As I look at ThruLines Matches under 15cM, roughly half of them have a Side (Maternal or Paternal) which is different from the Side of the Common Ancestor proposed. What’s up?
AncestryDNA has determined a “side” (Maternal or Paternal) for most of my Matches. Pretty slick! And very helpful!! For above-20cM Matches they appear to be fairly accurate. This is despite the fact that all of my Paternal and half my Maternal Ancestor were mostly from Colonial Virginia. I was expecting a lot of Matches to be “Both”, but relatively few are. The bulk of my Matches are in the Maternal and Paternal categories. And below 15cM, the Maternal or Paternal “sides” are not aligning with the “side” for many of the ThruLines Common Ancestors. Side note: it appears that Ancestry is now only reporting one ThruLines Common Ancestor per Match – they used to report two or three if they found them….
What are the possibilities?
1. The AncestryDNA “sides” may be incorrect. I’d like to think (hope?) that the science behind them is valid and that they are largely correct. Most of mine above 20cM appear to be.
2. The ThruLines may be incorrect. This is a genealogy area (not DNA). With my 50 years of genealogy research, I already know many of the descendants of my Ancestors, and I run a check (not-GPS-comprehensive) on each ThruLines reported. I used to spot about 5% with errors (some of which were easily fixed), but now there are more and more as AncestryDNA appears to have become fairly aggressive at finding Common Ancestors. It appears they have loosened up the algorithms to allow “close” name variants and “close” dates, resulting in more false results. But even with the ThruLines I review and accept the Common Ancestor from a genealogy point of view, there are roughly half which don’t agree with the “side”.
We cannot have it both ways… or can we?
When AncestryDNA determines a Maternal “side”, does that guarantee that 100% of the Match’s atDNA can only be on my Maternal side? I really think that is absurd! Particularly when you consider most of my Ancestry is from Colonial Virginia. Surely my Colonial Virginia Matches could descend from Ancestors who would be on both sides of my Ancestry. In fact, I have several of my own Ancestors who, due to distant pedigree collapse, are on both sides of my Tree.
I think it is entirely possible that the bulk of a Match’s atDNA could align with my Paternal or Maternal DNA, but that some of the Match’s segments could be from the other side. I’m scratching my head over whether or not this could occur half of the time.
3. Both Ways! My conclusion is that we can have it both ways! I have a colored Dot for cases with both “sides”, but I’ve decided not to let that, by itself, stand in the way of accepting a ThruLines Common Ancestor as valid.
I’m curious about your overall experience and observations about conflicting “sides”. You are encouraged to add your insights in the comments.
[35AA] Segment-ology: AncestryDNA Side vs ThruLines Side by Jim Bartlett 20240213
segment-ology 
I also agree. I have found man from both. I understand there may be mistakes but both are of value
LikeLiked by 1 person
Ciao jim ho avuto una corrispondenza 0,4percento (25,5cM)con 4 segmenti il piu grande 6,6 anche sua sorella ha 3 segmenti cisa vuol dire?
LikeLike
I have not noticed any wrong sides, or sides in conflict with Thrulines.
There are quite a lot so have not yet checked them all
LikeLike
Bob, It takes a *lot* of time and work for me to get to the bottom of my ThruLines Common Ancestors (some in the saved-6cM range). The issues I brought up are generally at the lower end of the list. And some really good gems are also found there too… Jim
LikeLike
As usual, a very interesting article. I share your experience.
Until recently, I was confident that all but 5% of my Common Ancestor Hints were correct. I take the time and trouble to add the pedigrees of these matches to my Ancestry tree, relying on primary sources. Of the 5% that I could not confirm, I could tell from the shared matches that many of the Hints were likely on the right track. Probably less than 1% were completely out of left field, and they were always very small matches.
But recently, as you have noticed, there has been a significant increase in the Hints that I cannot confirm with sources. It seems that Ancestry has made its algorithms less stringent, to no added benefit in my experience.
This is my experience with over 20 tests that I manage.
By the way, I rarely if ever use the Thrulines tab. I always search my matches for Common Ancestor Hints. I find Thrulines to be fairly clumsy. Or do you use Thrulines as a synonym for Common Ancestor Hints?
It is not uncommon to get a Hint for a maternal match which has been designated paternal, or vice versa. If the Hint can be confirmed, and especially if the shared matches also confirm it, I have assumed that the designated side was wrong, and that designating sides cannot be foolproof – even though it is a very helpful tool that I love.
