Prev
First off, that didn’t take very long! At RootsTech, just over one month ago, I took a DNA test at the MyHeritage booth. I didn’t have to mail it back. Instead the MyHeritage people delivered all the samples they collected at RootsTech directly to the lab in Texas.
A couple of weeks before RootsTech, I was selected to receive a free MyHeritage DNA test kit at RootsTech. I wasn’t going to turn down that opportunity. Previously, I had tested my uncle and myself at FamilyTreeDNA. I was interested in comparing the results and seeing what MyHeritage, the new DNA kid on the block, was going to provide.
The first email I got was yesterday morning. 
Clicking on the View DNA results link led me to their site
So this was going to give me my ethnicity make-up.
I know the ethnicity results from the various DNA companies vary. Each is based on a base of several thousand people they use, and the assumption is that the people know accurately their ancestral origins. This is tricky for them to do, because if one of the people is 50% Swedish and 50% Mexican, then anyone who matches this person will get a little of both when they really are only related on one side. Obviously, the companies will not include this extreme case in their base, but the illustration is accurate, because the same thing can happen 2 or 3 or 4 generations back, and therefore allocate a significant percentage of inaccurate ethnicity to a person’s whole.
The second inaccuracy in ethnicity percentage is because a person does not get the same amount of DNA at each ancestral level from each ancestor. For example, the normal case is having 32 great-great-great-grandparents. Therefore, each should average just over 3% of your ethnicity makeup. So if two of your g3-grandparents were from Sweden, you’d expect that 6% of your ethnicity report would be from Sweden. But DNA does not pass down evenly. The amount of DNA passed down from each g3-grandparent can vary greatly. You might not get any from some and could get as much as 6% or even 8% from others.
In my case, none of that should be a problem. I am a good test case for how good the base is, since as far as I know, anything less than 100% Ashkenazi for me is likely incorrect. All my lines as far back as I can go don’t indicate anything otherwise. Even my Ancestral Birthplace Chart is boring, with my father’s side all Romania, and my mother’s side the country right next door: Ukraine.
So let’s see the results at MyHeritage:
Hmm. They got 83.8% right. The East Europe 3.8% could be argued that they got the locale right but the group wrong. I really have to laugh at the 1.0% Eskimo/Inuit though. Maybe that’s a result of my living through the frigid winters in Winnipeg all my life, that my genes have evolved into Eskimo.
Let’s compare to my FamilyTreeDNA ethnicity estimates.
They only got 79% right with an 11% locale correct for Eastern Europe. Wonder where they got the 2% British Isles from. And my total is just 99%.
My uncle was only tested at FamilyTree DNA and he is 100% from Romania. He came out to 89% Ashkenazi Diaspora, 2% Eastern Europe and 8% Eastern Middle East. Well, like my total, that also totals only 99%. And how does my uncle get Eastern Middle East but I get Asia Minor?
To me, the ethnicity results provide me with no information (although maybe I’ll flaunt my being part-Eskimo). What’s really important to me are the matches.
At FamilyTreeDNA, I currently match to 9,637 people. The high number is likely because I match to most of the people of Ashkenazi heritage who have tested there due to the great amount of endogamy in this population. If the Ashkenazi could map everyone just like the Icelandic people did, we’d be able to use an app like they’ve got to determine how we’re related. Unfortunately, our records don’t go back to 1000 A.D. and to make things more difficult, our people were one of the last to adopt surnames and that happened in the early 1800’s, only about 5 generations ago. So of those 9,637 people, I only have confirmed relationships of two: my uncle and a 3rd cousin.
MyHeritage DNA is a new DNA testing company. They only started up last year but with the enormous reach and large worldwide membership of their MyHeritage site, they are growing quickly. I was interested to see how many matches I had. That number initially turned out to be 260. They are shown for me on 26 pages, 10 per page, in order of decreasing shared DNA. My first three entries look like this:
They provide the name of the person, sometimes a picture of the person, their approximate age (I like that), where they are from, the possible relationship range, a percentage of shared DNA (I find that useless if the cM is given), the shared cM, the number of shared segments, the largest segment in cM, and the size of their tree at MyHeritage along with a link to their tree.
