Swiss Cheese Multi Layered Approach to Genetic Genealogy

We need tools to ensure we are interpreting our DNA results correctly. The best tools for this are a chromosome browser and well documented family trees.

In the case of a Pandemic the multi layered protection is used to plug any holes which may allow the virus through. The more layers the more likely holes are covered.

When using genetic genealogy to infer relationships we need to make sure we use multiple tools so we cover any holes which may lead to making false assumptions regarding how we are related to our matches. 

Some say there is no way to say for certain where the segments we share with our matches come from. True, we cannot say where they originally came from but we can say who we received the segments from in the past few hundred years. 

How do we ensure the segment is Identical by Descent?

Using shared matches 

Looking at the single segment I share with my match below without any context it's meaningless. 

Checking shared matches I add context for this matching segment. I tested my mother's DNA and she doesn't share this match but my paternal aunt does. Now I know the segment is from my paternal side. Since my aunt doesn't share this segment with our shared match I haven't confirmed it's IBD (Identical by Descent). 

People who dismiss the use of a chromosome browser state that the likelihood of relatives and distant cousins sharing the exact same segments is low. Since many of our 3rd great-grandparent and generations earlier ancestors left many descendants the chances of sharing the same segment with others is good. It's just a matter of odds. 

MyHeritage has some of the best DNA tools. Sometime after this first match showed up on my match list at MyHeritage I noticed an uncle of this original match showed up. Below you see my predicted relationships to my matches and their predicted relationships to each other.   

I now check the original matches uncle's information.  He is a generation back and shares even more segments with me. So now I have more than a single segment to compare with other matches. We see segments shared between my aunt and our original testers circled below. These segments are IBD and over 15cMs. My aunt being so close genetically to me we know the larger segments I share with her are IBD.

Do any other matches share these same segments? Yes, below you see another match who shares a portion of the same segment. This match doesn't match my mother either, so we know it's paternal. 

Looking at the chromosome browser to visualize the segments shared with our original tester's uncle we see she shares long stretches of DNA with him. 

Using shared Ethnicity as another layer

It was pretty easy to determine how these two matches were related to us. Looking at the shared ethnicity our matches are predicted to be Eastern European/Balkan, and they are currently living in Austria. They also share the same genetic groups. My paternal grandfather Rudolph Kapple was born in Austria. Looking at shared matches with trees they also relate through the Burgenland area of Austria. Do they all have overlapping segments? No, but some do. 

Trees are important but not always necessary

Having a well documented tree is important but not always possible. Can we confirm how we relate to this match without a tree? Looking at the trees a few shared matches posted they probably were related through my paternal great-grandmother Mary Kurta-Kappel's line? I contacted this match to confirm how these matches are related to us? Since the predicted relationship to my aunt was 2nd cousin to the older man I assumed they shared Kurt/Jost great-grandparents? Without a tree I didn't know for certain. 

A documented tree was unnecessary in this particular case because one of the matches had photos of members of my own family sent to them by my great-grandmother. She sent them to her sister who returned to Austria. My great-grandmother, Mary, lived in Chicago until her death. One photo, among others, was taken in Chicago and my aunt June, who I knew very well, was in the photo. 

Photo of Mary Kurta-Kappel and her grandchildren, June and Junior, owned by Austrian cousins

Testing close relatives and second cousins helps us to find IBD segments and confirm them. Two 1rst cousins once removed on my paternal grandfather's side tested along with second cousins in that line allowing me to collect a number of segments to compare with matches, as you can see below. I know these segments are from my paternal grandfather. My paternal grandmother was Irish, English, Scots-Irish, and French Canadian and not Eastern European. 

The segments I received from my Austro-Hungarian grandfather line up well with the chromosome ethnicity paint browsers at AncestryDNA and 23andMe. That is another confirmation that these segments are IBD.

A more difficult case

I would like to use DNA to try to identify a set of my mother's 3rd great-grandparents. I'm searching for Sarah Campbell's parents. I'm guessing her father's name might be James, but James may have been her brother and not her father? 

Sarah's maiden name was Campbell I assume she was of Scottish or Scots-Irish descent. With so many matches sharing this same early American Scots-Irish heritage ethnicity results aren't as useful. These matches have tested at Ancestry so I can't compare segments in many cases. A couple of these matches have also tested at other companies that share segment information. 

Without segment data for many of the shared matches at AncestryDNA well documented trees are important. Some shared matches have no trees. The few who do often aren't documented at all. If I were to use these matches as a form of proof I would need to document the trees and make sure we don't share another line in common. Having IBD segments and at least one or two people with documented trees to compare with alleviates the need to document additional trees. If you have IBD segments on a confirmed line, and at least one match has a well documented tree, documenting another dozen trees isn't as necessary. Shared segments can lessen the need to do extensive research on your matches undocumented trees. 

Sarah Campbell-Wray died before the 1850 US Census. She couldn't tell us where she thought she was born. A daughter of hers, on a later census, stated her mother was born in Tennessee. My mother has a number of matches with descendants of George Lafayette Campbell of Greene County, Tennessee. The trees posted for these matches aren't showing any other relationships to these families. 

I've been able to collect a few more segments shared by descendants of  George Lafayette Campbell.  I'm keeping a look out for more matches on this particular line and hope some will agree to share segment information if they tested at AncestryDNA. This would give me more segments to catch matches with. 

  

When looking at the trees I check to make sure I'm not just collecting segments of close relatives of my matches. Siblings and parents sharing the Campbell segments I'm interested in wouldn't be as meaningful as more distant George Lafayette Campbell cousins sharing the same segments. I have found the matches I'm comparing with are generally 3rd cousins and farther back. 

