Tuesday, March 10, 2020

Working With My 23andMe "Family Tree"

Have you tried out the 23andMe "Family Tree" feature?  It lets you add names to your own ancestors to it, and then hypothesizes how some of your 23andMe DNA matches descend from your ancestors.  23andMe described their "automatically created" family tree in   https://www.23andme.com/topics/ancestry/family-tree/.

On your 23andMe home page, click on the "Family and Friends" link and see "Family Tree Beta" on the dropdown list:

When I clicked on the "Family Tree Beta" list, I saw the "tree" blow.  I entered data for my parents and for my parents and other ancestors that I could determine were correct.

So my tree above has me, my parents, my paternal grandparents and one set of paternal great-grandparents.  I figured those out from the DNA matches on the tree at the bottom of the chart above.  I know who the other ancestor couples are, just not which one of the couples connect to the next couple up the chart.

There are 16 DNA matches on the chart, each with an estimated relationship based on how 23andMe figured them out.  I have blanked out the match names and relationships.  Here are the 16 DNA Matches and what I know about them:

*  NJ - 0.90%, 3rd cousin, estimated 1x removed, on maternal side of the tree. No known common ancestor.
*  WH - 0.78%, estimated 3rd cousin, on maternal side of tree. No known common ancestor.
*  TK - 0.84%, estimated 3rd cousin 1x removed, on maternal side of tree. No known common ancestor.
*  TB - 1.54%, estimated 2nd cousin 1x removed, on maternal side of tree.  No known common ancestor. States is an adopted child.
*  CF - 1.16%, estimated 3rd cousin, on maternal side of tree.  No known common ancestor.

*  KH - 1.01%, estimated 2nd cousin 2x removed, on paternal side of tree.  No known common ancestor.  Also an AncestryDNA match.
*  CR - 1.22%, estimated 2nd cousin 2x removed, on paternal side of tree.  Common ancestors are my grandparents Frederick Seaver/Alma Bessie Richmond.
*  AP - 0.72%, estimated 2nd cousin 2x removed.  On paternal side of tree.  No known common ancestors.  Tree indicates may be sibling of JR.
*  JR - 1.40%, estimated 2nd cousin 2x removed.  On paternal side of tree.  Known common ancestors are my 2nd great grandparents James Richman and Hannah Rich, so a 3rd cousin.  Also an AncestryDNA match.
*  DB - 0.79%, estimated 3rd cousin 1x removed, on paternal side of tree.  No known common ancestor.  Also an AncestryDNA match.

*  TS - 0.73%, estimated 3rd cousin 1x removed, on paternal side of tree.  No known common ancestors.
*  PC - 0.73%, estimated 3rd cousin, on paternal side of tree.  No known common ancestors.
*  LB - 0.73%, estimated 3rd cousin 1x removed, shown on paternal side of tree.  No known common ancestors.  also an AncestryDNA match on my maternal side of tree, with shared ancestors to Kemp and/or Sovereen line.
*  AJ - 0.72%, estimated 3rd cousin, shown on paternal side of tree.  Also an AncestryDNA match.  Common ancestors are 2nd great-grandparents James Kemp and Mary Jane Sovereen on maternal side of my tree.
*  JW - 0.94%, estimated 3rd cousin, on paternal side of tree.  A known 3rd cousin, with common ancestors of 2nd great-grandparents James Kemp and Mary Jane Sovereen on maternal side of my tree.
*  DW - 1.40%, estimated 3rd cousin, on paternal side of tree.  A known 3rd cousin, with common ancestors of 2nd great-grandparents James Kemp and Mary Jane Sovereen on maternal side of my tree.  Sibling of JW.

My conclusions from all of the above is:

*  CR relationship is correct, but is attached to the wrong common ancestors (should be Frederick Seaver and Alma Bessie Richmond).
*  JR relationship is correct, but is attached to the wrong common ancestors (should be James Richman and Hannah Rich).
*  At least 4 of the 16 matches (LB, AJ, JW and DW) should be on the maternal side of my tree.
*  If I can identify common ancestors for CF, then I might be able to determine the connection to the adoptee, TB.

The basic problem with this 23andMe "Family Tree" is that there are not enough DNA matches with a well-developed family tree to determine which ancestors are identified on the tree.  Only the two known paternal matches (CR and JR) help me identify my paternal grandparents and paternal great-grandparents.

If LB, AJ, JW and DW were on the maternal side of the 23andMe family tree, then I could identify more of my ancestors on the maternal side of my tree.

Why are LB, AJ, JW and DW on the paternal side of my tree and not on the maternal side as I know them to be?  Perhaps because they share a segment on a specific chromosome pair with TS and PC, and 23andMe assumed they were all on one of the pair, because the companies cannot tell which segment is on the paternal or maternal side of the specific chromosome pair.

How are other researchers dealing with this automatically generated "family tree?"

It is an interesting solution to the problem, but I've found that it is incorrect for a non-trivial percentage of the DNA matches considered.

                                =============================================

The URL for this post is:  
https://www.geneamusings.com/2020/03/working-with-my-23andme-family-tree.html

Copyright (c) 2020, Randall J. Seaver

Please comment on this post on the website by clicking the URL above and then the "Comments" link at the bottom of each post.  Share it on Twitter, Facebook, or Pinterest using the icons below.  Or contact me by email at randy.seaver@gmail.com.

2 comments:

Louis Kessler said...

Randy, 23andMe does not use other people's family trees to determine your family tree. It creates it simply from your matches and shared matches, sort of the way clustering works

Randy Seaver said...

Louis,

I should have said that at the beginning of the post. It is evident from the 23andme description article.

If more matches had a family tree linked to their 23andme results (I know, they don't have one now, but they did before with a MyHeritage link), then it would be easier to determine common ancestors.