That is all very nice. MyHeritage is of course trying to use the DNA testing to get more people to use their services. This is a great initial step and they seem to be doing all the right things so far.
One odd thing in their relationships. I wonder why they state: “1st cousin twice removed”. I would sooner them state “2nd cousin” which is the same genetic distance. It is more likely your match is at the same generational level to you than for them to be 2 generations before or after you.
The big question is whether the MyHeritageDNA match information is compatible with the match data from other services. MyHeritage and FamilyTreeDNA use the same company in Texas to analyze their DNA tests. You would think the test results should be similar.
I did find a few of my matches who tested at both companies. Here’s the comparison:
MyHeritage is too optimistic about the Possible Relation. With endogamy, the relationships should be lessened at least to what FamilyTreeDNA has.
My Heritage Total cM is less than FamilyTreeDNA’s. That is okay. All that means is that FamilyTreeDNA is including smaller segments than MyHeritage. FamilyTreeDNA includes segments as small as 1 cM in their total. MyHeritage likely only goes down to, say, 3 cM or 5 cM.
But it’s the largest cM that bothers me. For this the two companies should have the same values, but don’t. And they’re not out by a small amount either. MyHeritage’s largest segment in all cases are larger than FamilyTreeDNA’s. I have no explanation for this, but it is indicative that the two sets of analysis have something significant that is different between them.
What MyHeritageDNA haven’t done yet, and it remains to be seen if they do, or if they hold out like AncestryDNA, is whether they provide you the ability to download your match data. Currently, if you want a list of the people you match with, you’ll have to go through your pages and record the info yourself, one by one. Nor is the segment match data supplied. As a result, I cannot check the individual matches to see why they differ from FamilyTreeDNA.
MyHeritage does allow you to download your raw data, and you can import that into GEDmatch. So currently, the only way you can use your MyHeritage data with Double Match Triangulator is through GEDmatch.
None-the-less, it’s a good start for MyHeritage. They’ll grow quickly and likely join the big-3: AncestryDNA, 23andMe and FamilyTreeDNA as the 4th major player in the DNA-testing circuit. I hope they make the decision to implement some DNA analysis tools and allow you to download your own data. And lets also hope that they don’t become one of those companies that sells your data to others, and hide that in their terms of agreement.
Now, what can I do to find my Eskimo relatives?
Followup: March 13, 2017: Ann Turner and Annemieke van der Vegt pointed out on the ISOGG Facebook group that the raw data can be downloaded from MyHeritage. There are 3 little dots that you can click on and the download option will appear. The raw data then can be uploaded to GEDmatch. I’ve updated my post to reflect this info.
Update: April 4, 2017: FamilyTreeDNA did a major update to their Ethnic Makeup algorithm. Many people have said it is much improved. It is as well for me, with more Ashkenazi accounted for.
Here’s my new results:
Ashkenazi up from 79% to 92%. British Isles and Asia Minor are gone. I don’t believe the 7% West and Central Europe. It should be East Europe. The trace of West Africa is new and perplexing.
My uncle improved as well. He went from 89% Ashkenazi to 96%, with trace amounts from Southeast Europe (okay), West Middle East (maybe), South Central Africa (huh?) and Central Asia (nope). You think maybe my uncle’s Central Asia is where my 1% Eskimo at MyHeritage came from?
Update: May 31, 2017: I transferred my FamilyTreeDNA raw data over to MyHeritageDNA a couple of months ago to see how it would compare. Both my tests were done on the same person, i.e. me, and analyzed at the same lab, since MyHeritageDNA and FamilyTreeDNA both use the same lab in Texas.
I would have expected my ethnicity as analyzed by MyHeritageDNA, whether from the MyHeritageDNA kit or from the FamilyTreeDNA Kit to be very similar. Here’s how they look:
I would only consider the Ashkenazi, East European and Balkan to be correct. The discrepancies in each line between the two tests are surprising. In one test, MyHeritageDNA says I have Iberian, North African and Eskimo. In the other it says I have NorthWest Europe and Irish/Scottish/Welsh.
Identical siblings should not give different ethnicity results, and neither should raw data from the same person analyzed for ethnicity by one company.
Some people comment that “ethnicity estimates are not an exact science”.
Well I would say it’s more like “an approximate art”.
RSS
Trackback
Permalink