Wednesday, May 3, 2017

Evaluating AncestryDNA Matches

AncestryDNA now has 4 million testers, making their database of testers the largest in the field. The new numbers caused me to think about what this large comparison database has done for my genealogy?

I've written many times about how I work with my DNA results. I've been using segment data to build a chromosome map with "Genome Mate Pro". That got me thinking. Without the segment data what has AncestryDNA provided me with? I have been able to verify my paper trail based on some strong matches at Ancestry. By strong matches I mean they share a substantial amount of DNA.

These days I generally only look at the matches predicted to be 4th cousins or closer. Actually I generally only feel most interested in 3rd or closer. Just out of curiosity. I always hope they might be persuaded to upload elsewhere. It's also interesting to see how we are related. I have no brickwalls in those generations however. Without the segment data they provide nothing of genealogical relevance. So far they've provided me with no additional family information either. Some have had basically no information at all and are adopted, or are estranged from their birth families. Maybe some day someone will come forward with some pictures or additional info?

I received a message from a match that I sent a request to compare with at GEDmatch. I got her reply yesterday. She basically said she was afraid GEDmatch wasn't secure enough. She also said she didn't see any use for segment data. She made connections at AncestryDNA without it. I explained that if your match doesn't have a large family tree you won't know if you are related to that person on more than one family line.

To get a clearer picture regarding what I've gotten and what AncestryDNA, sans segment data, can provide I decided to take a look at how many matches actually have a tree with an ancestral couple related to me. I searched using the hints filter.

Instead of using my results I used my mother's. My father had a great deal of immigrant ancestry, with some French Canadian. I'm related to most French Canadian matches on several lines, making identifying the common DNA ancestors impossible.

First Cousins


1rst Cousins once removed


2nd Cousins once removed


3rd Cousins and once removed


4th Cousins and once removed


5th cousins and once removed


6th cousins and once removed

7th cousin

According to AncestryDNA my mother has four 3rd cousins. Her age probably threw the calculations off. Instead she has 24. Most are once removed, but a few are 3rd cousins.

Taking this a step further how many family lines are represented? Out of the 135 matches that have an ancestral couple matching, 79 are from one family. The Roller/Zirkle family.More than half.

Fifty-six other matches, with matching ancestors, are scattered throughout other families, with Browning the second most represented.

Eve Zirkle was my mother's 3rd great-grandmother, and over half of the Roller/ Zirkle matches appear to descend from her. According to the book "Genetic Genealogy in Practice", by Blaine T. Bettinger and Debbie Parker Wayne, a third great-grandparent could share as much as 3.125%, on average, with you. Looking at my chromosome map it does look like Eve Zirkle shares that amount, or more, with my mother.  Did she get more than the usual amount of DNA from Eve Zirkle, which would account the overwhelming number of matches coming from this family?  Her matches are with both Zirkle, Eve's father, and Roush, Eve's mother's family.

I've marked the Zirkle/Roush matches segments on this map:

To answer the question whether 3rd great-grandmother Eve Zirkle contributed more than the average of 3.125 percent of DNA I needed to see what that amount of sharing would look like in a chromosome browser.. Here is roughly what an around 3.125% share would look like.

If the chromosome lengths are about accurate it does look like my mother may have inherited a little bit more than the average DNA share from her 3rd great-grandmother. The Zirkle and Roush family tended to marry cousins, which could account for more than average amounts of DNA passed down.
What has this demonstrated to me? Zirkle/Roush descendants predominate my match list, and using the shared match feature the Roller/Zirkle, and Zirkle/ Roush, numbers overwhelm all other matches. This is true at the other companies also

 I do believe if AncestryDNA provided a segment data sharing option I could fill in the chromosome map for other family lines, which are less well represented. Without the segment information over 1000 matches aren't connected with any family because they lack large enough trees, or have no trees at all. You can jump to conclusions using the shared match feature, but you never know whether you're related to someone more than once?  Or it's a coincidence you share the same match? A segment map is the only way to confirm which line someone is related on.

Out of the 135 matches, with matching tree ancestors, only about 15 have uploaded to a place where we can compare segments.

During this exercise I did find a couple more matches who have uploaded elsewhere. One of these matches shares two different family lines with me. One being the Zirkle/Roush line. Adding her information to "Genome Mate Pro" I was able to determine she matches through Zirkle/Roush, and not my Browning line.

At this point I continue to check for close matches at AncestryDNA out of curiosity, and in hopes someone may come forward with more information. I also check for my brickwall Campbell line, which so far hasn't produced any results. I keep checking GEDmatch for more AncestryDNA matches. Sadly few AncestryDNA matches upload there, and the ones who do can't be easily found on the AncestryDNA match list. They also often don't have good trees, or any trees posted.

The last point I would like to make is I used the Google chrome extension to come up with the many of the facts I presented. I was able to quickly search my notes, and search for user names.This tool provides better search capabilities than AncestryDNA offers..