Mega QAPDM Models thread

[drb234]

I've been experimenting with those rights.

Attachment 147272

Attachment 147273

Standard errors aren't that great, but I wonder if the p value would cause some to ignore it.


And then I tried just using 2 sources:


Attachment 147274

Attachment 147275

Slightly better with just 2 sources. If one shoots for Z scores 3 or close to it.

Still experimenting. This is what I have for now.

low standard errors > p value (unless p val is below 0.1/0.05ish)
i dont think that set of rights would work well for ancestry/23andme files. they're pretty low coverage so they're difficult to model accurately as a result

[drb234]

delete

[drb234]

modeled Beethoven (who was half flemish / half rhenish german)

[Gannicus]

low standard errors > p value (unless p val is below 0.1/0.05ish)
i dont think that set of rights would work well for ancestry/23andme files. they're pretty low coverage so they're difficult to model accurately as a result

On Illustrative for qpAdm I'm using just my DNA file from MyHeritage.com.

I had to add a few more references to get those SEs down a bit more.

[Birchy]

Austria IA La Tene + Norway IA

[Gannicus]

In the spirit of St Patrick's Day, I modeled myself as 100% medieval Irish.



References:







^ Austria IA LaTene + Norway IA to compare to Birchy

I used Drb's rights from a couple posts ago.

[Bulkster]

In the spirit of St Patrick's Day, I modeled myself as 100% medieval Irish.



References:

[Birchy]

Denmark IA + Scotland IA (with DRB’s refs)

[Gannicus]

Modeled myself as 100% VK202. The dataset on IllustrativeDNA has him labeled as: Pict (Orkney).



Rights:





I also match him on 23&Me's Historical matches




I don't know if he's 100% Pictish ancestry or mostly with some Scandinavian. From what I was able to look up he is majority Pictish admixture.

[Gannicus]

France Yonne IA + Norway IA with Drb's newest rights:



Z scores would be over 3 with this model.


Modified Drb rights model: