search this blog

Tuesday, August 18, 2020

Housekeeping stuff


I'm about to phase out the use of the Global25 datasheets with modern-day samples. In large part, this move is due to the uncertainty about the populations that these individuals represent and the resulting (often idiotic) discussions here and elsewhere about their usefulness.

This uncertainly exists because many, perhaps most, of these people are classified based on their self identity, which may or may not reflect their genetic origins.

Thus, I'll no longer be updating these datasheets and, from next week, I'll also stop linking to them at this blog (like here). The links will remain live for the next few months, so that users can adjust to the change.

However, modern-day samples sequenced from archeological remains, and thus, as a rule, painstakingly classified by experts based on their burial contexts and genetic characteristics, will continue to be featured in the Global25 datasheets.

In other words, as far as the Global25 is concerned, all of the modern-day samples from the living are out, but all of the modern-day samples from the dead will remain, and indeed I'll be adding more of the latter as they become available.

I'm planning to eventually create several sets of Global25 datasheets based on individuals and populations from different periods, including the modern era. But I'll probably need some help with that.

Also, please note that comment moderation will now be the rule here rather than the exception. And I'll be cracking down hard on trolling, insults and any sort of potentially defamatory material, so no more crazy stuff, or else.

See also...

New rules for comments

106 comments:

Wot said...

Of the current existing modern data, how much can be salvaged based on the criteria you use?

Davidski said...

None, because all of these samples are from the living.

We have to shift the focus to ancient samples, and if users are interested in more recent ancestry, then they should focus on the Iron Age and Medieval periods.

There will be a lot of samples available from the Middle Ages soon, so this won't be a problem.

Ymir said...

If G25 coordinates in the future will be still compatible with modern (not-updated) sheets?

Davidski said...

Yes, the old and new datasheets will all be compatible with each other, but I won't be linking to the current modern datasheets.

CHG Chad said...

David, I guess the Armenian samples played a role to this? Anyway, sounds cool to me and as you mention the IA and MA samples will be million of times better to have and discuss.

Davidski said...

Yes, I saw some complaints about the Armenian samples too.

Let's see what happens when the new Medieval Armenian samples arrive. I bet they won't be representative either to someone on the internet, but I didn't choose them, so not my problem.

Anonymous said...

I’ve always being more interested in the ancient samples anyway and just for the record my moderns were accurate

CrM said...

@Davidski

"There will be a lot of samples available from the Middle Ages soon, so this won't be a problem."
"Let's see what happens when the new Medieval Armenian samples arrive"

Does the "soon" apply to Medieval Armenian samples too?

Wise dragon said...

Respect your decision a 100% but please, please don‘t turn to a second anthrogenica, a forum that is being moderated to death. I really appreciated that on your blog free speech was very much alive. Surely you have to sort out trolls who troll only for the purpose of trolling. With that being said some people are just nutcases or mislead or both is better to let them express their nonsense and educate them by debunking and refuting them. I have seen some folks coming to their senses after their ass was handed to them with arguments. In my opinion what makes a forum really interesting is discourse where people can debate passionately and hard without heavy moderation.

Onur Dincer said...

Are there any ancient, medieval or modern-era Anatolian, Balkan or Caucasus genomes from the dead in the pipeline?

Ariel said...

I honestly thought that the g25 modern spreadsheets had problems and some of the complaints might have been valid, but overall it was a great tool, and the problem probably arose from the fact that we don't have a lot of public data from the living.
Btw, are we expecting any other major paper? Or this year is officially a waste?

Wot said...

Oh that makes sense, the Armenian samples had so little Steppe, what is the deal with them?

CHG Chad said...

@ Wise Dragon

I agree 100%.I started to post here because of the free speech. Anthrogenica is a really good forum but it contains PC policies and the majority of moderators acting like the Police.David you should also passing our comments even if Some of them are maybe a little bit retarted or Boring.There are not many anthroforas you all know it.

@ Wot

The Armenian samples are fine. They all plot between South Caucasus and Mesopotamia.Armenians are still a very west Asian ethnic group despite if their steppe decreased or increased during the periods.The problem with Athenians is their geographical position. Divided between 'eastern and western' Armenia,thought western Armenia is more close to their ancient homeland before Turks and Persians dominate these lands.

Onur Dincer said...

@Wot

Oh that makes sense, the Armenian samples had so little Steppe, what is the deal with them?

The academic Armenian samples from Armenia are normal Armenian samples, there is nothing unusual in their results. Only the academic Armenian samples from Maikop have some problems, there are some outliers with recent Russian admixture among them.

Low steppe is normal for the modern Armenians, all of them have low steppe unless they are recently mixed with non-Armenians.

Michalis Moriopoulos said...

I understand the reasoning, but I'm still disappointed. Comparing ancient to modern peoples is part of what makes this hobby fun in the first place.

Aram said...

I missed that discussion ( was it on Anthrogenica?) about modern Armenians but I also think that Medieval Armenians will be a surprise.

Just imagine that probably as much as 60-70 percent of ancestry of modern Armenians living in modern Republic are recent (100-150) migrants from West. Some from as distant places as Cilicia or Malatya.

Aram said...

This is a good move to collect Medieval data. But I guess for many ethnicities especially for those who don't have state this can be problematic.

Onur Dincer said...

@Davidski

Could you make a one-time exception and upload the recently published modern African genomes in this paper to your Global25 modern populations?

andrew said...

It's your blog and your call in all things here, but while you are housekeeping anyway, I'd like to make a suggestion for the next time you decide to make a blog appearance facelift.

Could I put in one vote for a larger font size in your body text?

I made this change at my blogs several months ago and it really does make for a more comfortable blog reading experience.

Samuel Andrews said...

You are removing living DNA samples because DNA studies do a poor job confirming the person is of the ancestry they claim to be? I don't understand. Everything in G25 PCA indicates DNA samples from living people are of ethnicity they claim to be.

zardos said...

@David: I would like you to reconsider your decision, because I think the modern samples are very useful and there being many samples included which are rare in what commercial testing companies and other datasets offer. I also think that the regional variation could be expanded and the silly complaints by people which are not satisfied with this or that should not take away such helpful tools with a lot of potential.

Even more, the ancient samples are ideal to check and compare with the modern ones. I hoped for more regional samples and probably more academic ones, but not that this area being abandoned altogether.

But of course, everybody just has to accept your decision and I can understand it because especially some national communities are particularly annoying if its about their "perfect sampling" ideas and personal bias. Or those which don't understand that many samples, even academic ones, are not always ideal. I guess you got too annoyed by the personal attacks by a dozen or so individuals which always complain.
Still I think its a pity that those few did discourage you from rather expanding than abandoning the modern data sets.

RRN said...

I am very sorry to read you, and I regret that in the end the criticism campaign from dishonest people influenced your decision.

I understand your decision, although I would have preferred you to have kept the modern references, updating them over time.

I suppose you know that many, when the results of the ancient samples do not like them, they will criticize them inthe same way.

Davidski said...

@Wot

Oh that makes sense, the Armenian samples had so little Steppe, what is the deal with them?

Some of the new Medieval Armenians have even less steppe ancestry.

CHG Chad said...

@ Aram

I am reading your comments for ages and you are always wrong when it comes to your nation's genetics and historical facts. You are from Eastern Armenia? Western Armenia is the major area for the ethnogenesis of the Armenian people.Urartu and other 'pre-and proto' Armenian kingdoms-confederations too place mostly at western Armenia(eastern Anatolia if you want it this way).Western Armenia = those parts in what was previously in Ottoman Empire, Roman Empire and so on, rather eastern Armenia(under Sassanid-Iranics).Western Armenia is way bigger than Eastern Armenia in size.So of course Western Armenia has much higher importance for Armenians historically.Denying this is denying Armenian history and heritage.Not any different from what the genociders did.Also there is not big genetic differences between Western Armenians and Eastern Armenians as mainland Greeks and Anatolian Greeks have or as Balkan Turks and Anatolian Turks have or Northern Italians and Southern Italian have or even Northern French and Southern French have.You got a really wrong image about western Armenia.I dont want to fight or anything but it is time to put things in a place especially with the Armenian issue, witch is getting ridiculus annoying. For the last time, the Academic samples from Armenia were fine with exception the maijkop witch it was mixed with a Russian individual.

FrankN said...

Does this decision concern all modern samples, or only Europe/ Near East.

