Date
1 - 8 of 8
DNA match discrepancy ? #dna
I've just encountered a puzzling pair of DNA matches which I hope those with more expertise in this area might be able to unpuzzle.
I found what looks like a significant match between my mother Nancy and Mary through MyHeritage – 123.8cM total/10 segments/largest segments 25.2, 20.2, 15.2, 14.1. In addition, MyHeritage showed that Mary and my mother triangulated with 2 other known cousins on my mother's paternal branch.
I contacted Mary and, happily, she replied. Among other things, she let me know that she'd also tested at Ancestry though under a different name. When I found the match between my mother and Mary on Ancestry, the match data was so radically different – 21cM/7segments/ largest 15cM – that I'm stumped to understand it. Nor do I see any known shared matches, even though one of those with triangulated matches at MyHeritage also tested at Ancestry.
I've looked at enough DNA match data that I'm not surprised by slightly different numbers reported by different testing companies. I understand that the companies test different areas but it is hard to believe that this could result in such widely divergent readings. In one case, according to the rule-of-thumb for evaluating and prioritizing matches I've learned, one test reveals a close(ish) match with likely common ancestors and the other test suggests the match isn't worth pursuing.
Can anyone enlighten me about what's going on? Thanks in advance for any help you can provide.
I found what looks like a significant match between my mother Nancy and Mary through MyHeritage – 123.8cM total/10 segments/largest segments 25.2, 20.2, 15.2, 14.1. In addition, MyHeritage showed that Mary and my mother triangulated with 2 other known cousins on my mother's paternal branch.
I contacted Mary and, happily, she replied. Among other things, she let me know that she'd also tested at Ancestry though under a different name. When I found the match between my mother and Mary on Ancestry, the match data was so radically different – 21cM/7segments/ largest 15cM – that I'm stumped to understand it. Nor do I see any known shared matches, even though one of those with triangulated matches at MyHeritage also tested at Ancestry.
I've looked at enough DNA match data that I'm not surprised by slightly different numbers reported by different testing companies. I understand that the companies test different areas but it is hard to believe that this could result in such widely divergent readings. In one case, according to the rule-of-thumb for evaluating and prioritizing matches I've learned, one test reveals a close(ish) match with likely common ancestors and the other test suggests the match isn't worth pursuing.
Can anyone enlighten me about what's going on? Thanks in advance for any help you can provide.
Lee David Jaffe
===============
Surnames / Towns: Jaffe / Suchowola, Poland ; Stein (Sztejnsapir) / Bialystok and Rajgrod ; Roterozen / Rajgrod ; Joroff (Jaroff, Zarov) / Chernigov, Ukraine ; Schwartz (Schwarzstein) / Ternivka, Ukraine ; Weinblatt / Brooklyn, Perth Amboy, NJ ; Koshkin / Snovsk, Ukraine ; Rappoport / ? ; Braun / Wizajny, Suwalki, Ludwinowski / Wizajny, Suwalki
Hi Lee,
I know the feeling. I am tested on all the platforms and what I do to prevent insanity is use the platforms to screen for potentially meaningful matches, and then switch over to gedmatch where I apply my own parameters to the comparison (and use the same parameters every time). This system guarantees that I'm always comparing apples to apples.
Between you and me and the web, it seems to me that match results on some of the platforms have been getting bigger and bigger over time (more total sharing, more segments, and larger segments). I imagine this has something to do with the hidden incentive the companies have to find closer matches for their customers--- to keep the customers happy they may have been adjusting their match parameters. Meanwhile some of the companies have implemented parameters they claim produce more reliable results (and which may end up making the genetic distance appear further than previously). I believe Ancestry is in the latter category.
