23andMe

As I continue to research my ancestry through DNA, using various tools and services, and gaining experience and perspective, my views of DNA services evolve. These are my thoughts about 23andMe at this time, nearly three years into my DNA research.

Pros:

(1) The biggest advantage from 23andMe is that they provide DNA-related health and trait reports, both interesting and potentially important. (23andMe is not authorized to provide medical information, but a 23andMe report would certainly be a good basis for seeking medical advice from your doctor.) They offer different analysis products, and I believe the least expensive does not include health reports, so make sure to order the level of analysis you want.

(2) 23andMe has a large number of DNA contributors. I have found many known relatives there and have identified many matches. Though my already well-developed family tree has made that easier, perhaps, than for others.

(3) 23andMe provides a list of DNA matches in common (ICW), as do the other services. They also indicate which of your common matches "triangulate", a much higher level of confidence that a match is related. On the ICW list are shown, too, the relationship of your principal match to you and to the ICW. (MyHeritage does this; Ancestry and GEDMatch do not.) Sometimes it is necessary to construct trees for your matches, a slow, labor-intensive process, and information about how some ICW are related to each other can help enormously in focusing on fewer possible branches.

(4) 23andMe uses a prominently displayed star next to each match in your primary list of DNA matches, that you can toggle on or off. This is very helpful for showing which matches you have placed in your tree. (Browsing through matches for matches to work on next, it's very helpful to easily see those already completed.)

(5) Ethnicity estimates seem as accurate as any, at least for my very homogenous ancestry. I think my best estimates come from my own family tree.

(6) 23andMe analysis includes the X-chromosome, which others do not. Since males inherit X chromosomes only from their mothers, a match on this chromosome can make tree research easier by eliminating some lines of ancestry. This has not led to identifying a match for me yet, but it is one more analysis tool.

(7) While they do not provide a detailed analysis of the Y-chromosome, they do identify a paternal haplogroup for males, which is a pattern found on this chromosome. Theoretically, this could be another tool to help connect to male relatives. In practice, I find it confusing because some haplogroups are closely related and a father and son may be identified with different haplogroups. If you know enough about haplogroups to recognize those that are closely related, perhaps this is not a problem. So I list this as a pro because it could be a useful tool, even though not yet for me.

(8) 23andMe has so far tolerated the use of third party tools, like 529andYou, to help gather DNA match information. (529andYou gathers lists of triangulations.) Though recently there has been more use of Captcha to, I assume, distinguish between people using data collection tools and robots.

(9) Though the lack of tree building is listed as a con below, the associated pro is that the user profile allows you to list birthplaces of your grandparents and family surnames and a link to a tree located elsewhere. This is enough information, often, to start a tree that can be continued by finding grandparents in census records (currently born before 1940).

(10) 23andMe allows you to download files for your raw DNA analysis, your matches and your shared DNA, which can be used for your own analysis offline.

(11) 23andMe seems to show matches down to about 7cM. (The bigger limitation is, perhaps, the number of matches displayed. I believe that MyHeritage and 23andMe limit the number of matches displayed, to about 8000 and 1000, respectively. When you have long-time American families, like my Dad's ancestry, these limits are reached before you reach the lower shared DNA threshhold, so there are not many matches shown below about 8cM. Ancestry's limit is, rather, the lower DNA match threshold, which they recently raised from 6cM to 8cM. All of these limitations are to avoid overwhelming [most] users with matches well beyond their interest and to reduce the load on their servers). So this could be a pro or a con.

Cons:

(1) 23andMe is not a genealogy service. There is no sister company with historical records, there is no family tree building feature. As part of your descriptive profile, you can list family surnames, your grandparents' birthplaces, and a link to your family tree. Personally, I don't need the paid access to historical records and have a public family tree with a link, so don't find this "con" a disadvantage. However, the lack of trees does make research quite a bit more difficult and I find my self searching for relatives more often on Ancestry and MyHeritage because of this.

(2) Managing or researching others' DNA can cause confusion. I have access to some DNA tests. Because I don't want to be seen as impersonating someone, when I send a message to a DNA match I explain that I am not their DNA match and what my relationship is to their match. Then I send a message, but since it's not my account I'm not sure how the message sender is shown. I usually include an outside e-mail address to make communication less cumbersome. Ancestry makes it easy to assign a management role to me. (Though I'm not sure how clearly communications are identified with them, either.)

Overall, I generally recommend this service, especially if you would like to get DNA-related health and trait reports.

My Genetic Genealogy: Pros and Cons of Too Many Matches

I've been working with DNA kits on 23andMe, MyHeritage, and AncestryDNA. One of my first observations was, after beginning with 23andMe and seeing about 1000 DNA matches, that MyHeritage's 3,ooo matches was ridiculous. Who will ever have time to go through and try to link 3,000 matches! 23andMe is now providing about 1200 matches and MyHeritage is now about 8000. Really? But now I've crunched some numbers and am having second thoughts.