I mean, there is still a lot of areas that are insufficiently covered by aDNA, and may for the foreseeable time remain so. I would hate to lose stuff like Onge, Papuans, Aeta, Indian tribals, also some SSA variety or Modern Jews. How will we in future be able to look at the genetic impact of the Arab expansion, if there are neither ancient nor modern Arabs in the G25? How shall we figure out the geographic origin of some Roman gladiator or soldier?

LGK said...

It isn't necessarily so much about deception as it is people just being misinformed or ignorant about their descent, even within the last few generations, without genealogical and/or local historical knowledge.

For instance, a friend would be considered 100% Greek, and from a particular region; but upon investigation found their great-great-grandparents included ethnic Italians and Aromanians. Moreover, their known Greek great-great-grandparents were not exclusively local residents but some had migrated from afar, across known ancestry clines (e.g. south vs north Greece).

All it takes is one generational step where that knowledge is not communicated, and you get someone of "X" nationality presenting as X ethnicity, when really they may be a mix of X, Y and Z within a meaningfully recent number of generations.

Michalis Moriopoulos said...

@LGK

When you have enough samples from a given group, it's pretty obvious who's an outlier and who isn't, no? I don't see what the fatal problem is here.

Even in the case of ancient samples, it's obvious that the two Anatolia IA samples are not representative of most of their contemporaries on the Plateau. We know that because we are able to compare them to their Bronze Age predecessors AND to modern Cappadocian Greeks and Turks.

How many historical non-living samples from the modern period (1500 onward) can we reasonably expect to see? I can only think of the Roopkund Lake people and the century-old Andaman Islander. And we only know the Roopkund Lake B people are Aegean Greeks because we have modern Aegean Greek samples to compare them to.

I just fear not keeping up a living moderns list will undermine the project, especially for underserved regions like Africa. Why not just have a disclaimer saying "living samples are often based on self-identification and therefore cannot be guaranteed to be representative"?

Onur Dincer said...

@Michalis

I just fear not keeping up a living moderns list will undermine the project, especially for underserved regions like Africa.

That is my biggest concern. Most people take genetic tests (including Global25) to learn about their relationship to modern populations, most people are not even aware of or interested in ancient DNA. I fear this change will cause a significant drop in Global25 sales.

Why not just have a disclaimer saying "living samples are often based on self-identification and therefore cannot be guaranteed to be representative"?

Will probably be sufficient to shut the mouths of most of the vocal but tiny minority of dissenters.

When you have enough samples from a given group, it's pretty obvious who's an outlier and who isn't, no? I don't see what the fatal problem is here.

Even in the case of ancient samples, it's obvious that the two Anatolia IA samples are not representative of most of their contemporaries on the Plateau. We know that because we are able to compare them to their Bronze Age predecessors AND to modern Cappadocian Greeks and Turks.


Fully agree.

How many historical non-living samples from the modern period (1500 onward) can we reasonably expect to see? I can only think of the Roopkund Lake people and the century-old Andaman Islander. And we only know the Roopkund Lake B people are Aegean Greeks because we have modern Aegean Greek samples to compare them to.

We probably will not see any medieval or modern non-living samples from most countries in the foreseeable future. Recent periods of most countries do not get enough attention to get funding for such costly tests.

@Frank

I mean, there is still a lot of areas that are insufficiently covered by aDNA, and may for the foreseeable time remain so. I would hate to lose stuff like Onge, Papuans, Aeta, Indian tribals, also some SSA variety or Modern Jews. How will we in future be able to look at the genetic impact of the Arab expansion, if there are neither ancient nor modern Arabs in the G25? How shall we figure out the geographic origin of some Roman gladiator or soldier?

Very good example.

CrM said...

@Anatolian Farmer

"Western Armenia is the major area for the ethnogenesis of the Armenian people."

This is the Azeri average, Azeris being people with extra Iranian and Turkic ancestry:

Target: Azeri
Distance: 2.0533% / 0.02053347
33.8 TUR_Tepecik_Ciftlik_N
25.6 IRN_Wezmeh_N
16.4 Yamnaya_RUS_Samara
10.2 GEO_CHG
8.4 Levant_PPNB
5.6 RUS_Devils_Gate_Cave_N

Now this is an Azeri ghost without any excess Iranian and Turkmen related ancestry (made by subtracting Ganj Dareh Historic and Turkmen from the Azeri average), compare it with the Armenian average:


Target: Azeri2
Distance: 2.9341% / 0.02934122
41.4 TUR_Tepecik_Ciftlik_N
25.0 IRN_Wezmeh_N
17.0 GEO_CHG
10.0 Levant_PPNB
6.6 Yamnaya_RUS_Samara
0.0 RUS_Devils_Gate_Cave_N

Target: Armenian
Distance: 2.7624% / 0.02762442
46.6 TUR_Tepecik_Ciftlik_N
20.0 IRN_Wezmeh_N
18.4 GEO_CHG
8.2 Levant_PPNB
6.8 Yamnaya_RUS_Samara
0.0 RUS_Devils_Gate_Cave_N

Distance to: Azeri2
0.02118157 Armenian
0.02206592 Assyrian
0.02469313 Georgian_Jew
0.03227480 Iranian_Jew
0.03484650 Armenian_Hemsheni
0.03529600 Turkish_Trabzon
0.03546857 Mountain_Jew_Chechnya
0.03837891 Greek_Trabzon

Point is, Caucasus Albanians used to be nearly identical to modern Armenians, who had little relations with West Armenia(East Anatolia), even the late Neolithic samples from Azerbaijan were closest to Armenians.

I also tried subtracting the Steppe ancestry from MBA-LBA Armenian samples, and the result is pretty Armenian-like.

Target: ArmLBA-Catacomb
Distance: 3.5430% / 0.03543033
46.0 TUR_Tepecik_Ciftlik_N
26.8 GEO_CHG
19.8 IRN_Wezmeh_N
6.4 Levant_PPNB
1.0 Yamnaya_RUS_Samara
0.0 RUS_Devils_Gate_Cave_N

Target: ArmMBA-Catacomb
Distance: 3.2751% / 0.03275101
50.4 TUR_Tepecik_Ciftlik_N
27.0 GEO_CHG
20.2 IRN_Wezmeh_N
1.8 Yamnaya_RUS_Samara
0.6 Levant_PPNB
0.0 RUS_Devils_Gate_Cave_N

Target: Lchashen-Lola
Distance: 3.0817% / 0.03081659
44.0 TUR_Tepecik_Ciftlik_N
29.4 GEO_CHG
17.4 IRN_Wezmeh_N
9.2 Levant_PPNB
0.0 RUS_Devils_Gate_Cave_N
0.0 Yamnaya_RUS_Samara

Another thing regarding the Steppe ancestry in modern Armenians, it's not as little as people think, you just need the relevant samples. It's interested to note that Armenia showed a large variety of Steppe-related ancestry, ranging from Progress-like to Steppe Maykop-like. And if Armenian speakers came directly from the Steppe, then it's possible that their Steppe was more like Kubano-Tersk rather than Yamnaya/Catacomb.
https://i.imgur.com/b0nr17b.png

Istakhr said...

I completely disagree with this move. Analyzing modern populations with aDNA is like half of population genetics. How are medieval samples supposed to replace thousands upon thousands of the diverse groups today? The uncertainty about a remain's ethnicity are probably greater than modern samples. Isn't the whole point of population genetics to see how people have changed compared to aDNA?

A compromise could be to have to separate sheets; one that is vetted and one that lists samples as is from studies. Samples get into the vetted sheet by checking if the sampling methods were sufficient for the population via supplementary data from the studies. If passed, the samples will be checked for outliers via PCA or admixture analysis. Certain criteria could be set so the community could help.

Davidski said...

@Istakhr

The datasheets with the modern, living samples are still up, and they will be for a while, and you can keep using them if you're confident that they produce reliable outcomes for you.

But, for a variety of reasons, I have to shift the focus to samples from archeological sites.

Gabriel said...

@Davidski

Are any Medieval samples from France, Germany or England coming soon?

Davidski said...

@Gabriel

Are any Medieval samples from France, Germany or England coming soon?

Yep.

Aram said...

G2a farmer

You have a very superficial knowledge of Armenian history and You claim that I am wrong? You don't know that Sasanids were controlling practically all Greater Armenia except some minor regions in the west. Current understanding of West East division is due to recent wars between Russia and Ottomans. And Eastern Armenia is not equal to modern Republic.