--
Adam Cherson,NY, NY
Benyakonski, Kherszon, Rubinovich, Solts, Grodsinski, Levine, Cohen, Kaplan, Lubetski, Karchmer, Horwitz, Rabinovich, Zussman (Lida, Voronova, Dieveniskes, Konvaliski, Smarhon, Vilna)
Genomics Publications and Presentations: https://independent.academia.edu/AdamCherson
I know the feeling. I am tested on all the platforms and what I do to prevent insanity is use the platforms to screen for potentially meaningful matches, and then switch over to gedmatch where I apply my own parameters to the comparison (and use the same parameters every time). This system guarantees that I'm always comparing apples to apples.
Between you and me and the web, it seems to me that match results on some of the platforms have been getting bigger and bigger over time (more total sharing, more segments, and larger segments). I imagine this has something to do with the hidden incentive the companies have to find closer matches for their customers--- to keep the customers happy they may have been adjusting their match parameters. Meanwhile some of the companies have implemented parameters they claim produce more reliable results (and which may end up making the genetic distance appear further than previously). I believe Ancestry is in the latter category.
--
Adam Cherson,NY, NY
Benyakonski, Kherszon, Rubinovich, Solts, Grodsinski, Levine, Cohen, Kaplan, Lubetski, Karchmer, Horwitz, Rabinovich, Zussman (Lida, Voronova, Dieveniskes, Konvaliski, Smarhon, Vilna)
Genomics Publications and Presentations: https://independent.academia.edu/AdamCherson
Sarah L Meyer
MH has some imputation problems where they "connect" some small segments that may be close. They also include segments under 7cMs. So they tend to inflate their numbers. Ancestry excludes the segments under 7 which would bring down the match size and then TIMBR takes out pile up areas. I am working with a genetic genealogist and he recommends transfer for both kits to Gedmatch if there is any possibility of solving the match..
--
Sarah L Meyer
Georgetown TX
ANK(I)ER, BIGOS, KARMELEK, PERLSTADT, STOKFISZ, SZPIL(T)BAUM, Poland
BIRGARDOVSKY, EDELBERG, HITE (CHAIT), PERCHIK Russia (southern Ukraine) and some Latvia or Lithuania
https://www.sarahsgenies.com
--
Sarah L Meyer
Georgetown TX
ANK(I)ER, BIGOS, KARMELEK, PERLSTADT, STOKFISZ, SZPIL(T)BAUM, Poland
BIRGARDOVSKY, EDELBERG, HITE (CHAIT), PERCHIK Russia (southern Ukraine) and some Latvia or Lithuania
https://www.sarahsgenies.com
R Jaffer
Please don't discount MH results. I have a predicted 3-5 cousin match with whom I share 62.3 cM over 6 segments with the largest being 17.4 cM. All the segments are over 7 cM. I never looked at the match nor contacted her because of the level of match. However, I saw her connection to my first cousin at 102cM over 7 segments with the largest being 21.5 cM. His predicted relationship was
1st cousin twice removed - 3rd cousin once removed. This sparked my interest especially because I saw a shared surname. When contacting this match, I learned that the family had no idea of where the immigrant had been born because he had been taken as a cantonist at about nine years of age. This match and I are third cousins. I have the revision lists that have her great grandfather as the older brother of my great grandfather in 1848, and in 1854 he was reported to have been taken in 1852. The problem with so many matches is that often they go through women for whom we don't know surnames, and thus will never be able to determine a mutual ancestor. This match who was born in the US never tested at Ancestry because she lives abroad. Therefore, I don't know how Ancestry would report the match.
Roberta Jaffer
Massachusetts
I appreciate all the answers, received here and in private, explaining some of the mechanics behind the discrepancy in DNA match data from MyHeritage and Ancestry. It seems that the majority of responses report that MyHeritage combine smaller segments and includes questionable segments, inflating their totals. Ancestry, on the other hand, eliminates smaller segments, pile-up areas and other questionable segments, to provide a more conservative (and realistic?) report, according to these responses. Many suggested comparing matches on GEDMatch as a more reliable platform.
As it turns out, I have collected match data from several different sources into a table, so I decided to check into this. I don't have a lot of instances where someone tested on MyHeritage and Ancestry to be a representative sample, but I thought this was a start.
