Which is more accurate, the McDonald Analysis or Eurogenes?

[7eleven]

Which do you find to be more accurate, Dr. McDonald's analysis or Eurogenes?

[The Exiled King]

I like both really. You kind of have to pick and choose what Eurogenes test you like best though. I really like Dr. McDonald's chromosome painting, it seems that it goes back further than say 23andMe's Ancestry Composition.

If I had to choose one overall it would actually be a combination of Eurogenes and Dr. McDonald's Chromosome Painting since the percentages are about equal for both ie. Sub-Saharan African, Asian, etc.

[xajapa]

In my mind, they view your results from a different perspective. McDonald's graphs and maps tries to pinpoint your actual ancestry. Eurogenes gives a general ethnic composition (like northern European, West Asian, etc). I don't see it as trying to identify the exact place where your ancestors came from. So, after looking at McDonald's results, I found they really did capture my recent ancestral heritage. Eurogenes, on the other hand, gave a good picture of my more distant ancestry, but couldn't capture the places from where my most recent ancestors came, but, I don't think that is its intention.

[Vesuvian Sky]

IMO, what most of the 3rd party calculators claim they are detecting per their categories is rather questionable.

McDonald's BGA clearly presents all 23 chromosomes and their autosomal regional characteristics in the best most digestible of terms. Based on the geographic area, he then gives the populations your most like modally in 3 categories. I found his results compatible w/ my FTDNA output and reasonable. Actually expected it to come out the way it did per his methods.

3rd party calculators like Eurogenes use Oracles based on binary modal classes which tend to come out not so accurately particularly for the secondary population class but also at times the primary.

On the other hand, it can vary per person regarding accuracy but I'd say your McDonald's BGA will usually be the most accurate, meaningful BGA for all 23 chromosomes.

[Su]

McDonald's BGA sucks for Turks since the reference populations in Middle East are not accurate.

[Jackson]

IMO, what most of the 3rd party calculators claim they are detecting per their categories is rather questionable.

McDonald's BGA clearly presents all 23 chromosomes and their autosomal regional characteristics in the best most digestible of terms. Based on the geographic area, he then gives the populations your most like modally in 3 categories. I found his results compatible w/ my FTDNA output and reasonable. Actually expected it to come out the way it did per his methods.

3rd party calculators like Eurogenes use Oracles based on binary modal classes which tend to come out not so accurately particularly for the secondary population class but also at times the primary.

On the other hand, it can vary per person regarding accuracy but I'd say your McDonald's BGA will usually be the most accurate, meaningful BGA for all 23 chromosomes.

Depends what calculators you use though, some are really accurate, other's not very accurate. Partly this can be due to the user not being included in the data set of the initial run that splits it into components, but for many of the tests this isn't a problem.

Definitely i would use both McDonalds and the Eurogenes, Dodecad calculators. McDonald's chromosome painting is quite good, and the PCA is quite good also. But PCAs are a bit vague in some respects - However McDonald is very accurate, and it agrees with a couple, or a few of the calculators on GEDmatch, so those are the ones i use. May be different for other people though.

[Vesuvian Sky]

Depends what calculators you use though, some are really accurate, other's not very accurate. Partly this can be due to the user not being included in the data set of the initial run that splits it into components, but for many of the tests this isn't a problem.

Definitely i would use both McDonalds and the Eurogenes, Dodecad calculators. McDonald's chromosome painting is quite good, and the PCA is quite good also. But PCAs are a bit vague in some respects - However McDonald is very accurate, and it agrees with a couple, or a few of the calculators on GEDmatch, so those are the ones i use. May be different for other people though.

This is all very true. And to be fair to Eurogenes, at K9 its fairly tite. But beyond that, I often question how real a Volga-Ural component is in me, East European, West Central Asian...especially at the amounts suggested. Even asked him about the West Central Asian component/Baloch and he said its correlated somehow w/ the North Euro component. Its kind've like the S. Asian component to a degree and on McDonald's the only time I came up w/ something like this was during the second time I resubmited (very recently) and he felt it was a no-call and not very real. Also, on MDLP-K22, any S. Asian like component completely disappears. So I find its a matter of sketchy categorizations at times.

For Admix calculators though I like MDLP and Harrappa world. The Oracles for both these at the bi-modal level I felt were decent and categories fair but even some of the results I take with a grain of salt. Dodecad gave me the worst results for bi-modal population share.

[xajapa]

For Admix calculators though I like MDLP and Harrappa world. The Oracles for both these at the bi-modal level I felt were decent and categories fair but even some of the results I take with a grain of salt. Dodecad gave me the worst results for bi-modal population share.
Those various admix calculators do seem to work differently, depending upon your ethnic ancestry: Eurogenes and MDLP for northern European, Dodecad for southern European, Harappa for Near Eastern/ South Asian. Harappa wasn't very accurate for me, but then, neither was MDLP. As has been mentioned already, the populations used by the calculator determine the results. Some populations just don't serve as a good proxy.

