Wednesday, April 6, 2016

Recap Of Sunday's SL Discussion: Working With Segments

We've been having DNA discussions in the virtual world of Second Life (just consult the "Genealogists In Second Life" page at Facebook for more details). Our next discussion will be on Sunday May 1, 2016.
Here is a recap of our last meeting Sunday April 3, 2016.
This slide shows a view from the 23andme chromosome browser

  • The browser shows our 23 pairs of chromosomes. One row from our Mother the other from our Father
  • It shows where my Mom, a match, and I share DNA.
  • Mom shares from side to side on each chromosome (the purple lines).
  • Match is represented by the red segments
  • Does she match my Mom?
  • We would check either my Mom or my matches match list and see if they match
  • We would then check the chromosome browser to verify they match on the same chromosomes in the same place
  • Another view (above) of a chromosome browser shows the pair of 23 chromosomes, plus the X
  • I've listed some alleles (i.e., A's, G's, C's, T's) which are used to determine who matches 
  • One line across each chromosome always represents a parent. The companies can't tell which alleles come from which parent. Only testing parents can work this out, in order to list the alleles separately and correctly for each parent. Without that its up to software to figure this out
  • AncestryDNA phases results without parents. This often produces good results, according to them. They acknowledge a small error rate
  • The same positions are tested for each parent
  • Because of that segments two, or more, matches may look like they match in the chromosome browser. You must know whether the match is a match on the paternal or maternal side to know whether overlapping segments represent a match
  • I tested my mother so I can narrow down the possibilities
  • I borrowed someone else's grandparents to demonstrate how having grandparents segments is useful in establishing IBD segments and finding a more precise relationship with matches.
  • Here you can see the possible 3rd cousin match also has the same matching segments as she had with my mother. She matches my maternal grandmother on the same segments
  • As it turns out she is also Nicaraguan. Just like my grandmother
  • These segments are certainly IBD

  • You can see (in the slide above) that the other set of grandparents don't share the same red segments. This confirms there is no match on the paternal side
  • This view shows the segments that my grandparents gave me through my father. The side to side shares from parents are divided up by segments they got from their parents. Siblings receive different assortments of DNA from grandparents.
  • Grandparents DNA is further segmented with DNA from their parents
  • This chain of inherited DNA continues back in time, until we not longer share DNA with certain ancestors
  • Notice here how the paternal grandparents segments fit together like puzzle pieces (above)

  • If you don't have living grandparents you can recreate the segments your grandparents gave to you by testing and comparing segments with 2nd cousins
  • There is also a way to test siblings to find start and end points for grandparents shares

  • Downloading segment data from Family Tree DNA , 23andMe, and GEDmatch allows you to compare segments from matches from all three places
  • You can upload and compare the segments at third party sites such as Kitty's Chromosome Mapper. Or you can download a free app from Genomate Pro (good idea to donate too). This software will allow you to store and compare your segment match data.

  • Triangulation is useful for all testers. The definition of triangulation is having a matching or overlapping segment with two matches or more
  • It's especially useful if you don't have close relatives who have tested. In such cases a good triangulation can establish a segment as identical by descent, or not a false match
  • Triangulated shared segments, or overlaps, should overlap by at least 7 cM's
  • When triangulating without establishing segments as identical by descent, using close relatives, it's best to use segments in the IBD range. According to Dr. Tim Janzen 15 cM and 2500 SNP segments are more likely to be IBD
  • In September of 2015 Vee got in contact with me through 23andMe's messaging system. She asked me to look over her list of names, and said she had a tree on Ancestry. I determined the most likely connection was through the surname Grenier.  I have French Canadian Ancestors with that surname. Her ancestors did not come to the US through Quebec, however. Instead they came from France to New York in the 1700's
  • Last week I began using Genome Mate Pro. This app allows you to look at shared segments on each chromosome from various sources. I noticed Vee, my Paternal Aunt, a Paternal First Cousin, and a third cousin on the paternal side all matched on chromosome 6 in the same place. I found 2 others who also matched
  • I had to dig a little to find trees for the two additional matches. When I compared everyone, except my 3rd cousin, and closer relatives, I found they all shared many of the same surnames, and all had Southern roots 
  • It also dawned on me that the names shared by these 3 common matches were also the surnames associated with  all my NAD's
  • Looking through the NAD's again I believe our common connection has something to do with the surname Douglas. I had first thought Troxell was the common surname, but one match in the NAD's doesn't share that surname. Still trying to work out our connection because I don't have Douglas on my tree?
  • All of these matches share 25 to 32 cM segments, and around 8000 SNPs, on chromosome 6. This makes it nearly certain the shared segments are identical by descent
Someone at the discussion asked "Once you have verified your segment matches, did you then establish documentation through public records etc..?" Yes. You need to also compare documented trees to see how you might be related to a match. Unfortunately it's difficult to document trees at Family Tree DNA. Or maybe it's just not as straight forward? You can add stories and notes to your Family Tree DNA tree, which many of us, including me, haven't done.
We also discussed the fact that it's difficult to get AncestryDNA matches to respond to messages, let alone upload to GEDmatch.
Another problem with AncestryDNA that we talked about is the lack of specific segment data, which hampers our ability to make the correct DNA connection with our matches. Matches can match through more than one couple; so segment mapping would help to determine which couple the DNA likely came from.
Another problem someone brought up was the need to meet the genealogy proof standard, regarding the lack of segment data at AncestryDNA. Scholarly genealogy journal articles, which refer to DNA testing, include exact segment data. Without the exact data and comparisons your proof argument won't hold up to scrutiny.  
Next meeting we will discuss GEDmatch.