PDA

View Full Version : Understanding your admixture results



Kurd
09-12-2018, 04:17 AM
Folks,

I have a new article titled "How to interpret your ancestry admixture results" at http://www.eurasiandna.com/2018/08/18/how-to-interpret-your-ancestry-admixture-results/

https://i.imgur.com/1jVTUMG.jpg

It is a work in progress.

Kale
09-12-2018, 05:23 AM
Very informative. For part of the target audience, your admixture run might confuse them as well. 3% South-Asian in English and Czech?
When components don't encompass the variation of the samples run through afterwards, weird things happen.
That and it could reflect shared ancestry, rather than one-way flow, e.g. Iran_N into both Paniya and Central+West-Asians.
In the case of English & Czech, Iran_N and CHG might be close enough to conflate.

As always though, I admire your generativity in taking the time to explain a lot of these tough to understand topics.

Hanna
09-12-2018, 08:45 AM
Which gedmatch calculator do you think is the best for picking up East Asian admixture?
Asking the same for the Caucasus component.

Kurd
09-12-2018, 02:31 PM
Very informative. For part of the target audience, your admixture run might confuse them as well. 3% South-Asian in English and Czech?
When components don't encompass the variation of the samples run through afterwards, weird things happen.
That and it could reflect shared ancestry, rather than one-way flow, e.g. Iran_N into both Paniya and Central+West-Asians.
In the case of English & Czech, Iran_N and CHG might be close enough to conflate.

As always though, I admire your generativity in taking the time to explain a lot of these tough to understand topics.

Yes , that’s the implication of the 3%. It’s all about allele frequencies. For example, the Czech or English subject may have the A allele at position rs1234. From all the component references, Paniya may have the highest frequency of A at rs1234. Let’s say 10 out of 18 Paniya references have the A allele. This translates to approximately a frequency of 55% . However, if say some Caucasus population was added to make a K6, we May find that Caucasians have a frequency of 80% at position rs1234. In this case the A allele would be more associated with Caucasians and not Paniya

It’s all give or take sometimes to get something you may have to give up a little of something else

Kurd
09-12-2018, 02:35 PM
Which gedmatch calculator do you think is the best for picking up East Asian admixture?
Asking the same for the Caucasus component.

None if you are trying to gauge total east Asian gene flow post bronze age , because they are all loaded with West Eurasian references who are themselves East Asian admixed post bronze age . They also have too high of a ratio of non-references to references . They are fine however for general clustering of populations as you can see from the oracles however they do minimize intra-population variation

Kurd
09-14-2018, 11:57 AM
Users with ethnic backgrounds from Iran (all parts), Iraq, Turkey, or the Caucasus are welcome to post screenshots of their 23andMe results for comparison in the comments section of the article to see how similar they are to the Kurdish subject results posted there, since there don't appear to be many such results posted online.

To post an image in the comments section, users can paste either the "direct link" or "BBCode" of their image in the comments section.

Please ensure that the image you post is yours and not someone else's.

lukaszM
09-14-2018, 12:20 PM
Very interesting Kurd! One question, what you think of using Sardinians instead Basques for European component reference?