PDA

View Full Version : Reference populations for AncestryDNA



sktibo
10-10-2016, 10:32 PM
12078

I had given up on trying to find out where their population samples come from, until I came across this little image...

Does anyone have a larger version?

Thank you

Ron from PA
05-30-2017, 01:00 PM
Interesting looks like nothing in Northern Scotland. Can't see if they cover SW Scotland or not, which is where the Scot I do have would be located. Looks like it goes a little into SE Scotland maybe to Edinburgh.

NewAlbion
06-10-2017, 08:19 AM
Interesting looks like nothing in Northern Scotland. Can't see if they cover SW Scotland or not, which is where the Scot I do have would be located. Looks like it goes a little into SE Scotland maybe to Edinburgh.

Yeah, I wish they would cover Argyll-Bhute I have a feeling I have a strong genetic affinity for that part of Scotland.

sktibo
06-18-2017, 01:55 AM
16989

I wanted a closer look at where those littler markers were.

Cornella
06-18-2017, 01:07 PM
Yeah, I wish they would cover Argyll-Bhute I have a feeling I have a strong genetic affinity for that part of Scotland.

If they use the 1000 Genomes samples, these include 4 people with 4 grandparents born in Argyll and Bute.

sweuro
06-18-2017, 01:39 PM
If you want to see averages of many populations you can use the Excel spreadsheet (Dataset 3 ) here :

https://www.nature.com/articles/ncomms14238#supplementary-information

Apex n Harmony
08-04-2017, 06:00 PM
16989

I wanted a closer look at where those littler markers were.

Did you happen to find an image showing AncestryDNA's reference populations for other parts of the world?

sktibo
08-04-2017, 09:25 PM
Not yet no

JerryS.
08-04-2017, 10:02 PM
quite a few gaps for France, Scotland and the Balkans. even the Italian heel is underrepresented.

Nqp15hhu
08-04-2017, 10:53 PM
So there is a limited reference population for Scotland? Interesting.

sktibo
08-04-2017, 11:07 PM
So there is a limited reference population for Scotland? Interesting.

It looks like none of their references actually come from Scotland, if some do they're from the Southeast

JerryS.
08-04-2017, 11:19 PM
It looks like none of their references actually come from Scotland, if some do they're from the Southeast

didn't the northeast coast of Scotland have significant Scandinavian influences, and wouldn't that omition skew the population sample? same for Italy's heel and the Balkan/Albanian influences?

Nqp15hhu
08-04-2017, 11:34 PM
I didn't realise their sample was so small. I had thought it was pretty large as it did come out recently that they had several million samples, but I guess that isn't the same as a Reference Population.

Does this mean the Ancestry results aren't reliable?

sktibo
08-04-2017, 11:37 PM
didn't the northeast coast of Scotland have significant Scandinavian influences, and wouldn't that omition skew the population sample? same for Italy's heel and the Balkan/Albanian influences?

Well Orkney and Shetland have the largest Scandinavian influence, the average on the POBI Orcadian regions comes out to 26.83%. The Northwestern Islands of Scotland appear to have the second largest amount of Scandinavian influence - POBI's NW Scotland has an average of 21.5%. I imagine the Isle of Lewis would have more of this, there were three Lewis samples off the top of my head but most of the samples for this region were actually from Argyll and Islay. The Northeast, Aberdeenshire area, has less Scandinavian than this at an average of 16.1%.
So I suppose the Northeast coast of Scotland doesn't actually have that much Scandinavian influence compared to the Northwest coast and the Northern Isles / the Northern tip. The Western Isles / Hebrides along with the Northern Isles were the primary areas of Scandinavian influence in Scotland, so I suppose this makes sense historically speaking. I have no idea about Italy

mwauthy
08-22-2017, 05:06 PM
12078

I had given up on trying to find out where their population samples come from, until I came across this little image...

Does anyone have a larger version?

Thank you

It seems odd that they label France in the Europe West category when 2/3 of the country is lacking reference samples. It also explains why people in the southern half of France are getting such inflated Iberian scores.

Apex n Harmony
08-22-2017, 05:17 PM
Does anyone know how exactly a customer's DNA is compared to the reference panel? I've read the ethnicity white paper, but I still don't fully understand the details.

For example, the Caucasus region includes 58 samples (although we don't how many samples were taken from each country in the defined region, or whether at least one sample was taken from each country). Is each customer's DNA compared to each of the 58 individual samples from the Caucasus, or does AncestryDNA take the 58 samples and create a sort of composite Caucasus "profile" to which each customer's DNA is compared?

mwauthy
08-22-2017, 05:32 PM
I'm not sure but my guess is that they associate certain snps in the DNA with certain groups. Then they do a composite with the various reference samples.

