PDA

View Full Version : Using Global25, how different are the Irish and the English?



Sikeliot
11-15-2019, 02:28 AM
Someone else might be able to better figure out how to model them than me, someone with more exposure to NW Europe than me. But how similar, or different, does it tell us the Irish are from the English?

JerryS.
11-15-2019, 02:41 AM
and German from Dutch....

Molfish
11-18-2019, 09:50 AM
I don't think Global25 can tell the whole story regarding the difference between the Irish and English, especially considering the English sample is rather small and the ancient samples are as yet too few and sparse to adequately compare, especially for Ireland.

It's important to look at academic Fst charts as well as something like G25
https://i.postimg.cc/529FzLtt/Dsucif-TXg-AEGz1a.jpg
https://i.postimg.cc/4NnTPXcg/FST-2.jpg
https://i.postimg.cc/XvCNBY9J/FST-9.jpg

The overall picture is that while being descended from essentially the same stock, the Irish are far more Insular than the English, who are probably mostly post Bronze Age continental Celto-Germanic descended.

firemonkey
11-18-2019, 11:25 AM
All I know is my English ancestry is the largest ancestry re my paper trail. Irish usually turns up as my highest component in G25 .

Jessie
11-18-2019, 02:48 PM
I don't think Global25 can tell the whole story regarding the difference between the Irish and English, especially considering the English sample is rather small and the ancient samples are as yet too few and sparse to adequately compare, especially for Ireland.

It's important to look at academic Fst charts as well as something like G25
https://i.postimg.cc/529FzLtt/Dsucif-TXg-AEGz1a.jpg
https://i.postimg.cc/4NnTPXcg/FST-2.jpg
https://i.postimg.cc/XvCNBY9J/FST-9.jpg

The overall picture is that while being descended from essentially the same stock, the Irish are far more Insular than the English, who are probably mostly post Bronze Age continental Celto-Germanic descended.

There looks to be some drift going on there. You can see it in relation to Norway and Iceland. You can also see the effect of drift on Ireland and Wales also.

sktibo
11-18-2019, 03:02 PM
I found the admixture estimate from the population genomics of the viking world to be very interesting, left is modern populations, right is ancient
34719

Capitalis
11-19-2019, 05:14 PM
I found the admixture estimate from the population genomics of the viking world to be very interesting, left is modern populations, right is ancient
34719

If I'm reading this correctly, there was no other 'southern' source used than Italy, which I guess is why the model output Irish as being ~9% Italian, and English as ~18% Italian. David Reich will be happy, anyway. I think this must be a model using only the Viking samples from their paper, or am I wrong?

sktibo
11-19-2019, 06:40 PM
If I'm reading this correctly, there was no other 'southern' source used than Italy, which I guess is why the model output Irish as being ~9% Italian, and English as ~18% Italian. David Reich will be happy, anyway. I think this must be a model using only the Viking samples from their paper, or am I wrong?

I am not sure where they sourced their modern samples from, Sco/Ire being modeled as ~9% Italian cannot be using only the samples from their paper because those are not ancient samples. The right hand side chart is using ancient samples, left is modern.

As for the right side ancient chart, I wasn't yet able to find what exactly "UK B" is.

Molfish
11-20-2019, 01:16 AM
I am not sure where they sourced their modern samples from, Sco/Ire being modeled as ~9% Italian cannot be using only the samples from their paper because those are not ancient samples. The right hand side chart is using ancient samples, left is modern.

As for the right side ancient chart, I wasn't yet able to find what exactly "UK B" is.
Could you explain what the chart says regarding modern British/Irish populations and their makeup? I can't make heads nor tails of it.

sktibo
11-20-2019, 01:23 AM
Could you explain what the chart says regarding modern British/Irish populations and their makeup? I can't make heads nor tails of it.

Population or target is the row, percentage is the column. So, Wales = 58% UK, 22% Denmark, 1% Sweden, 3% Norway, 1% Poland, 13% Italy, and 2% Finland with this model.