Also I never said that West Armenia was not important. Hayassa was in west Armenia and it is almost certainly the source of Armenian endonym. But Hayassa didn't fell from Moon. It formed as a result of migration from modern Armenia who themselves migrated from Steppe.
There is no other solution. As for Urartu it was a multiethnic place and it was not about ethnicities but State and dinasty. You need to read Zimansky to understand this

Aram said...

Btw I am mostly from Western Armenia. Erzerum region. But my ancestors migrated before the Genocide. I live in East Armenia and as I stated most Armenians in modern Armenia do not have their deep roots in this republic. Except regions like Karabakh and Syunik.
So I think it is good idea to have Medieval era samples from Republic of Armenia.
And I know that the difference is not huge and it is absolutely not related to Armenians because that East West cline existed long before Armenians existed. Most probably since post LGM period.
Quite contrary Armenians were probably the most important homogeneising force that made populations in those Highlands close to each other.

Ariel said...

For example: The Italian medieval samples are literally all over the place, from East-Med to northern Europe, same from Germany where there are a ton of outliers, Spain is complex too. You still need to interpret which samples are outliers and which ones are not. Arguably 18/19th century samples will be to closest to moderns but there is no reason to study them. Same for ancients, think about Romans, Celts or Greeks that have outliers and diversity even early on, what do you do? You need to make a decision and there is always going to be people that are not happy about it. Definitely when it comes to some of the groups, like Swiss-Italians and Piedmont in the modern g25 spreadsheet, there is no way they are representative. But you need to consider that there might have been migrations inside the Italian peninsula since the 14th/15th century and some urban populations in Northern Italy might have significant central Italian or South Italian ancestry, and it might be too old for the individuals to know, it's clear from the spreadsheet. Same for the Iberian peninsula, Germany, France, when there is obviously a lot of modern mixing. Going about and find the pure samples and make them the standard for a given population is extremely complicated, and I understand why David hasn't done that. And it will always be arbitrary to some degree. Maybe someone will make a user's spreadsheet with high quality samples from a given region. 23andme results are extremely consistent for example, and could be taken as a benchmark.

CHG Chad said...

@ Crm


Armenians are between Arslantepe and kura-Araxes. To model them, you will need a BA Anatolian source and something related to Kura-Araxes-Iran C,(thought K.A is Iran C mostly).They are not CHG shifted like Georgians, Laz,Trabzon or any other North Caucasus group etc. They are balanced when it Comes to Iran N and CHG components.And still their main component is ANF related, Witch means a source similar to Arslantepe played a major role to their ethnogenesis. Arslantepe-Medin Btw it was a Hittite-Hurrian hotspot.Later with the formation of Urartu they have become probably a little bit more West Asian and finally with the arrival of proto-armenians they eventually indoeuropicized.Some of the academic samples might be a little bit mesopotamia/Levant shifted But not crazy things.Armeniand due to their geographical position have come in contact with Assyrians and intermarriages between them was not rare.Their ydna Witch exception R1b and J2 markers there are Some weird J1 lineages among them, like Some Pontic Greeks from eastern Pontus have. Probably Jewish relation.The point is that eastern Armenians or western Armenians, there are not going to be Huge diffrences between them. Nothing like Mainland Greeks and Anatolian Greeks,North Italians and South Italians and So on.

Norfern-Ostrobothnian said...

When are we to expect the Uyelgi samples to become public?

CHG Chad said...

@ Aram

I agree with the most, thought the Armenian kingdoms covered both western and Easter Armenia before Roman-Byzantine Empire and Sassanids division.But still I will insist western Armenia is much more important for the ethnogenesis of Armenians but anyway forget it.This is a detail.My question to you is simple.Why you and some other Armenians dont like the academic samples in G25?What crazy do they have? You expected to see similar results like the samples from Areni?There is not enough steppe? What?What's so crazy with them?The decision that David takes has to do with issues like the Armenian samples,The Greek Peloponnese samples and i am pretty sure with Italians especially those from South, who cannot accept their East Med genes. Remember a person in AG attacking David that G25 Italian samples have error and they do not represent them.With a few words for 3,4 ethnic groups with their complexes we are going to lose this wonderful master piece.There are not ancient samples from Trabzon or Northeast Anatolia, so now I am going to lose the only thing related with my own people witch it was the modern samples in G25.Good job kiddos..

CrM said...

@Anatolian Farmer

Was Arslantepe important in the ethnogenesis of Caucasus Albanians? Because as I have said before, it would seem that Azeris without excess Turkic and Iranian ancestry would be nearly indistinguishable from modern Armenians, suggesting that an Armenian-like genetic profile could have developed on its own in the Caucasus in the Bronze Age (ignoring the LN Aze samples).

I would like to know what happened between KAC and Trialeti period in Armenia, because Armenia MBA-LBA + Lchashen can't be modeled as a mix of KAC and Steppe, and indeed would require something like Arslantepe to cut it, but would BA Anatolian ancestry reach as far as Azerbaijan?

After KAC but before Steppe influx in Armenia there were people that were slightly more CHG shifted but otherwise not too dissimilar from modern Armenians, their Anatolian ancestry being pretty high still, so the post BA East Anatolia might not be very important.

Target: ArmMBA_withoutSteppe
Distance: 3.2751% / 0.03275101
50.4 TUR_Tepecik_Ciftlik_N
27.0 GEO_CHG
20.2 IRN_Wezmeh_N
1.8 Yamnaya_RUS_Samara
0.6 Levant_PPNB

Target: ArmLBA_withoutSteppe
Distance: 3.5430% / 0.03543033
46.0 TUR_Tepecik_Ciftlik_N
26.8 GEO_CHG
19.8 IRN_Wezmeh_N
6.4 Levant_PPNB
1.0 Yamnaya_RUS_Samara

Target: Armenian_withoutSteppe
Distance: 2.9467% / 0.02946656
49.8 TUR_Tepecik_Ciftlik_N
21.4 IRN_Wezmeh_N
19.6 GEO_CHG
8.6 Levant_PPNB
0.6 Yamnaya_RUS_Samara

Target: Armenian
Distance: 2.7624% / 0.02762442
46.6 TUR_Tepecik_Ciftlik_N
20.0 IRN_Wezmeh_N
18.4 GEO_CHG
8.2 Levant_PPNB
6.8 Yamnaya_RUS_Samara

Samuel Andrews said...

90% of the time living people are good representatives of the ethnicity they represent. So I don't understand this decision.

@Davidski,

Do you think many of the German samples are not representative of Germans? Are Medieval, Black-Death era, samples from Germany less mixed than "living" samples?

Wot said...

@Onur Dincer

My problem with how the Armenian samples look in G25 is that they have less Steppe than even Arab Levantine, Mesopotamians and also less than Assyrian and West Iranians, they look almost like Georgian transplants, even if you say they are derived from a mix of the 15-30% Steppe samples we have from Bronze Age Armenia the value is still ridiculously low.

Ricardolindo said...

Please, reconsider. I like to see the ancient genetic components of present day populations in Global25.

Mr. J said...

"Soon" probably means within 6 months for these new Medieval samples.

Onur Dincer said...

@Crm

You are subtracting the wrong populations from the Azeris. The Turkic population that contributed to them was most likely very similar to the Central Asian medieval ancient genomes we have. The non-Anatolian-admixed-looking one of the two "Ottoman" Turkish ancient samples is very similar to the Karluk and Karakhanid ancient genomes. The modern Turkmens have large amounts of Iranian ancestry from the last 1000 years to be of any use. As for the Iranian population that contributed to the Azeris, you need to use a Western Iranian-speaking population with practically no detected Turkic admixture (so the Mazandarani, Zoroastrian and Kurdish averages could do, but Ganj Dareh Historic seems to have detectable Turkic admixture) as most probably it was the Turkic ancestors of the Azeris that brought the Turkic admixture to Iran in the first place.

Gill said...

Other users: Just save the latest moderns list and share it on Google Drive/OneDrive or something for the rest of us. If it's not "official", he can tune out any drama resulting from it.

Also, he said:

"I'm planning to eventually create several sets of Global25 datasheets based on individuals and populations from different periods, including the modern era."

Some modern pops are going to be useful for reference. From South Asia there's Balochi, Paniya, Onge, etc. And Papuans, Sardinians, Karitiana, Yoruba, etc commonly used populations like these might continue to be officially provided but it doesn't affect us either way, just save the list.

