51] or when modeling EAJ using North-Italians.46 In an opposite scenario, AJ may have no ancestry at all from the ancient Levant. This could be the case if an unsampled Italian population (with more Levantine-like admixture than in modern South-Italians) is the source of all the Levantine-like ancestry seen in AJ. At present, we believe all types of scenarios are plausible. Co-analysis of ancient DNA data from the Middle East and the Italian peninsula from Antiquity and the early Medieval period would make it possible to distinguish them. Further complicating the interpretation of the Middle Eastern origins are multiple demographic changes that have also affected the Middle East during the past two millennia. Most notably is African admixture, as documented in multiple populations [109-111], particularly Saudis [112, 113] and Egyptians [114]. Bottlenecks and population structure were identified in Druze, Bedouin, and Lebanese…and temporal changes in ancestry were observed in Syria [117] and Lebanon…”
The authors seem to be taking a very cautious position that is open to all conceivable possibilities. They mention previous estimates of 30-50% Levantine ancestry in Ashkenazi, which is what many autosomal studies seem to have found, and the possibility of them being correct (if Imperial-era ME admixture into Southern Italians is taken into account).
They also mention the possibility of an unsampled Italian population, with even more Levantine and/or Levantine-like admixture than any Italian population we know of, being the source of AJs. (This would seem somewhat supported and more speculative, since we have no evidence of such a population existing, but perhaps possible in theory.)
Despite the authors’ caution, it might be significant that the first possibility (of ME admixture in Southern Italians giving misleading results) is the one that they choose to include in the main paper.
They look to be stating that even in the scenario of the majority of AJ ancestry being from “European converts”, many of these converts would likely have to have come from a European population rich in their own Levantine/Middle Eastern admixture (in order to fit the data).
In either scenario, the authors seem to be concluding that there woulc be significant Middle Eastern admixture in Ashkenazi (above what their model using South Italy as a reference detects if taken at face value) but that they are still unable to be sure as to when and how this admixture came to be there (whether through substantial direct descent from Judeans/Levantines, from a theoretical population of heavily Middle Eastern-admixed Italians, or some combination of both).Skllagyook
You wrote:
“I already said that i dont mind keeping the range and adding a note tag regarding author’s reservations over the interpretation of the results, note that we are talking here about the interpretation of the result, not about the qpAdm result itself —- which is the table’s point to show—-, i find no merits in hiding or omitting the fact that qpAdm most plausible models / results is that of the south italian models…”
The issue seems to me that the authors say, in multiple places, that there are uncertainties and caveats that point to 19% being somewhat of a misestimate one way or another.
So aconcern is that the description of that figure as “most plausible” or “average” in the table could be a bit misleading because, though there is a quote in the ref, this is much less likely to be read, by the general reader, than the table itself or paragraphs. It thus presents a figure that, in the paper is hedged by caveats, in a stand-alone manner.Skllagyook (talk) 18:48, 27 November 2025 (UTC)[reply]
  1. ^ Lerga-Jaso, Jon; Novković, Biljana; Unnikrishnan, Deepu; Bamunusinghe, Varuna; Hatorangan, Marcelinus R.; Manson, Charlie; Pedersen, Haley; Osama, Alex; Terpolovsky, Andrew; Bohn, Sandra; De Marino, Adriano; Mahmoud, Abdallah A.; Bircan, Karatuğ O.; Khan, Umar; Grabherr, Manfred G. (2025-05-16). “Tracing human genetic histories and natural selection with precise local ancestry inference”. Nature Communications. 16 (1): 4576. Bibcode:2025NatCo..16.4576L. doi:10.1038/s41467-025-59936-3. ISSN 2041-1723.

Must Read

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top