At RootsTech, MyHeritage introduced cM Explainer, a new tool for all of their DNA customers that utilizes both the total matching cMs (centimorgans) plus the ages of the people involved, if provided by the customers, to estimate the relationship possibilities between two matches.
According to the MyHeritage blog article, here:
DNA Matches are characterized by the amount of DNA shared between two individuals, measured using a unit of genetic distance called centimorgans (cM). cM Explainer™ is unique in the way it uses both the centimorgan value as well as the ages of the two individuals (if known) to fine-tune its predictions, making MyHeritage the only major genealogy company to offer relationship prediction at this level of granularity and accuracy.
cM Explainer™ is fully integrated into the MyHeritage platform to shed light on any DNA Match found on MyHeritage, and is also available as a free standalone tool to benefit individuals who have tested with other DNA services.
Using cM Explainer
cM Explainer is automatically implemented for all MyHeritage DNA customers, so there’s nothing to do except utilize the tool in conjunction with the additional DNA tools already provided by MyHeritage.
Just click on DNA Matches if you’re a DNA customer, or under DNA Tools if you want to use the cM Explainer standalone tool.
Let’s look at my matches.
As you can see, two of my matches have provided their ages which appear in their match information.
The new cM Explainer probable relationship is listed as well. In Charlene’s case, she’s predicted to be a half first cousin, and Cheryl is predicted to be either a half first cousin, or a parent’s 1st cousin, which is another way of saying first cousin once removed.
Recall that “once removed” means one side of the descendant tree is one generation longer than the other. Cheryl and my Mom are first cousins (1C) and Cheryl and I are first cousins once removed (1C1R) or, said another way, I’m Cheryl’s first cousin’s daughter.
Some matches receive two listed “Probable Relationships,” but everyone can view additional estimates.
Click on the purple “Review DNA Match” button to view detail information.
My match’s segment information is provided, in addition to the possible relationships, in order of most probable first. To see additional information, click on the “show more relationships” link.
Charlene has a total of 5 possible relationships listed, each with its own probability calculated. One of my matches has a total of 8 possible relationships displayed.
Next, you’ll see the diagram of possible relationships.
Don’t forget to click on the “Relationships” dropdown in the upper right corner of the diagram.
You can click on Full relationships, Half relationships, or All Relationships.
I clicked on “all” which displays everything together.
Clicking the “Show probabilities for MRCA” box in the upper left-hand corner adds the probability that you and your match descend from a specific generation, or MRCA (most recent common ancestor.)
The highest or best probable relationship for cousin Charlene is calculated as 51.8% half first cousin.
The other possibilities are less likely. The second most likely is “Parent’s first cousin,” at 24.3%.
Charlene is my first cousin once removed (1C1R,) at the bottom. Stated another way, Charlene is my first cousin’s child, calculated at 4.5%, which should be genetically equivalent to a half first cousin at 51.8%. As you can see in the chart above, there’s VERY large probability difference between those two, which may be because of the expected comparative ages of the people in those positions involved.
Let’s take a minute to look at how half and “removed” relationships work genetically.
Half and “Removed” Relationships
In this example, John was married twice, to Mary and Sue. John had son Jim with Mary, and both daughters, Anna and Bonnie, with Sue. Their descendants took DNA tests.
In this chart, you can see that half-relationships of any kind carry half the average expected shared DNA as the full version of the same relationship. The yellow people, descendants of John and Mary, are half relationships to the green people because John was married to both Mary and Sue, having children with both wives.
The green people descend from full siblings, Anna and Bonnie, the children of Sue.
In the first generation, Jim and Anna are half siblings and share about 25% of their DNA. Anna and Bonnie are full siblings and share about 50% of their DNA. By extension, of course, Jim and Bonnie are half siblings too, sharing approximately 25% of their DNA, but not the exact same DNA as Jim and Anna share.
In the next generation, Jordan and Andrew are half first cousins and share about 6.25% of their DNA, while Andrew and Brad are full first cousins and share about 12.5% of their DNA.
Below the second cousin level, some cousins won’t match each other, but that doesn’t mean they aren’t cousins. It only means they didn’t happen to inherit a common segment of DNA from their common ancestors.
At the fourth-generation level, Jeremy and Abraham are half third cousins and share less than 1% or about 26.56 cM of their DNA, while Abraham and Betty are full third cousins and share about 53.13 cM of their DNA.
That half division of DNA occurred several generations earlier because Jim and Anna are half siblings which means that they only share half as much DNA as full siblings Anna and Bonnie. Of course, each subsequent generation will be a half relationship, and share roughly half as much as the full equivalent of that same relationship.