MacUalraig
08-22-2017, 05:34 PM
Does anyone know how exactly a customer's DNA is compared to the reference panel? I've read the ethnicity white paper, but I still don't fully understand the details.

For example, the Caucasus region includes 58 samples (although we don't how many samples were taken from each country in the defined region, or whether at least one sample was taken from each country). Is each customer's DNA compared to each of the 58 individual samples from the Caucasus, or does AncestryDNA take the 58 samples and create a sort of composite Caucasus "profile" to which each customer's DNA is compared?

It is based on allele frequencies as described under section 4.
"The next step is to estimate a customer’s ethnicity based on the DNA of the reference set of samples, as well as the DNA of the customer. We assume that an individual’s DNA is a mixture of DNA from a set of “source” reference populations. In the example below, a sample gets each allele at each SNP from one of four “source” reference populations.
Figure 4.1: Example Customer Genotype. The table indicates the genotypes of the sample. The colors of the SNPs correspond to their locations of origin, indicated in the legend beside the table.
In order to make estimates of genetic ethnicity, we simply use what we know about the frequency of the alleles of those SNPs in the reference populations.

In this example, we are looking at the frequencies of the two alleles at the first SNP in each of the four reference populations. A’s are more common in Eastern Europeans and people from the Iberian Peninsula, A’s and G’s are equally likely in the Irish (from the Ireland region), and G’s are more likely in Central Asians. If a sample from an individual has two A’s, it seems that Eastern Europe and the Iberian Peninsula are the more likely places from which he or she may have gotten these two alleles. In our example, the individual has gotten 1 A from Eastern Europe, and 1 A from the Iberian Peninsula.
AncestryDNA uses similar reasoning to make our actual estimates, but with a rigorous statistical model that incorporates SNP data from across the genome."

blah blah

mwauthy
08-25-2017, 01:09 PM
It would be interesting to know which of those reference samples scored 100% representative of their region. For Iberian Peninsula I believe it's some Basque samples. I'm curious about the rest of the regions.

Apex n Harmony
08-25-2017, 08:11 PM
It would be interesting to know which of those reference samples scored 100% representative of their region. For Iberian Peninsula I believe it's some Basque samples. I'm curious about the rest of the regions.

My father got pretty close: 96% Caucasus. Three of his grandparents were from the Gilan Province in Iran (we think), and the fourth was from Tehran (we think). I don't think I've seen a higher Caucasus score on AncestryDNA. I've seen an Armenian score 66% and an Azerbaijani score 69%, and I've seen Iranians score anywhere from 60% to 96%.

lucperri
11-14-2017, 12:07 PM
Oh, Look! Calais has gone back to Britain again. Mary Tudor would be so proud.

lukaszM
11-14-2017, 06:56 PM
12078

I had given up on trying to find out where their population samples come from, until I came across this little image...

Does anyone have a larger version?

Thank you

I don't believe they have reference sample from all over Poland. It's simply impossible because there are only few Polish regional samples published in some studies, and each from limited area (not to mention Estonian Poles).

Dibran
11-14-2017, 07:03 PM
I don't believe they have reference sample from all over Poland. It's simply impossible because there are only few Polish regional samples published in some studies, and each from limited area (not to mention Estonian Poles).

I forget who mentioned it, but, supposedly the new breakdown is based on autosomal DNA relatives, and their place of ancestry. so for example 15% of my DNA relatives are from East Europe, which is the percentage I have for autosomal. Idk how accurate/inaccurate that is with regards inferring ethnic composition.

Geborgenheit
11-14-2017, 07:28 PM
Interesting references: "Eastern European" almost reaching Swiss Alps, "Finnic" in Latvia.

RobinBMc
11-14-2017, 09:33 PM
I forget who mentioned it, but, supposedly the new breakdown is based on autosomal DNA relatives, and their place of ancestry. so for example 15% of my DNA relatives are from East Europe, which is the percentage I have for autosomal. Idk how accurate/inaccurate that is with regards inferring ethnic composition.

That only applies to the Genetic Communities, not to your ethnicity percentages.

RobinBMc
11-14-2017, 09:39 PM
I found the original image from https://thegeneticgenealogist.com/2013/09/12/ancestrydna-launches-new-ethnicity-estimate/ - but I don't know if that's where the OP found it.

In Ancestry's white paper, they have this 19804 which seems to show a thumbnail image of it. Shame they won't make the full thing accessible. It looks like a "my google map" so if they made it public, it would be really interesting to explore.

https://www.ancestry.com/cs/dna-help/ethnicity/whitepaper

JerryS.
11-15-2017, 01:25 AM
Ancestry DNA results are light for Mediterranean people of mixed ethnicity. I have a cousin that is literally half Calabresi and she only got 30% Italian/Greek. another is an honest one quarter Abruzzi and only got 17% Italian/Greek.

mwauthy
11-15-2017, 03:01 AM
Itís odd that any reference sample would get 100% for a region or 0%. I wonder how many different alleles they consider native to a region?