Karen total # largest
Ancestry 110 11 22
GedMatch 110 22
Mildred* total # largest
MyHeritage 124 6 31.7
GedM 126.4 29.7
Robert* total # largest
An 168 16 28
GedM 155.8 29.3
Jane total # largest
MH 664.6 20 110.3
FTDNA 656 16 119.75
Linda total # largest
An 124 16 18
FTDNA 116 12 21
G.A. total # largest
An 172 14 27
MH 146.2 10 25.5
Mike * total # largest
An 168 9 72
MH 148 7 69.9
GM 155.7 47.1
Penny * total # largest
An 179 14 41
MH 174.7 13 32.4
GM 153.5 38.1
Janet ** total # largest
MH 350.9 17 60.2
FT 363 16 66
Alan ** total # largest
An 598 15 67
MH 590.7 18 109.1
Nancy * total # largest
FT 83 6 36
GM 82.7 24.6
* known 3rd cousin
** known 2nd cousin
From this small sample, I don't see a pattern that makes a case that Ancestry is more conservative or MyHeritage inflates its reports. MyHeritage's numbers are close to and sometimes lower than Ancestry's. MyHeritage's numbers are also consistent with GEDMatch and FamilyTreeDNA in the few cases where 3rd party comparisons were available. I definitely don't see anything here that would explain a 4x discrepancy in the two match reports I posted originally. Does anyone have another explanation?
As it turns out, I have collected match data from several different sources into a table, so I decided to check into this. I don't have a lot of instances where someone tested on MyHeritage and Ancestry to be a representative sample, but I thought this was a start.
Karen total # largest
Ancestry 110 11 22
GedMatch 110 22
Mildred* total # largest
MyHeritage 124 6 31.7
GedM 126.4 29.7
Robert* total # largest
An 168 16 28
GedM 155.8 29.3
Jane total # largest
MH 664.6 20 110.3
FTDNA 656 16 119.75
Linda total # largest
An 124 16 18
FTDNA 116 12 21
G.A. total # largest
An 172 14 27
MH 146.2 10 25.5
Mike * total # largest
An 168 9 72
MH 148 7 69.9
GM 155.7 47.1
Penny * total # largest
An 179 14 41
MH 174.7 13 32.4
GM 153.5 38.1
Janet ** total # largest
MH 350.9 17 60.2
FT 363 16 66
Alan ** total # largest
An 598 15 67
MH 590.7 18 109.1
Nancy * total # largest
FT 83 6 36
GM 82.7 24.6
* known 3rd cousin
** known 2nd cousin
From this small sample, I don't see a pattern that makes a case that Ancestry is more conservative or MyHeritage inflates its reports. MyHeritage's numbers are close to and sometimes lower than Ancestry's. MyHeritage's numbers are also consistent with GEDMatch and FamilyTreeDNA in the few cases where 3rd party comparisons were available. I definitely don't see anything here that would explain a 4x discrepancy in the two match reports I posted originally. Does anyone have another explanation?
Lee David Jaffe
===============
Surnames / Towns: Jaffe / Suchowola, Poland ; Stein (Sztejnsapir) / Bialystok and Rajgrod ; Roterozen / Rajgrod ; Joroff (Jaroff, Zarov) / Chernigov, Ukraine ; Schwartz (Schwarzstein) / Ternivka, Ukraine ; Weinblatt / Brooklyn, Perth Amboy, NJ ; Koshkin / Snovsk, Ukraine ; Rappoport / ? ; Braun / Wizajny, Suwalki, Ludwinowski / Wizajny, Suwalki
superimmune@...
Lee,
I think it would be inaccurate to say that Ancestry is any more accurate or reliable than MyHeritage. My personal experience is that Ancestry tends to under-estimate the closeness of relationships, while MyHeritage tends to over-estimate. Both tend to be most accurate for the closest relationships, and have more difficulty with the more distant ones. The problem with Ancestry is that it can sometimes be too conservative in terms of erring on the side of caution, and consequently I have some distant cousins who I match to on every platform except for Ancestry.com, which has missed the match entirely. I am convinced that these people who match me on every other platform are NOT false matches, and actually are related to me, although how I've figured this out is really too long a story to get into here. From my perspective, the bottom line is that while you are more likely to get a false match on MyHeritage than on Ancestry, with Ancestry you are more likely to have a true match that doesn't show up at all.