However, when the generations are offset by one, or once removed (1R,) the DNA is halved again. Looking at the chart again, half third cousins (3C,) Jeremy and Abraham share about 0.39% or about 25.56 cM of their DNA. Abraham and Beverly, who are 3C1R (third cousins once removed) are ALSO expected to share about 25.56 cM, the same amount of their DNA. In this comparison, the halving occurs in the last generation by the generational offset, when comparing Abraham with Beverly.
Of course, Jeremy and Beverly share the smallest percentage of all, because they are Half third cousins once removed, so they would be expected to share only about 13.28 cM of their DNA, assuming they share any at all.
I wrote about the various percentages expected of each relationship level and compiled a comprehensive chart in this article.
Of course, MyHeritage has included the factor of age to attempt to refine the relationship more succinctly.
How Accurate is cM Explainer?
I created a chart of my closest matches who are known, proven relatives.
Results where My Heritage has provided exact, accurate predictions are shown in red.
My first thought when I saw this new tool was that all of the people with whom I shared a Theory of Family Relativity (TOFR), especially relationships I had confirmed (or at least not rejected) would be predicted in cM Explainer to be that same relationship. Well, I was wrong.
Of the 8 matches with whom I have an accurate TOFR, the relationships of 4, or 50%, are correct, but the other 4 are not, so clearly, MyHeritage is not relying on TOFRs for cM Explainer, at least not solely, if at all.
|Person||Total cM||# Segments||Actual Relationship||TOFR||MyHeritage cM Explainer|
|Alberta||744.2||25||Half niece *1 (2nd on list)||Y||1C|
|Dana||521.8||17||Half 1C1R (not on list)||N||1st C dau, half 1C|
|Charlene||477.5||24||1C1R *2 (4th on list)||N||Half 1C|
|Cheryl||477.2||23||Parent’s 1C (2nd on list)||N||Half 1C, parent’s 1C|
|David||460.4||17||2C (3rd on list)||N||Half 1C|
|Buster||409.9||16||1C1R *3||N||Parent’s 1C|
|Donald||381.7||17||1C1R (3rd on list)||N||2C|
|Kurt||378.9||16||Half great-nephew (not on list)||N||2C|
|Teresa||330.4||13||Half great-nephew (not on list)||Y||2C|
|Shirley||223.3||8||2C1R (2nd on list)||Y||2C|
|Sydney||217.8||10||Half great-nephew (not on list)||N||2C dau, 1C dau|
|Amos||182.7||8||1C2R (8th on list)||N||2C son|
|Denny||166.9||6||3C (2nd on list)||N||2C|
|Carol||87||7||2C1R (2nd on list)||N||3C|
*1 – Half aunt/uncle is equivalent to half niece/nephew – it’s simply a matter of perspective.
*2 – 1C1R (first cousin once removed) is the same relationship as a first cousin’s child, just said differently.
*3 – Your parent’s first cousin in your first cousin once removed (1C1R.)
In the Actual Relationship column, I’ve indicated the actual relationship, then if the actual relationship is shown on the chart of possibilities provided by MyHeritage, and if so, at which position.
For example, I’ve listed Alberta’s “Actual Relationship” as my half-niece. Additionally, there’s a comment at *1 below the chart. MyHeritage predicted Alberta as my first cousin, but the correct half-niece designation is shown second on the list.
In this case, I’m fairly sure I know exactly why the relationship miscalculation occurred. Alberta’s mother, my half-sister, was born to my father’s first wife. My mother was 22 years younger than my father, so my mother is roughly the same age as my half-sister. I am the same age as my half-sister’s oldest children. Therefore, we have an unusual generational difference where ages might be misleading.
In the second position, MyHeritage estimated Alberta as half-aunt, which is the same as half niece, depending on whose perspective you’re speaking from, so cM Explainer was close. In normal circumstances, 1st Cousin is probably the most likely relationship although having children separated by two decades certainly is not unheard of.
MyHeritage correctly predicted 6 of 19 relationships, for 31.6% accuracy.
The correct relationship was on the relationship list most of the time, but was omitted entirely 4 times. The common factor in the entirely missing relationships is that they are all half-relationships. While they were not all from the same family line, they did all involve long generations, meaning children born over a very long period. That’s not uncommon with half-siblings, and half relationships are notoriously difficult to sort from other candidate relationships. These situations might possibly be considered statistical outliers.