[Jackson]

Those various admix calculators do seem to work differently, depending upon your ethnic ancestry: Eurogenes and MDLP for northern European, Dodecad for southern European, Harappa for Near Eastern/ South Asian. Harappa wasn't very accurate for me, but then, neither was MDLP. As has been mentioned already, the populations used by the calculator determine the results. Some populations just don't serve as a good proxy.

That is very true. The MDLP oracle is fundamentally flawed as it is affected by the so-called 'calculator effect' that i mentioned above, although it could be a really excellent oracle if that was fixed. The actual breakdowns in it are very good if you comparing to others not included in the run.

I find Dodecad K12b and Eurogenes EUtest to be the most accurate for northern and NW Europeans. I think this in part helps because of the Gedrosia component in K12b - That effectively orients you toward Britain if you have over 9% or so of that, as there is a local peak of it in Britain and Ireland, and the Eurogenes EUtest splits the NW European component that is the major component throughout NW Europe - Makes it much easier to tell apart people in the big NW Euro cluster, and it makes it a higher resolution image of northern Europe in particular, especially as it breaks up the eastern European component too, in to East Euro and South Baltic.

I find MDLP to consistently be the most inaccurate, Harappaworld is not particular accurate, and Dodecad V3 suffers from the 'calculator effect', Dodecad K7b is alright but not great.

[Vesuvian Sky]

Those various admix calculators do seem to work differently, depending upon your ethnic ancestry: Eurogenes and MDLP for northern European, Dodecad for southern European, Harappa for Near Eastern/ South Asian. Harappa wasn't very accurate for me, but then, neither was MDLP. As has been mentioned already, the populations used by the calculator determine the results. Some populations just don't serve as a good proxy.

Dodecad V3's chromosome percentages were fairly compatible w/ my McDonald's BGA but the bi-modal population share is what came out lacking much accuracy and certainly percision. But this is probably due to it being bi-modal rather then tri-modal and why I believe McDonald's methods are infinitely better for mixed modal results. A 3rd category or possibility is what's really needed for results to sort/class cleanly. Because of that, I find that its reliability even for a southerner (like your's truely) is not all there. And yes your results are only going to be as good as the reference population.

MDLP-22 I felt was pretty good for me and has very up-to-date reference populations. Percentage wise, for my chromosomes, the results were reasonable and even the bi-modal population share classed cleanly and gave me results that had compatibility w/ McDonalds.

Really though, these calculators can only improve, and they'll get better with improved reference populations.

[Ibericus]

They are both good. But McDonald for me was a bit boring, my chromosome painting was pretty much all red (euro) and he guessed me as 100% spanish.

[Prince Carlo]

Dodecad is better than both them.

[xajapa]

Dodecad V3 suffers from the 'calculator effect', Dodecad K7b is alright but not great.
Thanks. Now that you mention this Dodecad V3 is the the least accurate for me. Eurogenes EU is my most accurate.

[ZephyrousMandaru]

The third-party calculators and PCA/MDS Plots, and a variety of other tests are more accurate and informative than Doug McDonald's BGA. Which has been around since 2008. In terms of McDonald's PCA Plot, Davidski's Eurogenes SPA surpasses it in resolution and precision. McDonald's BGA is pretty decent, but it's also much more ambiguous. The estimates are fractions, based on populational affinity instead of actual ancestral clusters.

Compared to the Dodecad Ancestry Project, Eurogenes Genetic Ancestry Project and Harappa Ancestry Project. Dr. Doug McDonald's BGA is severely lacking in the reference samples department. I also highly suspect that Doug's BGA has a time stamp on it, whereas the Eurogenes SPA Plot doesn't.

According to the most recent genetic data, Assyrians plot in historical Babylon (Southeastern Iraq). Whereas McDonald plots me in very far Northern Iraq (on the periphery of Turkey).

[Tropico]

Idk. They are both pretty accurate. They mirrored each other somewhat well. On Both I got a mix of Pred Spanish with some French, then Amerindian, North African and SSA.

[Hadouken]

the burgerking

[Black Wolf]

I find Eurogenes to be quite a bit better/more accurate than McDonald's analysis.

[Skywalker]

It all depends on the individual which one is best. I personally like Dodecad K7b the best for me. I get the lowest distance of 0.32 in the Oracle, and it has me as half Lithuanian and half Southern Italian_Sicilian which accounts for almost 1/2 of what I knew about my heritage growing up. I think Eurogenes is better for Northern/North Western Europeans, Dodecad is better for Southern Europeans, MDLP for Eastern and Northern Europeans, and HARRAPA if you have South Asian. Dr. McDonald was also the creator of Family Tree DNA's Population Finder. However they "tweaked" it from his original version. Personally I wish they would overall it already. Dr. McDonald has more Native American samples than most. MDLP has some, and Eurogenes is trying to get up to speed with that now with his new EU Test v2.