Whenever a new study is published linking a specific gene or SNP to a particular trait or health risk I go rushing to my DNA results to see what I have, so I thought I would share how I do that.
My latest worry is my coffee drinking habit.
There are so many coffee and caffeine studies that I am totally confused about what I should be doing. One caveat is that most of these studies are far from comprehensive and all of them need to be taken with a grain of salt. Have they really factored out all the other possibilities that could cause this result? Let’s face it, it’s early days yet in understanding what specific genetic variants do.
Personally when I read one of these articles, the first thing I do is google that gene name to find the associated SNP’s “rs” number so that I can find that variant in my results. For example let’s look at rs762551, a SNP involved in the metabolizing of coffee.
While it is easy to look up a gene or SNP if you have tested at 23andMe or GENOS by using their browse raw data functions, what if you have only tested at Ancestry.com? Or Family Tree DNA, which has deliberately chosen not to test health related SNPs?
It’s actually pretty easy to look up a SNP in your raw data if you have downloaded it.
Next Friday I am doing a workshop at the SCGS Jamboree in Burbank on using the tools at DNAGedcom and GEDmatch to solve unknown parentage cases. Since I like to have slides that are screenshots for my attendees, I decided to grab a few images using the latest and greatest GWorks on a missing father search case whose maternal half brother’s results were just in.
The basic technique for finding an unknown parent is to search the trees of close DNA matches looking for an ancestral couple shared among many of them. Build the tree down from that couple until someone is in the right place at the right time. The more you know about the unknown person(s) the easier this is. See the top of this post of mine for a summary – http://blog.kittycooper.com/2017/01/a-jewish-adoptee-finds-his-birth-family/
It has been a while since I used DNAGedcom. Why? Because now that Ancestry.com DNA has over 4 million testers, many adoptees get lucky and as soon as their DNA results are posted, they have enough good matches to figure out who at least one birth parent is. Also the use of a mirror tree with a second cousin’s information can often identify the family branch they are looking for (see http://www.borninneworleans.com/how-to/what-is-a-mirror-tree/ for that technique).
However when there are only third and fourth cousin matches, the remarkable tools at DNAGedcom.com can help you solve your mystery, but it is not intuitive or easy. It works best for folk with deep American roots, since that is the most tested population at Ancestry.com.
GWorks is a tool at DNAGedcom designed to automate comparing the people in all the different trees of your DNA matches. With an inexpensive subscription, you can use their “client” to aggregate the results from each different testing site and the people in the trees at ancestry. You can also manually upload Gedcoms which are easy to generate from ahnentafel lists (for that technique see http://blog.kittycooper.com/2016/10/text-to-Gedcom-using-ahnen2ged/ ). Then you get GWorks to compare all those trees looking for common people.
Much to my amazement, as I took screenshots of the GWorks process using my case, I was suddenly able to solve it! And he had NO good paternal side matches! It was all from the trees of fourth cousins. And done in two days using GWorks. (OK, once I saw that I had it, I stayed up until 2:00 am building the tree, I admit it, still…)
So how did I do this?
My third cousin Irene (found with DNA) complained to me via email that her ethnicity results at 23andMe were very different from Ancestry; what was going on? Irene’s recent ancestry is known to be half Norwegian (my side) and half Scottish with some Irish (about 6% expected) mixed in.
Ancestry’s predictions are on the left, 23andMe’s estimates are on the right
So I looked and saw that 23andMe (above right) has a very low estimate for Scandinavian for her, while the British and Irish (her Scottish) is fine. Ancestry DNA (above left) has the Scandinavian predicted correctly but calls her Scottish both Irish and Western European. At least it includes her in a Scots genetic community which is correct and very cool.
Family Tree DNA’s predictions
Since I had also uploaded her DNA results to Family Tree DNA, I checked there and saw that they correctly predicted 49% Scandinavian. However the Scottish shows as 36% British Isles followed by some traces including 9% East Europe? No one else saw that!
One of the problems is that our Viking ancestors visited and even settled in various parts of the British Isles making it hard to tell British from Scandinavian. Also there was more migration within Europe many hundreds of years back than we might realize. Finally each company uses different reference populations and techniques for their ethnicity estimates.
Predicting where your ancestors originated based on your DNA test is just not that accurate yet. The broad strokes are pretty good, but since the actual percentages vary quite a bit from company to company it is easy to see that there is much improvement needed.
Somewhat alarmed that 23andMe, which I had always thought was the most accurate of the three, did not predict Irene’s Scandinavian correctly, I decided to double check the predictions for my Norwegian-American Dad, as well as my brother who is tested at all three of the main companies.
No I am not planning a Friday Finds as a regular feature, just an occasional one. Many articles caught my attention this week so I thought I would share some of the posts I plan to read or reread this coming weekend.
How to organize your DNA data is a perpetual and perplexing question, touched with at the end of this excellent article from Legacy Tree Genealogists, “Going Beyond Ethnicity Estimates in DNA Testing”
Another article that might help you organize your results demonstrates a visual way to show the shared DNA among family members. Read this recent guest post by Lauren McGuire on Blaine Bettinger’s blog “The McGuire Method – Simplified Visual DNA Comparisons” to learn more.
The Journal of Genetic Genealogy (JOGG) has resumed publishing under the editorship of Leah Larkin and there are a number of articles that interest me in the most recent issue.
On my reading list are CeCe Moore’s The History of Genetic Genealogy and Unknown Parentage Research, Blaine Bettinger on citizen science and a review of GenomeMate, a tool many researchers use to organize their data. Personally I use spreadsheets because I started out that way. One for each family members’ segments, a list of contacts, plus a list of known relatives with kit numbers and email addresses. Click on DNA spreadsheets in my tag cloud to learn more.
And for my Norwegian research, I have already read and plan to reread this terrific article by Martin Roe, Let’s wring the census records. Most of these census records are online at the Norwegian archives (see my post on those archives)
In two recent blog posts I have shown how your WIKItree pedigree can be linked to your DNA results at GEDmatch.com and 23andMe, but I forgot to emphasize that you must set the privacy level to public tree for that to be useful.
One of the great features at WIKItree is the many privacy levels but the default for a living person is completely private which means that your family tree, starting from you, cannot be seen. Please reset that to private with public tree. This preserves the privacy of all the living just fine; for example, my brother is listed on my chart as private brother 1950s – unknown (see previous blog post for image).
To change the privacy level, go to the page for that profile which you must manage or have trusted access to. Click on the tab that says Privacy.