Onur Dincer said...

@Wot

My problem with how the Armenian samples look in G25 is that they have less Steppe than even Arab Levantine, Mesopotamians and also less than Assyrian and West Iranians, they look almost like Georgian transplants, even if you say they are derived from a mix of the 15-30% Steppe samples we have from Bronze Age Armenia the value is still ridiculously low.

All the full Armenian individuals whose GEDmatch results I have seen cluster very closely with the academic Armenian samples from Armenia. The academic Armenian samples have normal amounts of CHG for Armenians, not any higher, so the Georgian admixture theory does not hold. Also, you should compare the steppe levels of the Armenians to the geographically close populations, not to the distant ones such as the Levantines. The Armenians have higher steppe than the Pontians and Kartvelians and similar levels of steppe to the Assyrians, who have similarly high frequencies of Y-DNA hg R1b and even higher frequencies of R1a than the Armenians. The Western Iranians have higher steppe than the Armenians, but they are also IE speakers. If there is anything unusual in these results, it is the relatively high steppe levels among the Northern Semitic speakers, they seem to have absorbed a lot of IE peoples, I think the Iron Age Assyrian Empire played a pivotal role in this, it is also the polity that spread the Aramaic language all over the Levant, Mesopotamia and even parts of Anatolia.

The Bronze Age genomes from what is now the Republic of Armenia do not seem to have contributed to the modern Armenians in that big amounts (I have already covered that issue in the past on a previous thread of Eurogenes, you can google and find it), the Armenian ethnogenesis seems to have taken shape mostly in Western Armenia (in what is now eastern Turkey).

Istakhr said...

@Wot This isn't an issue with the Armenian samples. The only issue is that it would better if we had more Armenian samples, with separate averages for western and eastern Armenians. The current average is still quite representative of Armenians based on the personal coordinates I've seen from many Armenians. The reality is that there seems to be a group of people online who don't want to accept the average steppe ancestry of Armenians, and claim that their steppe is too low in comparison to many surrounding populations, despite being Indo-European speakers, therefore something must be wrong with the samples. Armenian steppe ancestry isn't even that low and I don't see an issue. It just means the ethnogenesis of Armenians is more intricate than other IE speakers. You could roughly model Armenians as a mix of Arslantepe_EBA and 'steppe heavy' EBA samples from Armenia.

CHG Chad said...

@ Crm

I do not agree with your post sorry. Azeris are very iranic overall with Some turkic.Your models are with ancient refrences. These populations didnt cobtribute to us. You Have to model with BA-IA-MEDIEVAL samples.

Conchubhar said...

Feel vy sry to hear that. I can understand the reasons behind your decision but u are making a big mistake. Most people are more interested in modern references than in ancient ones. Every modern reference include any link to ancient dna in its own structure. The biggest problem has to do in telling the difference between a geografic and a genetic model. If u know how to use the tool you just have to disregard geografic references that can provide you a wrong result. Once u are able to define your own genetic and customized model everything will be clear. But u have to know how to use this wonderful tool accordingly. I have been testing with most major labs for the last three years and just 23 and Me is able to provide you with a reliable result as they use the most powerful algorithm in the market. But they just offer you a regional disclosure by the information provided by our genetic matches. That´s a geografic model that in most cases has ntg to do with our genetic one. G25 is the only tool able to let u know about your real genetic structure for the time being. If u put a brake to the update of your modern references most people will decide to stay in the sidelines and you can lose your market as ur tool loses interest. I hope you will change ur mind some time in the near term. Forget about people who disagree with you without knowing how your tool really works and go ahead with your valuable job. Take care. Best

Randwulf said...

I will add to the pile of those that like having the modern G25 spreadsheets available. Of course, we can keep the last iteration, which I assume is the one that is there now.

Onur Dincer said...

@Istakhr

This isn't an issue with the Armenian samples. The only issue is that it would better if we had more Armenian samples, with separate averages for western and eastern Armenians. The current average is still quite representative of Armenians based on the personal coordinates I've seen from many Armenians. The reality is that there seems to be a group of people online who don't want to accept the average steppe ancestry of Armenians, and claim that their steppe is too low in comparison to many surrounding populations, despite being Indo-European speakers, therefore something must be wrong with the samples. Armenian steppe ancestry isn't even that low and I don't see an issue. It just means the ethnogenesis of Armenians is more intricate than other IE speakers. You could roughly model Armenians as a mix of Arslantepe_EBA and 'steppe heavy' EBA samples from Armenia.

You nailed the issue.

Onur Dincer said...

@Gill

You neglect the marketing issue. Without the modern reference coordinates linked from the main Global25 pages, less people might buy the test as most people buy it for the modern relationships.

CHG Chad said...


@ Wot


No you are wrong. David mention that the upcoming medieval samples from Armenia are low steppe admixed.You think the Areni samples represent modern Armenians but you are wrong. I know that you write as a nationalist with eurocentristic point of view, but you are already proven wrong.The steppe amounts among Armenians are fine duel with it.

Ryan said...

@Sam - "Do you think many of the German samples are not representative of Germans? Are Medieval, Black-Death era, samples from Germany less mixed than "living" samples?"

I think it's reasonable to expect mobility and intermarriage to have increased recently.

For example, I gave my ostensibly German grandfather's DNA to David for helping figuring out his and my ancestry. Here's what David came back with:

Ryan_C_GF
"German" 62.25
"Spanish_Cataluna" 15.55
"Czech" 13.1
"Tatar_Lipka" 4.85
"Sindhi" 4.25

I imagine it would be bad for someone like that to be in your reference sample.

Still can't figure out where the Hispanic came from. Thanks again though David.

CrM said...

@Onur Dincer
You're right, but it's a bit hard to determine how much Iranian-related ancestry they've got. The Turkic (Karakhanid) ancestry remains at the same amount more or less, at 13%.

Target: Azeri-Karakhanid13%
Distance: 0.8538% / 0.00853757
45.4 TUR_Arslantepe_EBA
40.0 Iranian_Mazandarani
7.8 RUS_Catacomb
5.0 IRN_Hajji_Firuz_C
1.6 AZE_Caucasus_lowlands_LC
0.2 KAZ_Karakhanid
0.0 ARM_LBA
0.0 ARM_Lchashen_MBA
0.0 ARM_MBA
0.0 AZE_Caucasus_lowlands_LN
0.0 Kura-Araxes_ARM_Kaps

Target: Azeri-Karakhanid13%
Distance: 0.8584% / 0.00858421
66.8 Armenian
25.2 Iranian_Mazandarani
3.4 AZE_Caucasus_lowlands_LC
3.0 RUS_Catacomb
1.2 IRN_Hajji_Firuz_C
0.4 KAZ_Karakhanid
0.0 ARM_LBA
0.0 ARM_Lchashen_MBA
0.0 ARM_MBA
0.0 AZE_Caucasus_lowlands_LN
0.0 Kura-Araxes_ARM_Kaps

Let's say their recent extra Iranian ancestry is at 25%. At the end you still get a very Armenian-like ghost. They have a lot of TUR_Arslantepe_EBA, and I wonder if that ancestry reached them directly, and if it wasn't some local continuiation of AZE_Caucasus_lowlands_LN.

Target: Azeri-Karakhanid13%-Mazandarani25%
Distance: 1.1445% / 0.01144494
89.0 Armenian
4.4 AZE_Caucasus_lowlands_LC
4.0 RUS_Catacomb
2.0 IRN_Hajji_Firuz_C
0.6 KAZ_Karakhanid
0.0 ARM_LBA
0.0 ARM_Lchashen_MBA
0.0 ARM_MBA
0.0 AZE_Caucasus_lowlands_LN
0.0 Iranian_Mazandarani
0.0 Kura-Araxes_ARM_Kaps

Target: Azeri-Karakhanid13%-Mazandarani25%
Distance: 1.1391% / 0.01139083
60.6 TUR_Arslantepe_EBA
20.8 Iranian_Mazandarani
10.4 RUS_Catacomb
6.0 IRN_Hajji_Firuz_C
2.2 AZE_Caucasus_lowlands_LC
0.0 ARM_LBA
0.0 ARM_Lchashen_MBA
0.0 ARM_MBA
0.0 AZE_Caucasus_lowlands_LN
0.0 Kura-Araxes_ARM_Kaps

Distance to: Azeri-Karakhanid13%-Mazandarani25%
0.01506586 Armenian
0.01696405 Georgian_Jew
0.01914827 Assyrian
0.02972379 Turkish_Trabzon
0.02978902 Iranian_Jew
0.03099546 Turkish_East
0.03219165 Mountain_Jew_Chechnya
0.03315682 Greek_Trabzon

Distance to: Azeri-Karakhanid13%-Mazandarani38%
0.01600243 Armenian
0.01959588 Georgian_Jew
0.02267267 Assyrian
0.02747859 Turkish_Trabzon
0.03145719 Greek_Trabzon
0.03211096 Iranian_Jew
0.03257122 Greek_Central_Anatolia
0.03542605 Mountain_Jew_Chechnya
0.03631591 Armenian_Hemsheni

@Anatolian Farmer

"Azeris are very iranic overall with Some turkic."