The Beginning

Browsing through matches on 23andMe, I started exploring a not-too-distant match for my father, 0.95% shared DNA, about 70cM, somewhere near average for a 3rd cousin. Except that Dad is in his early nineties and the match was middle-aged, so the relationship is more likely to be a 2nd cousin twice removed. This indicates a common ancestor of Dad's great-grandparents who immigrated to the United States.

The Genealogies

Dad's match was able to provide me with his family genealogy back to the early 1800s in Ireland. There was no intersection with my tree, which also geos back this far. Knowing that there is a connection, through the DNA match, the genealogies indicated that the family connection would have to be one or more generations earlier than the 0.8%  shared DNA suggested. Something's not right.

Different Relatives in Common

Comparing notes, we realized that Dennis's list of Relatives in Common (persons that were DNA matches to both him and to Dad) was different from Dad's list. I've noticed this with others, but hadn't delved into the explanation. So, FYI. Both lists were about 35 persons long, but only about 5 persons were the same on both lists. I asked 23andMe for an explanation.

The Relatives in Common list is created by taking your list of DNA matches - about 1200 at 23andMe - and selecting from them those that also share at least 5cM of DNA with the match you are comparing to. To make this less abstract. Suppose Dad's match is Dennis. [In what follows, Dennis and Keith are made-up names.] Dennis has a list of 1200 DNA matches, one of which is Dad. When he clicks on Dad, he is presented with a list of about 35 Relatives in Common. This list is created by taking Dennis's 1200 matches and selecting those who share at least 5cM (this is a VERY small piece of DNA) with Dad. If I look at Dad's list of all DNA matches, the very last one shares 0.27% (about 20cM). Dad's list of Relatives in Common must be from his list of matches, all of which share at least 20cM of DNA with him. The only persons who who show up on both Dennis's and Dad's lists share at least 20cM of DNA with both of them (though I don't know exactly Dennis's threshold), only about 5 persons. Note that both lists are valid, but this explains why they are different.

Cousin Keith

Dennis mentioned that his first cousin, Keith, was on his Relatives in Common, though it was not on Dad's. It turns out that Keith shares about 0.15% DNA with Dad, so doesn't make Dad's list of 1200 matches, so doesn't show up on Dad's version of the Relatives in Common. The second thing to note is that two first cousins should share about the same amount of DNA with Dad, while Dennis and Keith share 0.95% and 0.15%, respectively. This is a reminder that there can be large variations in inherited DNA. One possibility is that Dennis and Keith are related to Dad through different relatives, but further research showed this to be nearly impossible. Comparing to the genealogy research we were studying earlier, though, Keith's shared DNA indicates a common ancestor one or two generations further back than our immigrant ancestor, which could fit our observations better. My current hypothesis is that cousin Keith shares a more normal amount of DNA for the relationship with Dad, while Dennis inherited an unusually long strand of DNA.

What Does This Mean?

In this case, I seem to have gotten lucky that Dennis had an unusually long inherited strand of DNA that moved him above Dad's match threshold of about 0.27%. If not, I would not have seen this connection to investigate. This is disappointing. Much of my known genealogy ends with immigrant ancestors who are great-grandparents to my parents (whose DNA I am working with). My findings with cousins Dennis and Keith leads me to believe it is unlikely I will find connections to earlier ancestors in their countries of origin through 23andMe. Remember that my initial thought had been 1200 DNA matches is more than enough to work with. Now I see that it is not enough for the pre-immigration connections I eventually hope to make.

Not Quite That Bad

So far, in two of my ancestral lines, I was able to connect with many matches through 23andMe whose common ancestor was a pre-immigration family. Fortunately, there are older participants from these "clans" whose relationship to Mom/Dad were 3rd cousin once removed. The average shared DNA for 3rd cousins once removed is about 0.4%, so above the 0.27% threshold for 23andMe matches. But it is important to seek connections with older matches (say, 60 and up). It remains to be seen whether this population will decrease, from natural causes, or increase as more people get their family elders tested.

What About Other DNA Services?

AncestryDNA: I don't know the numbers for Ancestry. I haven't found a way to harvest their matches, Ancestry does allow downloads of this information, and I ran out of patience scrolling endlessly through who knows how many matches to find the end.

MyHeritage identifies about 8,000 DNA matches, down to about 8cM. Perhaps overwhelming. Perhaps absurd. But it does seem to allow the possibility of connecting back further in time. Identifying the ancestral line going so far back from smaller DNA segments will, however, require lots of luck and lots of work.

[I've assumed a very simple relationship between shared DNA and relationship, while in reality, it is not simple. A simple relationship is easier to understand, and I think allows me to make my point.]