Log in

View Full Version : Eurogenes K13 SNP Count


Ion Basescul
03-11-2020, 08:13 PM
Here's an interesting thought.

You need to be conscious of how informative your autosomal DNA is for testing with GEDmatch and G25.

This is important to understanding your ancestral profile, because the more data/SNPs are utilised, the closer to the truth are the results that you get.
With a low SNP count you could get false highs or lows in some components and believe in them for years.


To do this, the easiest way would be by looking at Eurogenes K13, as it's by far the most popular test.

Go to GEDmatch -> Eurogenes K13 -> Wait for results -> Check the number at the bottom left side of the screen.

Here's an example of me and my parents. I tested on the old 23andme (2015), while my parents tested on MyHeritage (2020).

Me
https://i.imgur.com/Kf9uno8.png

Mom
https://i.imgur.com/dR0AN2w.png

Dad
https://i.imgur.com/HZKP60s.png


As you can see, they lack a lot of data (SNPs) compared to me.
The discrepancy is so big, that I can safely say that their results are probably informative to get the general picture, but not accurate to compare against mine.

168k vs 78k SNPs means that their genomes cover only 46% of the information that Eurogenes K13 gets from my autosomal data.
So while it's fun to play around, I should bear in mind that their autosomal data, which gets fed on GEDmatch and out of which G25 coordinates are created, has half the resolution that mine does.


By this example, I wanted to let you know that while your results should be kind of close to your ancestral makeup, the resolution on amateur tests/calculators/tools (aka the SNP count used for autosomal calculators/tests) is quite important.


Since their data has half the resolution of mine, this is why on G25 and GEDmatch calculators my parents pull away from me, creating the impression that we are not related.


https://i.imgur.com/TJVpQSS.png


On MyHeritage however, which probably works with the same number of SNPs (aka resolution) for both MyHeritage tests and 23andme uploads, I fall in-between them, as is expected of a child. I'm closer to my dad, but that's also because I inherited more DNA from him.


Me
https://i.imgur.com/ZnMKZ1n.png

Mom
https://i.imgur.com/pzpwupx.png

Dad
https://i.imgur.com/eruIWr6.png


https://i.imgur.com/mArTr75.png
Leto
03-11-2020, 08:28 PM
Now both FTDNA and MH provide about 77-78k SNPs.
Bosniensis
03-11-2020, 08:32 PM
On K13 I got "174287 SNPs used in this evaluation" quite high number
Ion Basescul
03-11-2020, 08:34 PM
On K13 I got "174287 SNPs used in this evaluation" quite high number

That's normal for older tests, which had a higher resolution and the data of which are thus more accurate.
Dušan
03-11-2020, 08:39 PM
174842 SNPs

https://i.imgur.com/CrMA6xv.jpg
Leto
03-11-2020, 08:39 PM
On K13 I got "174287 SNPs used in this evaluation" quite high number
Your V5 kit counts only 58k.
Bosniensis
03-11-2020, 08:40 PM
Your V5 kit counts only 58k.

Yes V5 is bad, but FTDNA is better.
Ion Basescul
03-11-2020, 08:43 PM
Your V5 kit counts only 58k.

Yes, that's why I didn't test my parents on v5, even though I would have preferred 23andme for phasing.
MyHeritage has 20k more SNPs and I can at least add my data there. It also appears to standardise my data as the results seem fair when compared against.
Nurzat
03-11-2020, 08:48 PM
I've tested with 23andme far back in 2011:

"182477 SNPs used in this evaluation"
Leto
03-11-2020, 08:49 PM
I've tested with 23andme far back in 2011:

"182477 SNPs used in this evaluation"
Back then it just said 100 percent hwite :lol:
Leto
03-11-2020, 08:50 PM
Yes, that's why I didn't test my parents on v5, even though I would have preferred 23andme for phasing.
MyHeritage has 20k more SNPs and I can at least add my data there. It also appears to standardise my data as the results seem fair when compared against.
Do you think V5 is so terribly bad? In many cases it works fine, I'd say.
Kaspias
03-11-2020, 08:55 PM
2016 MH: 174798 SNPs used in this evaluation
2018 FTDNA: 152005 SNPs used in this evaluation
2020 23andMe v5: 60995 SNPs used in this evaluation
2020 23andMe v3: 182231 SNPs used in this evaluation
Kyp
03-11-2020, 09:03 PM
Father(2019)
78217 SNPs used in this evaluation
Me(2019)
78205 SNPs used in this evaluation
Mother(2018)
174091 SNPs used in this evaluation


all with MH,

Always felt my results were off
Ion Basescul
03-11-2020, 09:05 PM
Do you think V5 is so terribly bad? In many cases it works fine, I'd say.

It's fine as the first test, but just as I've shown with MyHeritage it will be at an accuracy disadvantage when used for G25/GEDmatch.
JosephK
03-11-2020, 09:18 PM
I don't think SNP count has much to do with accuracy; it depends on what exactly the calculators are looking at. I have several people's kits who have both high count (AncestryDNA) and lower (MyHeritage, 23andMe) and the lower counts are more accurate in terms of Oracle relationship with true/known heritage.
gixajo
03-11-2020, 09:38 PM
MH 2019

78220 SNPs used in this evaluation in k13

78845 SNPs used in this evaluation in k15

72428 SNPs used in this evaluation in k36
Ion Basescul
03-11-2020, 09:51 PM
I don't think SNP count has much to do with accuracy; it depends on what exactly the calculators are looking at. I have several people's kits who have both high count (AncestryDNA) and lower (MyHeritage, 23andMe) and the lower counts are more accurate in terms of Oracle relationship with true/known heritage.

I agree that it depends on the SNPs that are being used, but for something like Eurogenes K13 it's obvious that for those who tested on the newer generation of chips, a lot of SNPs are not being counted. This should result in less accurate results, because otherwise what's the point of those extra SNPs?
Leto
03-12-2020, 05:25 PM
174842 SNPs

[img]https://i.imgur.com/CrMA6xv.jpg


Here's an interesting thought.
...

How many SNPs do you get on Dodecad K12b? Do screenshots like that please :)