A Segment-ology TIDBIT
ThruLines is based on genealogy – it finds Common Ancestors based on your Tree and the Trees of others. However, it only reports Common Ancestors with your DNA Matches. So, in a sense it has a DNA component. But the connections TL finds are not based on shared DNA cMs, Chromosome location, segment Triangulation, Clustering or Shared Matching – it is based only on connections found through Trees (only on genealogy). And ThruLines only reports Common Ancestors with your DNA Matches.
This is a two edge sword:
- If you only want to work with DNA Matches, it’s a good thing.
- However, if you are a genealogist looking for cousins who might share records, pictures, stories, analysis, new branches, etc., it leaves something out. Remember that roughly half of our 4th cousins (4C) don’t share DNA with us, and roughly 90% of our true 5C don’t share DNA with us, and the vast majority of our more distant true cousins don’t share any DNA with us. This means that, although a program like ThruLines could find those non-DNA-sharing cousins for us, it doesn’t. Think of all that we are missing – think of all the lost opportunities.
Well… looking back on the #1 cutting edge of the sword – I’ve got to be a happy camper. I’m finding more ThruLines Matches than I can keep up with. By adding children and grandchildren of my Ancestors in my Tree, ThruLines is finding more Matches with Common Ancestors. And these Matches and their Trees are reinforcing my Tree (and pointing out a few soft spots…)
Back to work… Stay safe!
[22AR] Segment-ology: AncestryDNA ThruLines Missing Out – TIDBIT by Jim Bartlett 20200326
segment-ology 
Jim – I read and appreciate every word you write here. And i agree – ThruLines are only as good as the tree they came from – but they do provide a quick filter to those searching the same relative.
If I may, I would like to address a previous post about “clusters” and your thoughts about what significance are to be attached to identical starting points.
On my chromo 1 I have an identified paternal match of 25 cM beginning at 109,025,xxx and ending at 157,xxx,xxx. Under that “umbrella”, I am developing a cluster which starts at 110,332,xxx extending for 9 to 13 cM. There are 24 matches with the identical starting point – and some with indications of ancestors from my maternal side.
Because the paternal match starts before the 24 matches with identical start points, their start does not appear to me to be a chip address issue. Nor would I think the crossover was from my parents – but likely from my maternal grandparents or someone up their line. Am I thinking right?
Has anyone attempted to develop an algorhythm to sort and trace ancestors using such overlaps and their possible creations?
I would appreciate your comments.
Thank you,
Vance Wiley
LikeLike
Vance – thanks for the kind words of encouragement. I think we are all helping each other learn more – and your thoughts and questions help us stretch even more.
I have about 20,000 Match-segments in my spreadsheet. A few years ago, I had enough to basically cover all of my DNA – from the beginning of Chr 01 to the end of Chr 23 (X). Except for a few very small gaps, I had a shared segment that “covered” the 43 chromosomes I got from my parents. With segment Triangulation, this has resulted in 372 Triangulated Groups. These 372 TGs are adjacent to each other and cover each Chr. All new shared segments “fit” into one of those TGs.
You are correct that there are start points (and end points) that are fixed. These are fixed recomination points in our DNA. On average there are about 34 new recombination points created each time a parent passes their DNA down to a child. 34 cut points, over 22 chromsomes means about 56 segments. If you do Visual Phasing or use TGs to identify the recombination points or use DNA Painter, or any other tool to map your segments, you’ll see roughly 56 grandparent segments from each parent (usually a few more from the mother and a few less from the father).
This is all about our own DNA. When we look at *shared* DNA with a Match, it’s like a Venn diagram – we and our Match have our own personal segment from an Ancestor and what we see as a shared DNA segment is the overlap. Usually one end of this overlap is a recombination point – and we’d usually see this point more often than any other point. In fact we “cannot” see beyond the boundaries of the segments we got from our Ancestors. Well…. that’s not entirely true in every situation. If we are comparing with a Match who is a close cousin, the shared segments tend to be longer. And this means they may cover several crossover points. A 1C may share most of a large segment from a grandparent. But that segment is made up of smaller segments from the grandparents ancestors. But your smaller shared DNA segments from more distant Ancestors, will tend to behave better and be within the crossover points defined by the TGs.
Another explanation for small deviations (a few cMs) from recombination points is the result of fuzzy data. These days, with companies using different chip sets, and MyHeritage and GEDmatch incorporating forms of imputation, we sometimes get shared segments which are longer than they should be. Imputation adds some SNPs – it adds SNPs that are highly probable – but the SNPs that are different are what define the crossover points, so the imputed SNPs tend to create longer segments than what we really got from an Ancestor. And remember, there is no “signpost” in our DNA which says: “stop comparing ACGTs, you’re past a recombination point”. So some shared DNA segments will appear to lap over a crossover point – it’s either fuzzy data (just ignore the overlap part) or the Match is sharing a DNA segment from a closer Ancestor.
To answer your question about an algorithm – I think segment Triangulation provides a process to define crossover points. I’ve done that for all of my DNA. Now the problem is to determine which Ancestor (or ancestral line) provided the DNA for each TG. It turns out some are easy and obvious – even out to 7xG grandparents. Other TGs I know the grandparent, but cannot seem to get a 3rd cousin Match with a Common Ancestor. At AncestryDNA, I have ThruLines Matches with Common Ancestors for all my (known) 5xG grandparents – so the “coverage” is there, but the segment data is not… So close, and yet so far! I’m working on it. Jim
LikeLike
Great article. I forgot that ThruLines are NOT created from DNA. Thanks for reminding us.
LikeLike
Paul, I would love to see AncestryDNA develop a program that utilizes the DNA data they have – even behind the scenes. They could easily just tell us that which Matches have Triangulated Segments, for instance (very similar to what 23andMe and MyHeritage does).
Jim Bartlett jim4bartletts@verizon.net
LikeLike
I completely agree. If Ancestry just indicated with a little “T” for triangulation which of our matches were; that would be such a help. I find myself using MyHeritage and 23andMe more and more, JUST because of the triangulated information provided. For me it is always a Big Green Shaking Leaf aka hint for me to narrow my focus with the massive DNA data coming at us.
LikeLike