Can I trust their trees? Yes, I've found supporting documentation such as census and burial information. 

I have not been able to link Sarah Campbell with any siblings. If I could, that would also help break down this brick-wall. I wouldn't say I have been able to prove Sarah Campbell is related to the Greene County, Tennessee Campbells beyond any doubt. The DNA evidence is pointing in that direction however. I need many more matches and supporting evidence. 

As you can see it takes analysis and use of a variety of tools to ensure we are drawing the right conclusions regarding how we are related to matches. 

Segment information is important to collect because that is what matching is based on. It's a key piece of evidence just like a primary source document. You can use shared matches to predict how you are related to someone, but a  collection of IBD segments actually makes identifying matches much quicker and more accurately. Once you've mapped your segments you can refer to that map every time a new match shows up you are interested in. You can also add newly discovered segments. 

If you don't want to make a map you can generally use shared matches, with the chromosome browser, to identify the segment or segments. 

I would recommend testing, or uploading your results, at all of the DNA companies if you are trying to break down a brick-wall. The company with the best tools is MyHeritage and I would definitely suggest testing with them or uploading your raw data there. 

There can be holes, like those found in swiss cheese, when we use only one tool to trace our family trees. If we are interested in a tree with people who are genetically related to us DNA testing is necessary because of adoption and non paternity events. Of course in order to have a family tree we also need names collected from sources such as the census etc.. Layering all of this information we can produce a tree which accurately reflects our genetic family. 

Wonderful New Ethnicity Browser at Ancestry/Now we can compare

 I thought I was hallucinating when I saw a chromosome browser at AncestryDNA. They now have a beta ethnicity chromosome browser. 

With the new browser we can now compare these ethnicity results with 23&me. At AncestryDNA you can select their parent 1 and parent 2 phased results to display an ethnicity browser showing what you inherited from each parent. 

Since each company has their own ethnicity categories it's a little tricky. 

I actually decided to look at the Sweden Denmark chromosomes from Ancestry first. I have no Scandinavian or Danish results at 23&me. I have no one from these ethnic groups on my family tree going back 7 and more generations. What does 23&Me tell me about chromosome 3 and 8? Part of chr 3 is likely French Canadian as I will explain later. 

23&Me doesn't allow you to select a phased parent to display inherited ethnicity but I can figure out how they phased the results because my mother had indigenous roots and my father didn't. I know the top line of each chromosome is my mother and the bottom is my father. Looking at chromosome 3 23&me is speculating a segment in the same place as Ancestry is British and Irish instead of Swedish and Danish (I believe this is a French Canadian region on the chromosome as I explain later). 

Chromosome 8 is split between broadly Northwest European and Eastern European. Sounds like a portion of chromosome 8 is difficult to place in a category. 

Eastern European makes a lot of sense because my paternal grandfather was Austro-Hungarian. Chromosome 8 might be difficult to place because it likely represents my paternal grandfather who, being from the Austro-Hungarian empire, had a very mixed Eastern European heritage. 

AncestryDNA is giving me more Germanic, all on my father's side, compared to 23&Me. My paternal grandfather was ethnically German. 

23&Me combines French and German in their predictions. I'm guessing the segments from my paternal side are mainly German. I do have French Canadian ancestors but that line is farther back. The German segment on chromosome 9 is being attributed to my mother's side. She did have German ancestors also. AncestryDNA predicts this region of the chromosome is English and Northwest European. I will look at the ethnicity of matches in that region. 

AncestryDNA has an Eastern European Russian category which can be compared with Eastern European at 23&Me. My paternal grandfather was born near Graz, Austria. His family was originally from what is now Burgenland, Austria, today, but used to be Vas, Hungary before WWI. 

AncestryDNA and 23&Me's results seem to line up fairly well when comparing Eastern European predictions. 

23&Me predicts I have more Eastern European segments. They both agree chromosome 11 is from my paternal line and is all Eastern European.

Spanish has always been difficult for the companies to separate out. When I first got my results back, years ago, I had no Spanish admixture. My grandmother Graciela del Castillo was from Nicaragua and definitely has some Spanish ancestry. 

AncestryDNA separates Spanish and Portuguese but 23&Me doesn't so I displayed Spanish and Portuguese by selecting both at Ancestry. 

23&Me is giving me Spanish segments on both paternal and maternal lines. I only have Spanish descent on my mother's side, as far as I know. 

British, Irish, and Scottish are much more complicated to compare because I have substantial ethnicity from these ethnic groups on both sides. I'm not even trying to figure that out. 

I was not able to compare on a continental level with Europe but I could regarding Indigenous American. Comparing at that level the results are much more identical. 

To compare the indigenous results between AncestryDNA and 23&Me I had to select more than one category at Ancestry, that's why the colors are different. Here we see 23&Me segments on the left and AncestryDNA on the right. 23&Me segments are hard to see because the color is so light but they do match up well with AncestryDNA's segments. 

Neither company has been able to find my French. It's completely missing. I have French Canadian DNA matches on my Mason/Masson side. Segments in the regions below are from my French Canadian matches and should be identified as French Canadian. AncestryDNA misidentified this region on chr 3 as Swedish and Danish. The Normans did occupy part of France so this makes some sense. There are even more segments that are French Canadian not shown below. We need more French samples but testing in France is illegal so we aren't likely to see enough French results to assist with identification of the segments. 

Oddly AncestryDNA found French admixture in my mother's results. I have not found any French ancestors on my maternal side. 

AncestryDNA is moving in the right direction by offering this useful tool. If they offered an opt-in matching segment browser I would highly recommend AncestryDNA  as the very best place to test. We hope one chromosome browser leads to another.