Molfish
11-20-2019, 01:30 AM
Population or target is the row, percentage is the column. So, Wales = 58% UK, 22% Denmark, 1% Sweden, 3% Norway, 1% Poland, 13% Italy, and 2% Finland with this model.
Thanks, and the columns represent ancients of those countries?

sktibo
11-20-2019, 01:57 AM
Thanks, and the columns represent ancients of those countries?

I think they represent moderns, I'm not sure where to find out what they represent with certainty, they may have built their components from ancient samples, but the labels being modern countries indicates to me that these are likely modern reference populations and they are testing these groups against that.

Fig. S11.14: Population-based admixture estimates based on the spatial median, as described in the
text. a) estimates of the proportion of ancient populations in modern populations. b) estimates of the
proportion of ancient populations in ancient samples.

Not clear if it's using moderns or ancients, as it just says "estimates of the proportion of ancient populations in" and doesn't specify what it was built with.

Just found this trying to find an answer:

"Fig. 4: Spatiotemporal patterns of Viking and non-Viking ancestry in Europe during the IA, EVA and VA.

UK = ‘British-like’ / ‘North Atlantic’ ancient ancestry component. Sweden = ‘Swedish-like’ ancient ancestry component. Denmark = ‘Danish-like’ ancient ancestry component. Norway = ‘Norwegian-like’ ancient ancestry component. Italy = ‘Southern European-like’ ancestry component. See Table S11.2 for statistical tests. The ‘Swedish-like’ ancestry is the highest in present-day Estonia due to the ancient samples from the Salme ship burial, which originated from the Mälaren Valley of Sweden, according to archaeological sources."

So maybe they did build these with ancient samples. Odd naming system.

It gave a reference to supplementary figure 11.2. Upon going there, there's a lot more information on this and the answer isn't jumping off the page to me. https://www.biorxiv.org/content/10.1101/703405v1.supplementary-material
Have a look maybe? Page 44.

Molfish
11-20-2019, 02:06 AM
I think they represent moderns, I'm not sure where to find out what they represent with certainty, they may have built their components from ancient samples, but the labels being modern countries indicates to me that these are likely modern reference populations and they are testing these groups against that.

Fig. S11.14: Population-based admixture estimates based on the spatial median, as described in the
text. a) estimates of the proportion of ancient populations in modern populations. b) estimates of the
proportion of ancient populations in ancient samples.

Not clear if it's using moderns or ancients, as it just says "estimates of the proportion of ancient populations in" and doesn't specify what it was built with.
Either way, their result for England is quite similar to how I've modelled my mother's data in G25, bearing in mind they didn't source France/West Central Europe.

https://anthrogenica.com/showthread.php?18911-North-sea-populations-using-G25&p=622823&viewfull=1#post622823

Capitalis
11-20-2019, 07:07 PM
I think they represent moderns, I'm not sure where to find out what they represent with certainty, they may have built their components from ancient samples, but the labels being modern countries indicates to me that these are likely modern reference populations and they are testing these groups against that.

Fig. S11.14: Population-based admixture estimates based on the spatial median, as described in the
text. a) estimates of the proportion of ancient populations in modern populations. b) estimates of the
proportion of ancient populations in ancient samples.

Not clear if it's using moderns or ancients, as it just says "estimates of the proportion of ancient populations in" and doesn't specify what it was built with.

Just found this trying to find an answer:

"Fig. 4: Spatiotemporal patterns of Viking and non-Viking ancestry in Europe during the IA, EVA and VA.

UK = ‘British-like’ / ‘North Atlantic’ ancient ancestry component. Sweden = ‘Swedish-like’ ancient ancestry component. Denmark = ‘Danish-like’ ancient ancestry component. Norway = ‘Norwegian-like’ ancient ancestry component. Italy = ‘Southern European-like’ ancestry component. See Table S11.2 for statistical tests. The ‘Swedish-like’ ancestry is the highest in present-day Estonia due to the ancient samples from the Salme ship burial, which originated from the Mälaren Valley of Sweden, according to archaeological sources."

So maybe they did build these with ancient samples. Odd naming system.