A half first cousin should be genetically equivalent to a first cousin, once removed, based on the amount of expected DNA for those relationships.
However, in at least one case, these two relationships are calculated with different resulting probabilities. Half first cousin is 41.5%, and 1st cousins child (aka 1C1R or 1st cousin once removed) is 43.8%.
Keep in mind, though that MyHeritage is using the age of the two individuals in their calculations, which could alter the results based on the combination of factors calculated into the probabilities. It’s 85% likely that the match is one of those two relationships.
I’m interested in your thoughts on this new tool. How does it work for you? What about endogamy or pedigree collapse? Do you find it useful? How are you utilizing it in your research?
Shortly, I’ll do a comparison article with other tools to see how the publicly available cM Explainer tool stacks up against the rest.
Thanks to MyHeritage for making this tool free for all to use, here.
Follow DNAexplain on Facebook, here or follow me on Twitter, here.
Share the Love!
You’re always welcome to forward articles or links to friends and share on social media.
If you haven’t already subscribed (it’s free,) you can receive an email whenever I publish by clicking the “follow” button on the main blog page, here.
You Can Help Keep This Blog Free
I receive a small contribution when you click on some of the links to vendors in my articles. This does NOT increase the price you pay but helps me to keep the lights on and this informational blog free for everyone. Please click on the links in the articles or to the vendors below if you are purchasing products or DNA testing.
Thank you so much.
DNA Purchases and Free Uploads
- FamilyTreeDNA – Y, mitochondrial and autosomal DNA testing
- MyHeritage DNA – Autosomal DNA test
- MyHeritage FREE DNA file upload – Upload your DNA file from other vendors free
- AncestryDNA – Autosomal DNA test
- 23andMe Ancestry – Autosomal DNA only, no Health
- 23andMe Ancestry Plus Health
Genealogy Products and Services
- MyHeritage FREE Tree Builder – Genealogy software for your computer
- MyHeritage Subscription with Free Trial
- Legacy Family Tree Webinars – Genealogy and DNA classes, subscription-based, some free
- Legacy Family Tree Software – Genealogy software for your computer
- Newspapers.com – Search newspapers for your ancestors
- NewspaperArchive – Search different newspapers for your ancestors
- DNA for Native American Genealogy – by Roberta Estes, for those ordering the e-book from anyplace, or paperback within the United States
- DNA for Native American Genealogy – for those ordering the paperback outside the US
- Genealogical.com – Lots of wonderful genealogy research books
- Legacy Tree Genealogists – Professional genealogy research
Roberta: The percentages in the chart of half and removed relationships seem wrong to me. I agree that siblings share 50% of their DNA, but first cousins do not share 25%, they share closer to 12.5%. Similarly, the half relationships below half sibling are also incorrect.
You are exactly right. This is why I should never work late at night. Thank you and I’ve corrected the chart and surrounding text.
Is there a place to put the actual known relationship? It says one of my daughters is my mother the other two are correct and my grandmother my aunt. Thanks!
No, there isn’t. Feedback would be a good idea and maybe help refine the tool.
Thanks as always for your very informative newsletter.
I’m deeply concerned about the way that MH are using an absolute relationship tag, with little thought to the implications of this on some vulnerable users of the site.
Some folk are adoptees, the children of adoptees and/or survivors of child protection interventions. I am a social worker and have a situation currently where this is the causing me deep concern.
I’m don’t think that this is the correct place to discuss the issue, as the story is not mine, but I do feel that it needs to be addressed urgently.
I see a lot of possible relationships, so maybe I don’t understand what you mean. Please contact MyHeritage with your concerns.
Roberta, I am trying to discern the relationship to me, 2 matches I have from Ancestry, who I know are brother and sister to each other. Using my age 73, his age 37 and cM’s shared 832, her age 43 and 743 cMs shared, he is a half nephew and she is half niece or first cousin’s child. Both siblings match all my maternal and paternal matches on Ancestry. What I don’t understand about the relationship is even if I had a half sibling who gave up 2 children in the early 80s to one family, how can they match both my mother and father.and be half anything. Apparently, at least the sister has no idea about our relationship either. Does any of this make sense? I have banged my head about this for several years and hoped this would help me get a better idea but I’m still confused.
It looks like the tool got the correct relationship in some cases and not in others. I’d be interested in seeing the average probability assigned to the correct relationship, and as a bonus it’d be nice to see the same statistic for the predictions without ages.
I recently tested 125 known data points in this way both with and without the number of segments. It was interesting to see that the number of segments nearly doubled the average probability given to the correct relationship.