However you suggest that designating sides is accurate. In that case I agree with what you say – we may be related to a person in more than one way via both parents. I guess if we could see which chromosome(s) made up our match (via both matches uploading their DNA to gedmatch.com), and we painted such matches at dnapainter.com, we would soon become aware if our shared DNA did not reflect the paper trail connection that we had discovered.
By the way, responding to earlier comments, just because your maternal and paternal pedigrees are geographically separate, that does not mean that all their descendants remained geographically separate.
LikeLiked by 1 person
Jean, Thank you for your insightful reply and inputs.
Like you I use a combination of “Gold Stars” to validate each ThruLines result – particularly a consensus of Shared Matches which keeps me digging when there is an error someplace.
Also like you, I eschew the ThruLines tab – not helpful to me. I am determined to adjudicate every one the ThruLines hints – they all now go into my Common Ancestor Spreadsheet (over 9,000 rows now), and I clearly mark the few that I believe are incorrect. So I use the Common Ancestors filter, combined with the Unviewed filter (almost every day to insure I pick them all up) – and I add a range filter (like 6 to 13) as I get down the list in order to speed up the reload.. Then, every few months, I scroll down the entire list of Common Ancestors and usually find a few more – some that I’ve viewed before, and now have a TL (my Dots help me quickly scroll).
By putting these in my Common Ancestor spreadsheet, I form Family Group Sheets for each of my Ancestors, I quickly see Matches who are closely related, and I can easlily “fix” certain mistakes TL makes over and over again. Someday soon I’m going to take all that valid data and sort on cousinship and build my own Shared cM table to compare to Blaine’s.
One clarification – I think Side View sides are pretty accurate on the whole. BUT, I don’t for a minute believe that all my Matches are 100% on one side or the other (just like I don’t take ethnicity estimates “to the bank”). The SideView may be accurate for most of the Match’s DNA, but, particularly with under-15cM segments (way less than 1% of our DNA), it’s entirely possible the shared segment is on the other side of SideView. I’m now comfortable accepting a “side” miss-match, if that is the only issue.
Thanks, again, for your input – very helpful. Jim
LikeLike
The most frequent way the side on my matches is incorrect has to do with ethnicity. My maternal is all French Canadian, my paternal is all Italian. Sometimes I will see a match who is clearly French derived but marked paternal-side. That person always has a bit of Italian in their ethnicity although all of our matches are clearly French Canadian.
LikeLike
So it appears that, like the ethnicity estimating system, the SideView estimates are not comprehensive or correct all the time. I think they rely on “some” of the Match’s DNA matching “some” of our DNA. They ususally get it right, but are sometimes fooled… Jim
LikeLike
I have a few mismatches on Ancestry ThruLines but my paternal and maternal lines are pretty well separated geographically. I have couple of both sides matches but have not been able to verify them.
I notice that shared matches only go to 20 cM while common ancestors and DNA matches go to 6 cM. It would be very helpful if shared matches went all the way to 6 cM, too.
LikeLike
Gary – I’d also like to see the Shared Match list threshold reduced below 20cM. But it would be a huge workload for Ancestry. Only 7% of my 93,000 Matches are over 20cM. So 93% are not included in Shared Match lists. However, note that most of these Matches below 20cM have their own Shared Match lists (all of whom are > 20cM) – so we can learn (and Cluster) from a sort of “reverse” Shared Matching. Jim
LikeLike
I too have noticed this. The “side” is almost always correct. BUT, the common ancestor is incorrect. I have determined this by using my Ancestry matches who are also on GedMatch.
LikeLike
Linda – interesting… the atDNA segment determined at GEDmatch (or any of the other companies) is a little different. That segment is the one that links to someone and creates a Match. I agree with you that *that* segment came from one Common Ancestor that should match the side. If not, there should be another Common Ancestor who does match the side. With ThruLines we don’t really know what segment is involved… Jim
LikeLike
Yes, I have noticed that phenomenon also – the Side predicted by SideView is different from the genealogical side in the ThruLine. This is more common for matches under 10 cM. I suppose there are two reasons for this. The smaller segments are less unique than say a whole 20 cM chunk. For more distant matches (e.g. < 10 cM) it is more likely that I could relate on both sides due to the number of generations to the MRCA.
LikeLiked by 1 person
Greg – I agree. And I think the reason is, in addition to number of generations, is that 10cM is such a small percentage of the overall DNA of a Match, it might not be representative of the bulk of the DNA used to determine the SideView side. Jim
LikeLike