RobinBMc
11-16-2017, 02:48 PM
Ancestry DNA results are light for Mediterranean people of mixed ethnicity. I have a cousin that is literally half Calabresi and she only got 30% Italian/Greek. another is an honest one quarter Abruzzi and only got 17% Italian/Greek.

If a Mediterranean person is of mixed ethnicity, why wouldn't you expect to see lower results for Italy/Greece and some from surrounding areas? Isn't that what being of mixed ethnicity means, that you're going to find results in more than just Italy/Greece?

JerryS.
11-16-2017, 03:40 PM
If a Mediterranean person is of mixed ethnicity, why wouldn't you expect to see lower results for Italy/Greece and some from surrounding areas? Isn't that what being of mixed ethnicity means, that you're going to find results in more than just Italy/Greece?

if you are half southern Italian as in your father just got off the boat and your mother is English and German, it would be reasonable to see a 50% Italy/Greece segment, not 30%.... and so on if we use conventional arithmetic.

geebee
11-16-2017, 04:11 PM
if you are half southern Italian as in your father just got off the boat and your mother is English and German, it would be reasonable to see a 50% Italy/Greece segment, not 30%.... and so on if we use conventional arithmetic.

You seem to be assuming that just because someone comes from the same region as the reference population, that person will automatically show up as 100% from that region. That isn't necessarily true. In fact, if you look at "Genetically Diversity in the Europe South Region" (https://www.ancestry.com/dna/origins/DD9853DB-6ADF-44BA-9DB9-4491BCC22C18/ethnicity/EuropeS/diversity), what you'll see is that the "typical native" of the region has 72% Europe South, not 100%.

If you do the math, then, you might expect someone who had one parent from the region not to have 50%, but more like 36%. So in reality, 30% Italy/Greece for a person who is "half Calebresi" may not be too far off. This is especially so when you consider the a purely "Calebresi" reference population, if there were one, would not be likely to be identical to the "Italy/Greece" reference population actually used by Ancestry.

If you divide 72% by 4, the result is 18% -- pretty close to the 17% scored by the "one quarter Abruzzi", wouldn't you say? This is using conventional arithmetic. It just isn't beginning with the erroneous idea that everyone from the South Europe region has to score 100% South Europe. Some do; but some also score considerably less than the 72% I referred to at the beginning.

JerryS.
11-16-2017, 04:20 PM
You seem to be assuming that just because someone comes from the same region as the reference population, that person will automatically show up as 100% from that region. That isn't necessarily true. In fact, if you look at "Genetically Diversity in the Europe South Region" (https://www.ancestry.com/dna/origins/DD9853DB-6ADF-44BA-9DB9-4491BCC22C18/ethnicity/EuropeS/diversity), what you'll see is that the "typical native" of the region has 72% Europe South, not 100%.

If you do the math, then, you might expect someone who had one parent from the region not to have 50%, but more like 36%. So in reality, 30% Italy/Greece for a person who is "half Calebresi" may not be too far off. This is especially so when you consider the a purely "Calebresi" reference population, if there were one, would not be likely to be identical to the "Italy/Greece" reference population actually used by Ancestry.

If you divide 72% by 4, the result is 18% -- pretty close to the 17% scored by the "one quarter Abruzzi", wouldn't you say? This is using conventional arithmetic. It just isn't beginning with the erroneous idea that everyone from the South Europe region has to score 100% South Europe. Some do; but some also score considerably less than the 72% I referred to at the beginning.

thank you for explaining it this way. this makes a lot more sense. It was rumored that my Calabresi family came from Albania long ago, maybe some other Balkan country.... the story was the old man spoke three languages, Italian (Calabresi dialetto), English (broken but passable) and something else that no one else understood.

anyway, your explanation clears up the percentages quite well. thank you.

JerryS.
11-16-2017, 04:26 PM
You seem to be assuming that just because someone comes from the same region as the reference population, that person will automatically show up as 100% from that region. That isn't necessarily true. In fact, if you look at "Genetically Diversity in the Europe South Region" (https://www.ancestry.com/dna/origins/DD9853DB-6ADF-44BA-9DB9-4491BCC22C18/ethnicity/EuropeS/diversity), what you'll see is that the "typical native" of the region has 72% Europe South, not 100%.

If you do the math, then, you might expect someone who had one parent from the region not to have 50%, but more like 36%. So in reality, 30% Italy/Greece for a person who is "half Calebresi" may not be too far off. This is especially so when you consider the a purely "Calebresi" reference population, if there were one, would not be likely to be identical to the "Italy/Greece" reference population actually used by Ancestry.

