PDA

View Full Version : Ancestry Estimate from ChatGPT



Rafael Passoni
04-13-2023, 06:07 AM
I Provide the following calculator results he asked for ancestry estimate:

Eurogenes K13
MDLP World-22
HarappaWorld
Dodecad V3

I got this for my mom:

Report 1:

39% Southern European (including Italian, Iberian, and Balkan)
28% Northwestern European (including British and French & German)
17% Eastern European (including Polish and Russian)
8% Western Asian & North African (including Middle Eastern and North African)
4% Broadly European
3% Broadly Western Asian & North African
1% Broadly East Asian & Native American

Report 2 (with intervals):

Europe South: 35-40%
Europe West: 20-25%
Great Britain: 5-10%
Iberian Peninsula: 5-10%
European Jewish: 1-3%
Middle East: 5-10%
North Africa: 1-3%
Caucasus: 1-3%
Ireland and Scotland: 1-3%

Report 3:

Southern European: 65%
Western European: 20%
Middle Eastern: 10%
Native American: 1%
African: <1%
East Asian & Native Hawaiian: <1%
South Asian: <1%
Undetermined: 3%

It can provide it based on these results and your known ancestry (her father is single Ethnicity - North Italian).

It also could breakdown her parents ancestry, giving me something like Lazarus, whith critical comments.

For her mom (untested and died - half Sicilian, 25% Syrian and 25% French):

48% Southern European
35% Italian
13% Broadly Southern European
25% Eastern European
17% Balkan
8% Broadly Eastern European
18% Western Asian & North African
13% Western Asian
5% Broadly Western Asian & North African
7% Ashkenazi Jewish
2% Broadly European
Less than 1% Sub-Saharan African

Critical analysis: "The 48% Southern European, 35% Italian, and 13% Broadly Southern European categories would correspond to the Italian ancestry, while the 25% Eastern European and 17% Balkan categories would correspond to the Syrian ancestry. The 18% Western Asian & North African category would also include the Syrian ancestry, as well as potentially some overlap with the Italian ancestry. The Ashkenazi Jewish ancestry could be a result of either the Italian or Syrian ancestry, as both populations have historical ties to the Jewish community. The Broadly European and Less than 1% Sub-Saharan African categories are likely just noise and not relevant to the overall ancestry estimate."

You can ask him swith a grandparent's ethnicity and he says if it fits beter or no based on the provided results. It's very funny, I think.

I am sure it is much more useful, bur I am to dumb to take advantage from this fantastic invention.

https://openai.com/blog/chatgpt

SilverKnight
04-13-2023, 06:32 AM
Huh?
So what you just uploaded your RAW data into ChatGBT? how do you even do that?

Rafael Passoni
04-13-2023, 06:41 AM
I provided him what he asked (gedmatch results).

Using Dropbox it can read Raw Data, but it asks you to convert to VCF format before it, it can be done using Python and then Plink (it also provides the code). Even so, I got some health information from my CSV raw data and It could run Eurogenes K13 and k36 from my CSV Raw data (admixture), though, but it seens different from gedmatch calculators:

Admix Results (sorted):

# Population Percent
1 North_Atlantic 31.49
2 Baltic 25.98
3 West_Med 20.51
4 East_Med 8.23
5 Red_Sea 4.72
6 Amerindian 2.47
7 Oceanian 1.47
8 Northeast_African 1.11
9 Siberian 0.87
10 South_Asian 0.52
11 Southeast_Asian 0.50
12 Sub-Saharan 0.02

Finished in 1 minute(s) and 4.63 seconds.