My model was on a ghost without the extra Iranic and Turkic. My point was that without them Azeris overall are very much like Armenians.

Simon_W said...

@G2a-M406 Anatolian Farmer

"The decision that David takes has to do with issues like the (...) and i am pretty sure with Italians especially those from South, who cannot accept their East Med genes. Remember a person in AG attacking David that G25 Italian samples have error and they do not represent them."

Definitely, yes. There was also one in the comments to one of the G25 posts. It was great comedy, the guy made a complete fool out of himself. I can't find these comments anymore, they must have been deleted. A pity for their pedagogic effect. And then there was that Italian woman on AG who furiously claimed that Davidski mislabelled European Jews as Italians in a PCA. Apparently the DNA subject tends to attract conservative Italians with an anti-Levantine attitude more than the liberal open-minded ones. I also remember discussing some years ago with an Italian woman who believed that Davidski hates Italians and pushes an evil agenda against them, lol. She was drawing upon some utterances a user called Polako made in a forum about 15 years ago or more. As if people were always the same.

Aram said...

G2a M406 farmer

Are You sure that You read for ages my comments? :)
Should You have done that You would knew that I was never particularly caring how much Steppe ancestry do Armenians have. For me it was fully sufficient that Z2103 and other HGs from Steppe migrated to Armenia and this was the reason of IE language.
Also I never had anti Levantine bias. Quite contrary many times I was saying that the dilution of Steppe was probably the Levantine influence. But I was partly wrong on this matter because pops like Arslantepe changed a lot the situation.
As for those modern samples. Some people complain about low Steppe but some complain that the Steppe is overemphasized. And some ( even in academic circles) don't want pronounce the word Steppe. So here You are overgeneralizing. People are diverse.




Aram said...

CrM

It is very likely that those who migrated to Near East in MBA were not pure Catacomb/Yamna. The presence of I2c a typical EEF farmer lineage could mean that those migrants were something like this Catacomb+Cucuteni-Tryplia. And in Azerbaijan they could have get Azerbaijan LC.

@All

And once more about modern Armenians. They are in the right place. They are the closest pop to ancient SC Neolithic farmers. What else is needed? :)

Distance to: AZE_Caucasus_lowlands_LN:MTT001
0.03367549 Armenian
0.03465252 Turkish_Trabzon
0.03667525 Greek_Trabzon
0.04118676 Assyrian
0.04148422 Iraqi_Jew
0.04278293 Greek_Central_Anatolia
0.04323291 Georgian_Jew
0.04351851 Armenian_Hemsheni
0.04424074 Lebanese_Druze
0.04427804 Druze

Wot said...

@G2a-M406 Anatolian Farmer

I feel like those types of comment shouldn't pass moderation, you are accusing me of things without even really knowing who I am and where I am from, in fact I have no ties to Armenia or even West Asia as a whole.
My problem is merely one of trying to square the ancient samples we already have and the modern situation with the single source of academic Armenian samples.

@Onur Dincer

"All the full Armenian individuals whose GEDmatch results I have seen cluster very closely with the academic Armenian samples from Armenia. "

And I've heard the exact opposite from other people, I'm honestly not sure who to believe.

"The Bronze Age genomes from what is now the Republic of Armenia do not seem to have contributed to the modern Armenians in that big amounts (I have already covered that issue in the past on a previous thread of Eurogenes, you can google and find it), the Armenian ethnogenesis seems to have taken shape mostly in Western Armenia (in what is now eastern Turkey)."

I'll look into it, but it's certainly not the most parsimonious explanation, those samples are all so close to each other and have a peculiar genetic make-up, what are they? North-East Caucasians migrating south? Unknown steppe lineage?

Davidski said...

Based on what I've seen, it seems that there was quite a bit of steppe ancestry in Armenia well into the Iron Age.

But during later periods, like Hellenistic and Urartian, you'll see a genetic shift away from the steppe and closer to where modern Armenians cluster.

CrM said...

@Onur Dincer
"The Bronze Age genomes from what is now the Republic of Armenia do not seem to have contributed to the modern Armenians in that big amounts (I have already covered that issue in the past on a previous thread of Eurogenes, you can google and find it),

The MBA-LBA Armenian samples differ from Modern Armenians by having more Steppe and slightly more CHG ancestry, but aside from that, there isn't that big of a difference, most notably they're still very Anatolia-rich just like modern Armenians are (if you ignore the excess Steppe). My guess is that it isn't a matter of them being ancestral or not, but it's a matter of some social layers, the Steppe-rich Armenians weren't the majority during MBA-LBA, which carried on to modern times.
https://i.imgur.com/lH11F55.png

"the Armenian ethnogenesis seems to have taken shape mostly in Western Armenia (in what is now eastern Turkey)"

Doubt it, because Azeris without excess Turkic and Iranian (said ancestry is recent in them) don't differ much from modern Armenians. Hell, we even have Neolithic samples from Azerbaijan who were genetically closest to modern day Armenians. East Anatolia may played a role in reducing the CHG ancestry in Armenians, but then again Neolithic, Chalcolithic Azeris and Armenians weren't particularly CHG-rich either.

Wot said...

@Davidski

So if the Armenian average is indeed what we see right now we could posit maybe a Hungarian-like scenario where the original admixture was diluted by mixing with surrounding populations?

CHG Chad said...

@ Aram


I am very happy that we actually agree now.My whole point has to do that Western Armenia(eastern anatolia) belongs actually to Armenians historically and people like you,you should not forget it or deny it.As for the steppe profile among Armenians,i think during the periods it increased and also decreased..but since you are mostly R1b and you are talking a IE dialect no1 can doubt about your Pontic-Steppe roots.Keep in mind that the surrounded folks have limited steppe autosomal,unlike Armenians and Assyrians.Georgians,Laz,Trabzon and Colchians in general score limited steppe(not more than 6%).About Levantines..this is an other story.These people get their steppe not only from LBA/IA periods but also much later during hellenistic,roman and maybe middle ages from Crusaders.So,i cannot figure why people comparing Armenians with Levantines but anyway.

CHG Chad said...

@ Simon

Yes,i am not gonna blame Italians only.Our folks from Greece are not that better.This is a good lesson btw for all of us.Now lets play again with gedmatch:DDDDD

CHG Chad said...

@ Crm

I think you giving too much attention to SC culture.You should wait for Trialeti samples.SC is LN/LC culture.We need genomes from LBA/IA eastern anatolia and Armenian highlands.KA vanished,then Trialeti come.Immigrants from the northern parts arrived at the Armenian highlands bringing steppe autosomal.The area is complicated AF.But indeed Armenians are not CHG shifted like kartvelians are.I will insist that Armenians are between Arslantepe and KA-Trialeti,thought during middle ages some components increased and decreased.I will give you an example.When Seljucks arrived in Anatolia they were fully pretty much central asian like(with a Persian/Iranic profile as well).When the beyliks started to taking place,these central asians folks started mixing with Native Anatolians and islamicized everyone on their way.What it was the result?These asiatic folks to started losing their east asian/siberian genes.Armenians are probably a similar case.They might started very Steppe admixed but components like Iran N,Levant N or CHG dominated the Steppe folks.Anyway,lets wait for the upcoming medieval samples from Armenia and will see!!!!

Onur Dincer said...

@Wot

And I've heard the exact opposite from other people, I'm honestly not sure who to believe.

In the overwhelming majority of Armenians with fully known Armenian ancestry you will see populations like the academic Armenian and Assyrian samples as the first population listed in their single population distance Oracle results and with very low distances overall.

I'll look into it, but it's certainly not the most parsimonious explanation, those samples are all so close to each other and have a peculiar genetic make-up, what are they? North-East Caucasians migrating south? Unknown steppe lineage?