It gave a reference to supplementary figure 11.2. Upon going there, there's a lot more information on this and the answer isn't jumping off the page to me. https://www.biorxiv.org/content/10.1101/703405v1.supplementary-material
Have a look maybe? Page 44.

Thanks for all the explanation. I thought the 'Italian' component had to be from samples used in the paper because it's a bizarre choice to use as a southern reference, otherwise.

The main worry for me in any of these models is their confidence that they can differentiate between British- and Danish-like. This is why I dislike the POBI paper and all of its offspring (there was a Scottish version recently). When (e.g.) Razib quotes from the POBI paper that modelled English as up to 40% Germanic using modern references, I'm annoyed not because I want English and Germans to be separate, but because I don't think it's good science.

Of course, I made a calculator using modern references but that's a different matter (I'm a crank).

EDIT: supp. file 03, for anyone else looking. File 02 has archaeological info.

EDIT2: This is where I start getting confused by DNA papers.


We used fs2.0.8 (www.paintmychromosomes.com) using the protocol described with the software to paint 1675 modern individuals primarily from across Europe (UK, Italy, Poland, Denmark, Sweden, Finland, Norway, as well as China and Africa) who together form the “modern sample”.

This isn't really across much of Europe, and UK/Denmark/Norway seems tricky to split, but I'll read on.


The clustering is strong and perfectly stratified by population label. Each population is characterized by receiving higher ancestry from its own population, implying that each represents a unique aspect of genetic drift.

I follow so far. Looking at the bottom left of the matrix (fig. S11.2), their ability to isolate falls apart for Denmark and the UK.


We now use ChromoPainter v2 to paint each ancient sample against each modern sample.

Seems backwards but I've never read the supplements for this method before, so what do I know?


In terms of difficult to assign populations, Denmark is one. The individuals which best match a Denmark population also well-match the “UK181_g” population.

:)


The UK populations contain individuals from the PoBI study and so we can confirm that these are English and hence contains a high proportion of Anglo-Saxon ancestry.

:\ Where is the science behind this statement? POBI used modern Germans to claim that modern English are Anglo-Saxons. They might well be but this isn't science.


The “UK61_e” population which is well-matched by the ancient Orkney individuals contains individuals from Scotland and Northern Ireland, whilst “UK_24f” contains individuals from Wales.

Isn't this all backwards (modern to ancient), or am I wrong again? And why did they think modelling Irish as Italians made any sense? (Rhetorical questions).

sktibo
11-20-2019, 10:02 PM
Thanks for all the explanation. I thought the 'Italian' component had to be from samples used in the paper because it's a bizarre choice to use as a southern reference, otherwise.

The main worry for me in any of these models is their confidence that they can differentiate between British- and Danish-like. This is why I dislike the POBI paper and all of its offspring (there was a Scottish version recently). When (e.g.) Razib quotes from the POBI paper that modelled English as up to 40% Germanic using modern references, I'm annoyed not because I want English and Germans to be separate, but because I don't think it's good science.

Of course, I made a calculator using modern references but that's a different matter (I'm a crank).

EDIT: supp. file 03, for anyone else looking. File 02 has archaeological info.

EDIT2: This is where I start getting confused by DNA papers.



This isn't really across much of Europe, and UK/Denmark/Norway seems tricky to split, but I'll read on.



I follow so far. Looking at the bottom left of the matrix (fig. S11.2), their ability to isolate falls apart for Denmark and the UK.



Seems backwards but I've never read the supplements for this method before, so what do I know?



:)



:\ Where is the science behind this statement? POBI used modern Germans to claim that modern English are Anglo-Saxons. They might well be but this isn't science.



Isn't this all backwards (modern to ancient), or am I wrong again? And why did they think modelling Irish as Italians made any sense? (Rhetorical questions).

The modern population admixture didn't appear to lead to much in the end although it would seem that the Scottish paper K3 is a big improvement. These papers are valuable for the clustering and charts/graphs, but best of all was the K2 using anglo saxon and iron age briton components from the insular Celtic population structure and the genetic drift with icelandic gaels in the Scottish paper. Those are quite informative imo..

I feel I agree with you that we must be careful when interpreting these charts using modern or seemingly modern components