Hope this helps,
Sincerely,
Claudia Bullock
I think it would be inaccurate to say that Ancestry is any more accurate or reliable than MyHeritage. My personal experience is that Ancestry tends to under-estimate the closeness of relationships, while MyHeritage tends to over-estimate. Both tend to be most accurate for the closest relationships, and have more difficulty with the more distant ones. The problem with Ancestry is that it can sometimes be too conservative in terms of erring on the side of caution, and consequently I have some distant cousins who I match to on every platform except for Ancestry.com, which has missed the match entirely. I am convinced that these people who match me on every other platform are NOT false matches, and actually are related to me, although how I've figured this out is really too long a story to get into here. From my perspective, the bottom line is that while you are more likely to get a false match on MyHeritage than on Ancestry, with Ancestry you are more likely to have a true match that doesn't show up at all.
Hope this helps,
Sincerely,
Claudia Bullock
David Brostoff
On Nov 21, 2022, at 6:24 PM, Lee Jaffe <leejaffe54@...> wrote:
David Brostoff
When I found the match between my mother and Mary on Ancestry, the match data was so radically different – 21cM/7segments/ largest 15cM – that I'm stumped to understand it. Nor do I see any known shared matches, even though one of those with triangulated matches at MyHeritage also tested at Ancestry.Besides the large discrepancy in total shared DNA, the fact that on Ancestry your mother and Mary do not share one of their MyHeritage triangulated matches makes me ask whether you are certain that the Mary on Ancestry is the same as the one on MyHeritage.
David Brostoff
On Sun, Nov 27, 2022 at 06:00 PM, David Brostoff wrote:
I guess it is possible that the much smaller match between Mary and my mother on Ancestry might result to an even smaller match with those who have shared matches on MyHeritage, small enough to fall below Ancestry's threshold. Note that the 25cM largest segment on dropped to a 15cM largest segment on Ancestry. Since the shared matches were from my generation, I'd expect the match numbers to be about half. If Ancestry excludes segments smaller than 7, that might explain the missing triangulated matches.
Of course, along the lines suggested by your question, if the matches are so radically different, what are the chances that something more serious went wrong?
Besides the large discrepancy in total shared DNA, the fact that on Ancestry your mother and Mary do not share one of their MyHeritage triangulated matches makes me ask whether you are certain that the Mary on Ancestry is the same as the one on MyHeritage.I've been in touch with Mary, initially through MyHeritage messages, and she pointed me to her test on Ancestry. And we've compared the matches.
I guess it is possible that the much smaller match between Mary and my mother on Ancestry might result to an even smaller match with those who have shared matches on MyHeritage, small enough to fall below Ancestry's threshold. Note that the 25cM largest segment on dropped to a 15cM largest segment on Ancestry. Since the shared matches were from my generation, I'd expect the match numbers to be about half. If Ancestry excludes segments smaller than 7, that might explain the missing triangulated matches.
Of course, along the lines suggested by your question, if the matches are so radically different, what are the chances that something more serious went wrong?
Lee David Jaffe
===============
Surnames / Towns: Jaffe / Suchowola, Poland ; Stein (Sztejnsapir) / Bialystok and Rajgrod ; Roterozen / Rajgrod ; Joroff (Jaroff, Zarov) / Chernigov, Ukraine ; Schwartz (Schwarzstein) / Ternivka, Ukraine ; Weinblatt / Brooklyn, Perth Amboy, NJ ; Koshkin / Snovsk, Ukraine ; Rappoport / ? ; Braun / Wizajny, Suwalki, Ludwinowski / Wizajny, Suwalki