How is it not the most parsimonious explanation? Armenians cannot be modeled as a continuation of the Bronze Age ancient samples from what is now the Republic of Armenia without substantial mix from elsewhere. Also, the historical Armenian lands are mostly in Western Armenia (in what is now eastern Turkey).

https://drive.google.com/file/d/1yc_ibPzJMSIPWcbWEMBleK24dVIa3NJf/view?usp=sharing

https://upload.wikimedia.org/wikipedia/commons/f/f4/Roman_East_50-en.svg

Onur Dincer said...

@CrM

The MBA-LBA Armenian samples differ from Modern Armenians by having more Steppe and slightly more CHG ancestry, but aside from that, there isn't that big of a difference, most notably they're still very Anatolia-rich just like modern Armenians are (if you ignore the excess Steppe).

You are downplaying the magnitudes. The steppe ancestry shrank from the 25-30% levels to about 5%.

https://drive.google.com/file/d/1yc_ibPzJMSIPWcbWEMBleK24dVIa3NJf/view?usp=sharing

My guess is that it isn't a matter of them being ancestral or not, but it's a matter of some social layers, the Steppe-rich Armenians weren't the majority during MBA-LBA, which carried on to modern times.

Read what Davidski wrote:

"Based on what I've seen, it seems that there was quite a bit of steppe ancestry in Armenia well into the Iron Age.

But during later periods, like Hellenistic and Urartian, you'll see a genetic shift away from the steppe and closer to where modern Armenians cluster."

It is almost certainly to do with migrations from Western Armenia (in what is now in eastern Turkey), lowering the steppe and homogenizing the gene pool of the emerging Greater Armenia.

Doubt it, because Azeris without excess Turkic and Iranian (said ancestry is recent in them) don't differ much from modern Armenians. Hell, we even have Neolithic samples from Azerbaijan who were genetically closest to modern day Armenians. East Anatolia may played a role in reducing the CHG ancestry in Armenians, but then again Neolithic, Chalcolithic Azeris and Armenians weren't particularly CHG-rich either.

Read what I wrote above. Part of modern Azeri lands (in what is now the Republic of Azerbaijan) became a part of Greater Armenia in the antiquity and was undoubtedly affected by the genetic homogenization trend. But the majority of Greater Armenia is in the western portion (in what is now eastern Turkey) like I said, so that part of course would have played the biggest role in the Armenian ethnogenesis and genetics. Unfortunately we have no ancient genomes from that portion yet, and this is why we are still making these meaningless discussions.

Onur Dincer said...

@Aram

And once more about modern Armenians. They are in the right place. They are the closest pop to ancient SC Neolithic farmers. What else is needed? :)

It would be a stretch to expect that the territory of what is now the Republic of Azerbaijan remained genetically almost unchanged during the millennia from the Late Neolithic until the arrival of the Armenians to that region, when so much genetic change was happening in the surrounding regions. Besides, that region is in the periphery of what would be Greater Armenia.

Onur Dincer said...

We have digresses a lot with our discussion of Armenian genetics. We should instead focus on the big picture. The modern population references in Global25 are representative of their populations overall, and this is what really matters. It is also the case for Armenians, but that is a side issue, the thread is about the big picture.

CrM said...

@Onur Dincer

"You are downplaying the magnitudes. The steppe ancestry shrank from the 25-30% levels to about 5%."
"Read what Davidski wrote:"

I'm not downplaying anything. I have showed you a model of what MBA-LBA Armenians are like, and what they look like WITHOUT Steppe ancestry. Here's the thing, those BA Steppe-rich Armenians are not a continuation of CHG-rich KAC + Steppe, there was a third party involved in the formation of Arm MBA-LBA, and that party was Anatolia-rich. It was either related to Arslantepe or to Aze Lowlands LN, and they mixed with KAC BEFORE the Steppe influx. So on the transitional period between KAC and Arm MBA-LBA, you had Armenians that were genetically like modern ones albeit with slightly more CHG. What happened to them? Did they all receive Steppe ancestry? And if they did, what about the connection between KAC and Hurro-Urartians (who used to inhabit East Anatolia), could they not be those Anatolia-enriched Kura-Araxans who migrated to East Anatolia before the Steppe input, essentially being shielded from additional Steppe that East Armenia received (assuming that Steppe-influx really was universal in whole of East Armenia and was not limited to certain people)?

"Read what I wrote above. Part of modern Azeri lands (in what is now the Republic of Azerbaijan) became a part of Greater Armenia in the antiquity and was undoubtedly affected by the genetic homogenization trend."
They probebly were affected by slight genetic homogenization, yet aDNA show that they used to be Armenian-like to begin with, and if, as you say, "It would be a stretch to expect that the territory of what is now the Republic of Azerbaijan remained genetically almost unchanged during the millennia from the Late Neolithic until the arrival of the Armenians to that region", then it would mean that the Armenian genetic influence on them would have been even more great than the Iranian and Turkic, which I also doubt.

Norfern-Ostrobothnian said...

Could you add missing ancient samples from this datasheet? Many of these are already present in Global25 views but not on the calculator, like Comb Ceramic.
https://drive.google.com/file/d/1gHYqNrbQIOHiEEE4Y79Jj0e49NQFrdoO/view

Aram said...

Changes in Urartian and Hellenistic period were quite easily predictable. Both Urartian rulers and Artashessids (Hellenistic) were frequently practicing population deportations and relocations. This was done to weaken local chieftains (nakharars) and decrease the chances of revolts. For example the Urartian fort Erebuni was settled by people from Erzinjan which is quite West. Artashessids on the other side relocated Jews and Cappadocians.

Onur

I am fully aware about changes after Neolithic. But this sample do not only represent Azerbaijan Neolithic but also give a hint about Armenia Neolithic. So this type of ancestry was present on a larger scale. This type of matching is off course partly a random thing but partly not because subsequent changes were NOT affecting the whole Highland. For example KA had very little impact in Arslantepe and imho in many other places. So modern Armenians are representing a sort off cumulative average of all ancient events. That is why I said they are good.

Davidski said...

@Norfern-Ostrobothnian

Which ones exactly?

Some are too low coverage to get decent coordinates for, but I can post them in a text file here.

Aram said...

I must add that I agree with Onur. Many people who order G25 are interested to compare themselves to modern pops. They don't know the meaning of this ancient samples. At best they are interested by Iron Age.
So the lack of modern G25 samples will affect this type of demand.

Wot

Modern Armenians have a lot Proto Armenian ancestry. I know many Armenian specific haplotypes unrelated to R1b. Proto Armenian is not equal to Steppe Catacomb. Catacomb was PaleoBalkano- Armenian.
Proto Armenian was a population that formed in ancient Armenia and had 20-25% of Steppe + 75-80% of locale. Armenians can have more than 20-25 percent of Proto Armenian ancestry With more ancient DNA this will become more clear. Armenian ethnogenesis is very different from Hungarian because Hungarian had a Empire while Armenians didn't.

Modern Armenian samples were collected in this Institute. http://www.molbiol.sci.am/
Long before Yamba paper and all this Steppe saga.




Norfern-Ostrobothnian said...

@Davidski
Those are G25 coordinates, so I guess they have been good enough to be converted? I got them from a G25 Anthrogenica thread.

Aram said...

And finally about Steppe in Armenia. Steppe in Armenia is not anymore 5%. It is higher. To correctly calculate the Steppe one must remember that KA also had Steppe. So it is cumulative. That is why the baseline must be _before_ any Steppe came. The baseline must be before KA and must be pure. So Arm Chl is out of question. Here what we have before KA and now.

Target: Armenian
Distance: 1.2195% / 0.01219540
46.2 IRN_Hajji_Firuz_C
43.8 TUR_Arslantepe_LC
10.0 RUS_Catacomb

10 percent. Notice the excellent fit. But maybe we ignore CHG effect? So let's add Meshoko.

Target: Armenian
Distance: 1.2078% / 0.01207790
44.6 TUR_Arslantepe_LC
43.6 IRN_Hajji_Firuz_C
9.6 RUS_Catacomb
2.2 RUS_Darkveti-Meshoko_En

Only a slight decrease.
Now let's add Neolithic farmers.

Target: Armenian
Distance: 1.1756% / 0.01175596
39.0 IRN_Hajji_Firuz_C
37.6 TUR_Arslantepe_LC
13.4 AZE_Caucasus_lowlands_LN
9.6 RUS_Catacomb
0.4 RUS_Darkveti-Meshoko_En

Amazing. But let's see what happens if we finally add KA.

