Defaulter1831 Registered User
#31

pinkypinky said:
It's just as they add more and more people to the database, they refine the markers tested and what they mean.


More accurate i hope

1 person has thanked this post
Jellybaby1 Registered User
#32

It seems I'm no longer Polynesian!!

1 person has thanked this post
dubhthach Registered User
#33

Defaulter1831 said:
In 2016 i was:

85% Irish
5% Scottish
5% Northwest Europe (Normandy mostly)
5% Scandinavian/Viking

Now September 2018 i'm
98% Irish/Scottish
2% British/North Western Europe

Some posters here seem really erudite on the subject, i'm wondering what you think of, what seem to me to be, major changes. Where has my Scandinavian heritage gone.. is it in the Scottish and British/Western European dna?

Thanks very much.


Well you have to remember that the estimates are basically 'software calculators' that use 'reference samples' to infer percentages. In such a case the model is only as good as the data that is fed into it.

so to take an extreme case, if their calculator only had three sample populations:
1. Congolese
2. Japanese
3. Swedish

Then in such a scenario an Irish person data fed into it would probably show up as 100% 'Swedish'.

So obviously the first step is to try and build a sampleset from every and any possible population group.

Of course if you have small samplesets you run into another "issue", in that neighbouring populations tend to be more closely related to each other then distant ones. So for example your average Irish person and average Swede will have shared more recent common ancestors with each other then either would with someone with origins in say Mongolia. Likewise someone from Mongolia will have more recent shared ancestry with say someone from Northern China or Korea then they do with either Swede or Irish person.

In such a case you need to have relatively large samplesets for each population. This allows you to
(a) generate a better picture of what average person in a population looks like when it comes to genetic variation
(b) push certain components backwards in time to more ancient admixture

b. above is possibly what happened with Scandinavian bit in your case. You have to remember that tests such as Ancestry been autosomal are really looking for matches in last 200 years.

Now there's some research that points that modern Irish people appear to have a certain level of 'Norse' admixture in them. RCSI proposed this in their paper on the 'Irish DNA Atlas'. I'm not so sure on their methodolgy for number of reasons:
1. They use only modern sample populations from both Ireland and Europe
2. They don't have any ancient DNA from pre-viking Early christian period -- as a result no baseline for level of admixture flow
3. They have no ancient DNA from Scandinavia to provide a baseline for 'Viking period' population structure. (eg. Nordic Iron age/Early Viking period timeline)


Anyways to go back linguistic point of view it's worth pointing out that both speakers of Germanic languages (eg. English, Dutch, German, Swedish, Norwegian, Danish etc.) and Celtic language (Irish, Scottish Gáidhlig, Welsh, Breton etc.) would have shared a common language on order of 4-5k years ago in Proto-Indo-european.

4-5k years isn't a huge period for genetic differenation to build up. In comparison the spilt say between Western Europeans and East Asians is more on order of 30-45k years ago. (leaving aside that there have been levels of admixture back and forth since on smaller scale)

6 people have thanked this post
Defaulter1831 Registered User
#34

dubhthach said:
Well you have to remember that the estimates are basically 'software calculators' that use 'reference samples' to infer percentages. In such a case the model is only as good as the data that is fed into it.

so to take an extreme case, if their calculator only had three sample populations:
1. Congolese
2. Japanese
3. Swedish

Then in such a scenario an Irish person data fed into it would probably show up as 100% 'Swedish'.

So obviously the first step is to try and build a sampleset from every and any possible population group.

Of course if you have small samplesets you run into another "issue", in that neighbouring populations tend to be more closely related to each other then distant ones. So for example your average Irish person and average Swede will have shared more recent common ancestors with each other then either would with someone with origins in say Mongolia. Likewise someone from Mongolia will have more recent shared ancestry with say someone from Northern China or Korea then they do with either Swede or Irish person.

In such a case you need to have relatively large samplesets for each population. This allows you to
(a) generate a better picture of what average person in a population looks like when it comes to genetic variation
(b) push certain components backwards in time to more ancient admixture

b. above is possibly what happened with Scandinavian bit in your case. You have to remember that tests such as Ancestry been autosomal are really looking for matches in last 200 years.

Now there's some research that points that modern Irish people appear to have a certain level of 'Norse' admixture in them. RCSI proposed this in their paper on the 'Irish DNA Atlas'. I'm not so sure on their methodolgy for number of reasons:
1. They use only modern sample populations from both Ireland and Europe
2. They don't have any ancient DNA from pre-viking Early christian period -- as a result no baseline for level of admixture flow
3. They have no ancient DNA from Scandinavia to provide a baseline for 'Viking period' population structure. (eg. Nordic Iron age/Early Viking period timeline)


Anyways to go back linguistic point of view it's worth pointing out that both speakers of Germanic languages (eg. English, Dutch, German, Swedish, Norwegian, Danish etc.) and Celtic language (Irish, Scottish Gáidhlig, Welsh, Breton etc.) would have shared a common language on order of 4-5k years ago in Proto-Indo-european.

4-5k years isn't a huge period for genetic differenation to build up. In comparison the spilt say between Western Europeans and East Asians is more on order of 30-45k years ago. (leaving aside that there have been levels of admixture back and forth since on smaller scale)


Thanks very much for this excellent post.

1 person has thanked this post
tabbey Registered User
#35

Jellybaby1 said:
It seems I'm no longer Polynesian!!


Don't be too disappointed, you can still take a holiday in Polynesia if you wish (and resources permit)

1 person has thanked this post
pedroeibar1 Registered User
#36

tabbey said:
Don't be too disappointed, you can still take a holiday in Polynesia if you wish (and resources permit)

Perhaps it was not a good idea to wear a Bikini when spitting into the tube?

dubhthach Registered User
#37

Here's a blog post from Ancestry talking about their reference populations and the size of them:
https://www.ancestry.com/cs/dna-help/ethnicity/reference-panel

Key bits for us:


Region Number of Samples
Northern Africa 41
Africa South-Central Hunter-Gatherers 34
Benin & Togo 224
Cameroon, Congo, & Southern Bantu Peoples 579
Ivory Coast & Ghana 124
Eastern Africa 82
Mali 169
Nigeria 111
Senegal 31
Native American—North, Central, South 146
Native American—Andean 63
Central & Northern Asia 186
Southern Asia 600
Balochistan 53
Burusho 23
China 620
Southeast Asia–Dai (Thai) 80
Western & Central India 65
Japan 592
Korea & Northern China 261
Philippines 538
Southeast Asia–Vietnam 159
England, Wales & Northwestern Europe 1,519
Baltic States 194
Basque 22
Ireland & Scotland 500
European Jewish 200
France 1,407
Germanic Europe 2,072
Greece & the Balkans 242
Italy 1,000
Norway 367
Portugal 404
Sardinia 30
Eastern Europe & Russia 1,959
Spain 270
Sweden 372
Finland 361
Middle East 271
Iran / Persia 459
Turkey & the Caucasus 101
Melanesia 49
Polynesia 58
Total 16,638


I bolded some there, but you can see if you follow an arc from Portugal northwards to Norway you see decent size samplesets for the following: Portugal, Spain, France, Ireland ⁊ Scotland, England ⁊ Wales ⁊ NW Europe, Germany, Sweden, Norway

From a grandscheme of things these are our surrounding populations so if you have any non 'Ireland ⁊ Scotland' ancestry you would expect perhaps one of these (particulary the 'England, Wales ⁊ NW Euro' one).

Eventually they will probably spilt Ireland and Scotland perhaps when they have at least 500+ samples in reference panel from each country.

1 person has thanked this post

Want to share your thoughts?

Login here to discuss!