This tool is not helpful for distant relationships. For example, if you plug in 1 cM it gives you a 4th cousin as your likely match. I addressed some of these issues with the creator of the tool and argued that the Shared cM Tool is a better predictor when it comes to matches going beyond 4-5 generations, even though it gives cumulative probability for those. The creator argued that at some point beyond 4th cousins the shared DNA is more a result of mutation rather than anything else and that it is nearly impossible to tell a real match from IBS. Still doesn’t prove false the possibility that say a 15 cM can be legitimate, but way beyond 4th cousins. Ancestry DNA even includes a percentage amount denoted as “distant relationship”, and has tons of cousins shown as possible 5th, 6th, 7th, or 8th cousins.
Someone left a comment on the tool creator’s YouTube page that the explainer gave a 4th cousin prediction to a match that was a verified 13th cousin sharing a whooping 38 cM.
I am trying to learn a lot about my DNA history and meet new people on my biological tree. Based on this tool I have over 1,000 matches listed as probable 4th cousins. I just have a hard time believing that most of these people don’t go beyond my 3rd great grandparents as MRCA, especially because they come from completely different places and/or unrelated countries of origin.
If the match is low confidence, then probabilities should be adjusted accordingly, because it is not just about shared DNA. This tool also ignores the largest segment as a great indicaror to tell some distant relationships apart. All in all, it looks like the tool is best for 2nd cousins or closer.
Well, first I like the way MyHeritage has added the prediction tool. Very Handy ! and easy to understand for newer DNA testers I think. That said, as you point out, there are exceptions.
I have a large group of fairly extreme exceptions that I have been working on as a side project for the last month. It is breakthrough for me. I had many 6th level cousins (in 5 databases) who always predicted as about 3rd Cousins.
I vaguely remembered reading about “large single persistent segments” some time ago. I remember being warned that these could be much further back in time.
I have one match ( on one 69 cm segment ) who so wonderfully, has data on all 5 databases. I have puzzled this for quite a while. But not that I am triangulating these matches in the various databases ( except one, of course ) I have seen the light. The matches are valid. The MCRA for many that I can place in trees are all 6C, 6C1r, 5C1r, 5C25, etc. The MCRA are ancestors born in about 1750.
I have many 6th cousin matches with total cM between 40-92 cM. The largest single shared segments so far are 54 and 55 cM on 23&Me. The large segments occur on Chr 13.
For example, if I enter 71 cM into the DNAPainter CM Tool – The footnote says.
“† this relationship has a positive probability for 71cM in thednageek’s table of probabilities, but falls outside the bounds of the recorded cM range (99th percentile)”. It is the same for all with a big segment piece on Chr 13.
I could change the range of the recorded data in a hurry. But I am holding off until I finish my study.
Correction – the sentence
“But not that I am triangulating these matches in the various databases ( except one, of course ) I have seen the light” should have read “But NOW that I am triangulating …..
I don’t know if anyone else has had this problem with this tool. The Relationship Prediction Tool has gone haywire on my mother’s DNA kit. My mother just turned 86 years old today and my oldest sister and oldest brother are both in their sixties. This prediction tool has now listed my oldest sister as our mother’s mother. My oldest brother is now listed as our mother’s father, while I am still listed as her daughter. My son is listed as her nephew instead of her grandson. If I look at my oldest sister’s DNA kit, my mother is listed as her daughter and our brother is listed correctly as our brother and vice versus. There is no major differences between us siblings on shared DNA that could explain why this algorithm decided to make them the parents and keep me the daughter. When I say no major difference I mean 1.5 cms difference. I have never seen anything like this before. I have put in a ticket with My Hertiage earlier this past week and so far I have not heard from them. I have no way to go in and make any corrections on my family’s relationship. At least at Ancestry they do give you an option to select a relationship where as My Heritage just gives you one without any feedback from the customer.
All the reviews I see on this tool miss the usefulness of visualizing the predicted relationships and using comprehensible descriptions like “parent’s first cousin” and instead jump to complaining about the missed predictions. Just helping users grasp the possible relationships and using plain language instead of 2C1R is a big improvement imho. MH is like apple to Ancestry’s microsoft
I am sure the tool is very useful when it is working correctly. For me it is not working correctly. For it to change the relationships for my known immediate family and to give no way to correct the mess it has created is a legitimate grip. I was thrilled when I first saw it on my DNA match list but not so much once looking at my mother’s and siblings match list. I reached out to My Heritage over a week ago with no response nor any fixes. If that is how Apple performs, I think I rather stick with Mircosoft.