Target: Armenian
Distance: 1.1025% / 0.01102487
44.2 TUR_Arslantepe_LC
33.4 IRN_Hajji_Firuz_C
10.8 Kura-Araxes_ARM_Kaps
8.6 RUS_Catacomb
3.0 AZE_Caucasus_lowlands_LN

Steppe declined as expected. But only slightly. So Steppe in Armenia is clearly more than five percent. Almost certainly with new samples from Van and North Iraq this will change.

Now let's add Arm LBA.

Target: Armenian
Distance: 1.0869% / 0.01086939
42.6 TUR_Arslantepe_LC
31.8 IRN_Hajji_Firuz_C
10.8 ARM_LBA
8.4 Kura-Araxes_ARM_Kaps
5.2 RUS_Catacomb
1.2 AZE_Caucasus_lowlands_LN

Clearly there were Steppe elsewhere outside those LBA samples because they can't explain the Steppe. Now let's remove Catacomb and bruteforce things

Target: Armenian
Distance: 1.1290% / 0.01129021
39.4 TUR_Arslantepe_LC
29.2 IRN_Hajji_Firuz_C
26.4 ARM_LBA
5.0 Kura-Araxes_ARM_Kaps

Heh. Rumours about the death of Arm LBA were exaggerated. This gives the approximative value of Proto Armenian share in modern Armenia. 25 percent.
No Hungarian scenario.

Aram said...

To avoid once more talks about West East. I must add that Arm LBA Is just the closest proxy for various Proto Para Armenian groups. This do not mean they were not living in West Armenia. Simply we don't have samples from there. They could have lower Steppe. But the source of Steppe for them was Arm LBA not Catacomb.

Davidski said...

@Norfern-Ostrobothnian

Yes, those are G25 coordinates (scaled).

Norfern-Ostrobothnian said...

Well, if you could add the two Baltic CCC samples, that would be nice. They originate from this publication I believe.
https://www.nature.com/articles/s41467-018-02825-9

Davidski said...

They don't have enough data for reliable coordinates. But there's nothing remarkable about them anyway, certainly not in the context of Uralic origins.

Norfern-Ostrobothnian said...

@Davidski
I see. What would be a good ballbark for the amount of SNPs that they would work as a sample?

Davidski said...

At least 50,000 out of the usual 1240K.

Onur Dincer said...

@CrM

You make a lot of assumptions with the currently quite limited ancient DNA data. Let's wait for the ancient genomes from historical Western Armenia. Anyway, the main reason I do not want to continue this discussion is that this thread is not about Armenian genetics but about Global25 as a whole, so let's focus on that.

Norfern-Ostrobothnian said...

I checked the .anno file and Tamula1 & 3 have SNP coverage of 160,654 and 155,136 respectively, which isn't high quality, but there are other Baltic samples from the paper with less SNPs that are in the official datasheet. They seem to be at least useful to model Saami people, which would make sense in the context of Bolshoy and Kiukainen, as both formed on the basis of local CCC.

Target: Saami
Distance: 2.7896% / 0.02789617
25.2 Tamula1
24.8 Nganassan
24.4 Baltic_EST_BA
12.8 Yamnaya_RUS_Samara
8.0 Early_European_Farmer
4.8 RUS_Karelia_HG
0.0 ITA_Villabruna

Target: Saami
Distance: 3.4979% / 0.03497860
37.8 Tamula1
23.4 Nganassan
19.8 Yamnaya_RUS_Samara
12.8 Early_European_Farmer
6.2 RUS_Karelia_HG

Target: Saami
Distance: 4.1499% / 0.04149908
29.4 RUS_Karelia_HG
23.0 Yamnaya_RUS_Samara
22.6 Nganassan
16.8 Early_European_Farmer
8.2 ITA_Villabruna

Target: Saami
Distance: 2.9494% / 0.02949421
32.8 Baltic_EST_BA
24.6 Nganassan
17.6 RUS_Karelia_HG
11.2 Yamnaya_RUS_Samara
8.6 Early_European_Farmer
5.2 ITA_Villabruna
0.0 ITA_Villabruna

I'll check with qpAdm if they work there too, or if this is just a circumstantial amalgamation of Western and Eastern hunter gatherer proportions.

However, Latvia_MN2 has 758,978 SNPs and would thus be a more confident CCC sample.

As for universally useful samples, Andronovo samples would probably be well received, and most of them have a great amount of SNPs.

Onur Dincer said...

@Aram

Thanks for supporting my arguments on the Global25 issue.

Before finishing the discussion of Armenian genetics, I want to point out that we actually know what is now the Republic of Azerbaijan did not preserve its modern Armenian-like (minus the steppe) Late Neolithic genetics long as already by the Late Chalcolithic times it had turned into something much more Iranian-like:

https://drive.google.com/file/d/1bNjQVZ82h5jige78XTxYRUG6D1gKS6Ci/view?usp=sharing

I repeated this analysis using Catacomb instead of Yamnaya as the steppe source to test how much the steppe levels would be affected, here are the results:

https://drive.google.com/file/d/1hD5Ijb9KQQnELuHCP65vesoeJSFXn2as/view?usp=sharing

There is barely any change. So the higher steppe results in some of your analyses should be to do with using more recent ancient sources. But since in my analyses I compare modern Armenians with ancient populations from as early as the Late Neolithic, I have to use older ancient genomes as sources than you do. But my methodology does not affect the relative proportions of steppe ancestry much, so the point remains that in analyses in which the MBA-LBA genomes from what is now the Republic of Armenia get about 25-30% steppe modern Armenians get about 5% steppe.

Aram said...

Onur
I understand Your methodology. Simply the difference comes from the presence of CHG. 20% is masking Steppe. And the distances in Your model is higher. Mine is lower than 2% with just few samples. If You add Hajji Firuz and Arslantepe they will "eat" everthing else for Armenians.
As for Azerbaijan LC. It is Leyla tepe thus Chaff faced ware which means they have an influx from North Iraq and possibly partly NW Iran. Look that Levant also increase in Leyla tepe. It is in accordance with archaeology. But there wasn't total replacement.

Aram said...

Ancient samples from North Iraq and SE Turkey will change a lot our understanding. I suspect very basal people were living in that region. Plain regionaregion Anatolians were affected by WHG. They also can mask any Steppe ancestry. That is why we have such a problems with correct measuring of Steppe. And we get such different results.

Onur Dincer said...

@Aram

Yeah, our ancient genomes from West Asia are still quite limited and the current methodologies have difficulties in calculating the exact proportions of ancestry.

But - you will probably agree with this - the academic modern Armenian samples from Armenia are representative of the modern Armenian genetics overall irrespective of the difficulties in the calculations of things like exact proportion of steppe ancestry (which is a general issue, not just related to Armenians, and has nothing to do with modern population sampling).

Norfern-Ostrobothnian said...

The Comb Ceramic does seem to be useful - if not necessary - for the Saami. It also makes more chronological for the source of the EHG ancestry in the region than just pure EHG. Any BOO ancestry would accompany the local Asbesthos Ware variant of the CCC in Lapland.

Nganasan: 0.246±0.037
Poland_CordedWare.SG: 0.537±0.064
Latvia_MN_Comb_Ware.SG: 0.218±0.064
chisq: 13.622
tail prob: 0.478207973

https://docs.google.com/document/d/e/2PACX-1vS8Q_etk6YOWMAjJZO1YLldAbhgzgKYCw92xW0GvSuJwJG_IAyjyOJhDol8VJD4m17yEE8WZX5wHyL2/pub

With Narva the tail prob passes but it is noticeably lower, and the CWC is abnormally high. Not that Narva ancestry would make much sense anyways, and the tail prob without the Narva is actually higher.

Nganasan: 0.277±0.047
Poland_CordedWare.SG: 0.682±0.086
Lithuania_EMN_Narva: 0.040±0.057
chisq: 21.861
tail prob: 0.08152

https://docs.google.com/document/d/e/2PACX-1vSXcFr4fV3lvmd7OnHM-6b3GFTDoIxkZ4z4QMJJzjQxGx48Rl4xjoqFkbNbkg7rF0iNOAG-gwt3jIqf/pub

Now what do I think this means? If we backtrack, I believe that the proto-Saami speakers could have absorbed the Kiukainen and other Asbesthos ware peoples, which would have introduced the paleo-Saamic substrate. This may or may not include people like the BOO. If we go further back, the CCC might partly come from late Baltic CWC and post-Fatyanovo or even Volosovite and late-Kargopol people. Ultimately, the CCC might also partly come from proto-Uralics themselves, and may act as a proxy for all non-Steppe EHG or even WSHG. I would still place a big bulk of this ancestry descending from the various Asbesthos ware peoples.