If you divide 72% by 4, the result is 18% -- pretty close to the 17% scored by the "one quarter Abruzzi", wouldn't you say? This is using conventional arithmetic. It just isn't beginning with the erroneous idea that everyone from the South Europe region has to score 100% South Europe. Some do; but some also score considerably less than the 72% I referred to at the beginning.

thanks again, but I cannot see the reference map you posted, it requires membership to do so, mine has lapsed.

Dibran
11-16-2017, 04:41 PM
That only applies to the Genetic Communities, not to your ethnicity percentages.

supposedly the new breakdowns are genetic communities.

ollie444
11-16-2017, 05:53 PM
supposedly the new breakdowns are genetic communities.

Not the percentages, they're still the dodgy ethnicity guesses! Sub-regions/GCs seem to have little to do with the main regions and their percentages really.

mwauthy
11-16-2017, 08:35 PM
I donít understand why someone from Corsica and someone from Northern Germany would be considered from the same region. No wonder some ďnativesĒ receive 0% for their region. Iíve said this before but Ancestryís Europe West category has problems especially for French people or people with French ancestry. My mother who is 100% French Canadian received less than 1% Europe West. That means her French ancestors that settled the St. Lawrence River valley came from Great Britain or Ireland for her genetic community?

geebee
11-16-2017, 09:05 PM
thanks again, but I cannot see the reference map you posted, it requires membership to do so, mine has lapsed.

Maybe this will help.
19823

JerryS.
11-16-2017, 11:00 PM
I don’t understand why someone from Corsica and someone from Northern Germany would be considered from the same region. No wonder some “natives” receive 0% for their region. I’ve said this before but Ancestry’s Europe West category has problems especially for French people or people with French ancestry. My mother who is 100% French Canadian received less than 1% Europe West. That means her French ancestors that settled the St. Lawrence River valley came from Great Britain or Ireland for her genetic community?

It may seem trivial, but Corsica was Italian before the Napoleon wars, but yes, it would seem odd that a Corsican would have similar DNA to a North German. Regarding your wife's ancestors.... could it be that the French settlers mixed with the British descendants so much that the French got almost washed out?

mwauthy
11-16-2017, 11:58 PM
It may seem trivial, but Corsica was Italian before the Napoleon wars, but yes, it would seem odd that a Corsican would have similar DNA to a North German. Regarding your wife's ancestors.... could it be that the French settlers mixed with the British descendants so much that the French got almost washed out?

On paper all of my momís French Canadian Ancestors have French surnames going back to 1650. Itís not just my mom either but many French and French Canadians get Low Europe West percentages so itís most certainly not a nonpaternal event thatís causing these discrepancies.

If an ex Italian territory with a current French nationality is part of the same region as north Germans then it probably means the region is too large and needs division and refinement.

RobinBMc
11-18-2017, 01:40 AM
supposedly the new breakdowns are genetic communities.

They aren't new - they're same Genetic Communities as before, they have just be re-organized to be listed as sub-groups of the ethnicity percentages. But they are calculated differently than the ethnicity percentages. What you described earlier is how Genetic Communities are established, not the ethnicity percentages.

RobinBMc
11-18-2017, 01:50 AM
if you are half southern Italian as in your father just got off the boat and your mother is English and German, it would be reasonable to see a 50% Italy/Greece segment, not 30%.... and so on if we use conventional arithmetic.

I wish that were always true but unfortunately even people native to Italy, with their entire ancestry in Italy as far back as they can research, get results in neighboring regions. So even native Italians will not necessarily get 100% in Italy or South Europe because they are too mixed to begin with and share too much DNA with neighboring regions (and it's not just Italians - this applies to nearly everyone in Europe except maybe Ashkenazi).

My dad is half southern Italian - his mother was born in America but her entire ancestry is southern Italian as far back as I can trace (1700s for most branches). He gets 44% Europe South from AncestryDNA, 8% Southeast Europe from FTDNA (22% from Asia Minor), 23.7% Italian from MyHeritage (45.8% Greek). Ancestry is by far the most accurate but still not exactly 50%.

JerryS.
11-18-2017, 04:01 PM
I wish that were always true but unfortunately even people native to Italy, with their entire ancestry in Italy as far back as they can research, get results in neighboring regions. So even native Italians will not necessarily get 100% in Italy or South Europe because they are too mixed to begin with and share too much DNA with neighboring regions (and it's not just Italians - this applies to nearly everyone in Europe except maybe Ashkenazi).

My dad is half southern Italian - his mother was born in America but her entire ancestry is southern Italian as far back as I can trace (1700s for most branches). He gets 44% Europe South from AncestryDNA, 8% Southeast Europe from FTDNA (22% from Asia Minor), 23.7% Italian from MyHeritage (45.8% Greek). Ancestry is by far the most accurate but still not exactly 50%.

yes, another member explained this to me a few posts back, makes sense to me now. but thank you just the same.