So in short ZVEJ31/Latvia_MN2.SG would be a reputable sample in the official datasheet.

Norfern-Ostrobothnian said...

One thing of note is that I4554 (ZVEJ24) and I4435 (ZVEJ13) cluster with Karelia HG instead of Latvia HG. I4436 (ZVEJ14) is closer to EHG but falls between LHG and EHG. According to amtdb these samples would just fall into early CCC (4200 BCE - 3900 BCE). Along with ZVEJ31 they could form an early Baltic CCC group.

Norfern-Ostrobothnian said...

Although it does seem like ZVEJ24 and 13 are effectively EHGs with minor Narva ancestry, while 14 is closer to the later Tamula(3) CCC.
https://imgur.com/a/KrwyGOT

Aram said...

Onur

Just an addition. In this PCA it is visible that the Tepecik Ciftlik is slightly more northern shifted than the main Arslantepe cluster. So it's use can affect the result.

https://3.bp.blogspot.com/-t0r50UaE9ZA/Xt9BgaMJDwI/AAAAAAAAI8I/XMxVTm_S8uIegBZhOc1gwaBXekJIKcP3gCLcBGAsYHQ/s1600/Arslantepe_LC_Maykop_PCA.png

As for modern academic Armenians. As far as I know Armenian geneticians had very rigurous approach. Self declaring was not sufficient. The donor should also declare that all his grandparents were Armenians. So I don't doubt about their authenticity.
My only concern is that probably they were little to strict eliminating outliers which made those samples too compact.
And this has some negative repercussions in Gedmatch matching. Also regional diversity was not taken in account as in Georgia.
But now Yepiskoposyan's team had collected separate West and East sets. They used them in that recent paper.

Btw I was a candidate for West Armenian set. But one of my grandparents was not a pure westeros so I was disqualified.

Aram said...

In sum those modern Armenian samples are in accordance with ancient samples from our region.

Wot said...

@Aram

Your definitions and terms don't make much sense, what's the point of defining "proto-Armenian"? It's not a useful term and its creation seems merely a rhetorical device, if a "proto-Armenian" population was born out of 20-25% original Steppe admixture and 75-80% local admixture like you say, given that population was further diluted, how in the world is that not a "Hungarian" scenario? I'm pretty sure one could come up with a "proto-Hungarian" population too.

Norfern-Ostrobothnian said...

The Latvian Middle Neolithic groups are really heterogeneous and can be pretty confidently split between Narva and CCC, albeit closer to one another than EHG and older Narva.
https://ibb.co/XzWGQVC
The Tamula samples show further shift to the intermediate territory between EHG and Narva, and Tamula1 glares with it's minor shift toward CWC.

Onur Dincer said...

@Aram

Yes, the academic modern genome samples are being complemented with new ones as the years pass, so it is prudent to keep using them as references in analyses such as Global25. When the Behar et al. paper was published in 2010, the Cappadocian Turkish samples it used became the first academic Turkish genome-wide samples, but over the years new academic Turkish samples have been added to them, representing many more regions of Turkey, and with the non-academic Turkish samples of genetic projects such as the Turkish DNA Project we now have Global25 reference samples even for the Balkan Turks (divided into "Turkish_Deliorman" and "Turkish_Rumeli" in Global25 based on their region of origin in the Balkans).

Onur Dincer said...

@Davidski

Will you add the new samples in this new paper to your Global25 modern references before discontinuing your modern references (assuming you will discontinue them)? People were asking about them Anthrogenica.

Davidski said...

@Onur

It seems to be sequence data, not genotypes, so I won't spend the time converting it. If someone else does it, I can run them.

Onur Dincer said...

@Davidski

I see, thanks. I will ask the interested members on AG if they can do the conversion.

Aram said...

Wot
We have ancient DNA from Hungary and it is clear that they had more cases of haplogroup N than today. Obviously they had also that Siberian ancestry which is today absent. I don't see how You see parallels with Hungarians if the Yamnayan Z2103 is still one of the most frequent haplotypes in Armenia and there is still 8% of Steppe after 4000 year of that migration while the Siberian ancestry and N are practically lost in Hungary in less than 1000 years. So really I don't see how You can see parallels.

As for Proto Armenian term. Yes it is legitimate to use Proto Iranian or Proto Oghuz terms. Because if You do not that You will not correctly understand why modern Turks speak Turkic. Turkic homeland was almost certainly in Mongolia but Oghuz speakers came to Near East not from Mongolia in Iron Age but from Central Asia in Middle Ages. So they were a mix of Central Asian and Mongolia IA. This is the correct measuring of Turkic impact in Anatolia. As You can understand this will inflate the impact. The same is true for Proto Iranians. Proto Iranian community formed in South Central Asia as a mixture of Andronovo and BMAC which later expanded to Near East. I know this because they are young 3300 year haplotypes in Iran if possible BMAC origin.
The same is in Europe. Proto Celtic is not equal to Bell Beaker. Otherwise why You would have Basques full of Beaker ancestry. Or Proto Germanic is not equal to CWC.
So once more why if this is true for everyone why not for Proto Armenian or Proto Armenic which is a better term (proposed by Kassian) because it includes other extinct Armenic tribes like Mushki.
The Hungarian case is different. Even if You imagine a Proto Hungarian pop in Steppe the net result will not change much. As I said they virtually lost the Siberian for yet unknown reasons.
One theory that they passed a bottleneck and true Hungarians were killed in wars with Ottomans. Other possibility is this one https://en.m.wikipedia.org/wiki/Magyarization

Nevertheless the Hungarian case is not just a simple dilution.

Obviously You will not find something similar in Armenian history.

Aram said...

Onur

This is the map of Chaff faced ware. 4500-3500 bc.

https://adnaera.com/wp-content/uploads/2019/03/Europe-3800-3600-BC.png

And here a paper. Also with map.

https://www.persee.fr/doc/paleo_0153-9345_2010_num_36_2_5387

Arslantepe VII layer was CFW. We have 4 the oldest samples from this period. The rest are from younger VIB layer. This culture made most Armenia more basal. And it really had a lot off impact. That is why it should always taken into account when calculating Steppe in Armenia.

Onur Dincer said...
This comment has been removed by the author.
Onur Dincer said...

@Aram

Thanks for the article. Will look into it when I have spare time.

By the way, this may sound like hairsplitting, but the Proto-Oghuz actually lived in what is now Mongolia or somewhere just to its west during the Gokturk era and they migrated to the Aral steppe of what is now Kazakhstan several decades after the fall of the Second Gokturk Khaganate, mixed with the Pecheneg remnants there, another Turkic people, who had come to the Aral steppe from the east earlier, and formed their Oghuz Yabgu State of the Aral steppe. Thereafter, they become the Oghuz the medieval writers knew, so the term "Proto-Oghuz" is usually not used to denote the Aral steppe Oghuz but to denote their Oghuz ancestors who lived around what is now Mongolia before the westward migration. For this reason, I and many others prefer to use the term "Oghuz/Turcoman" (or something similar) to denote the Oghuz coming to Anatolia and environs from the Aral steppe rather than the term "Proto-Oghuz," as this is historically more accurate and it includes their newly acquired (with the Islamization of the Oghuz) name "Turcoman" ("Turkman" or "Turkmen") as well.

Aram said...

Onur

Terms are off course not crucial. Turkic migrations are well attested so we know the names of this tribes. Oghuz Turcoman is OK. I wanted to highlight the concept.
In most cases we don't know the names of this MLBA pops in Armenia so we must use some terms.

Onur Dincer said...

@Aram

I agree that we have to use terms like Proto-X at some point, but in many cases that some point is hard to decide due to the limited data and the conflicting theories based on the limited data. I myself have no issue a priori with calling a predecessor population of the Armenians that had already moved out of the steppe "Proto-Armenian," reserving for their ancestors living in the steppe a term like "Pre-Proto-Armenian." (assuming they already spoke in the steppe a predecessor language distinctly evolved towards Armenian).