Recently, a reader asked if I had any crossover statistics.
They were asking about the number of crossovers, meaning divisions on each chromosome, of the parent’s DNA when a child is created. In other words, how many segments of your maternal and paternal grandparent’s DNA do you inherit from your mother and father – and are those numbers somehow different?
Why would someone ask that question, and how is it relevant for genealogists?
What is a Crossover and Why is it Important?
We know that every child receives half of their autosomal DNA from their father, and half from their mother. Conversely that means that each parent can only give their child half of their own DNA that they received from their parents. Therefore, each parent has to combine some of the DNA from their father’s chromosome and their mother’s chromosome into a new chromosome that they contribute to their child.
Crossovers are breakpoints that are created when the DNA of the person’s parents is divided into pieces before being recombined into a new chromosome and passed on to the person’s child.
I’m going to use the following real-life scenario to illustrate.
The colors of the people above are reflected on the chromosome below where the DNA of the blue daughter, and her red and green parents are compared to the DNA of the tester. The tester is shown as the gray background chromosomes in the chromosome browser. The backgroud person is whose results we are looking at.
My granddaughter has tested her DNA, as have her parents and 3 of her 4 grandparents along with 2 great-grandparents, shown as red and green in the diagram above.
Here’s an example utilizing the FamilyTreeDNA chromosome browser.
On my granddaughter’s chromosome 1, on the chromosome brower above, we see two perfect examples of crossovers.
There’s no need to compare her DNA against that of her parent, the son in the chart above, because we already know she matches the full length of every chromosome with both of her parents.
However, when comparing my granddaughter’s DNA against the grandmother (blue) and her grandmother’s parents, the great-grandmother shown in red and great-grandfather shown in green, we can see that the granddaughter received her blue segments from the grandmother.
The grandmother had to receive that entire blue segment from either her mother, in red, or her father, in green. So, every blue segment must have an exactly matching red segment, green segment or combination of both.
The first red box at left shows that the blue segment was inherited partially from the grandmother’s red mother and green father. We know that because the tester matches the red great-grandmother on part of that blue segment and the green great-grandfather on a different part of the entire blue segment that the tester inherited from her blue grandmother.
The middle colored region, not boxed, shows the entire blue segment was inherited from the red great-grandmother and the blue grandmother passed that intact through her son to her granddaughter.
The third larger red boxed area encompassing the entire tested region to the right of the centromere was inherited by the granddaughter from her grandmother (blue segment) but it was originally from the blue grandmother’s red mother and green father.
The Crossover
The areas on this chromosome where the blue is divided between the red and green, meaning where the red and green butt up against each other is called a crossover. It’s literally where the DNA of the blue daughter crosses over between DNA contributed by her red mother and green father.
In other words, the crossover where the DNA divided between the blue grandmother’s parents when the grandmother’s son was created is shown by the dark arrows above. The son gave his daughter that exact same segment from his mother and it’s only by comparing the tester’s DNA against her great-grandparents that we can see the crossover.
What we’re really seeing is that the segments inherited by the grandmother from her parents two different chromosomes were combined into one segment that the grandmother gave to her son. The son inherited the green piece and the red piece on his maternal chromosome, which he gave intact to his daughter, which is why the daughter matches her grandmother on that entire blue segment and matches her great-grandparents on the red and green pieces of their individual DNA.
Inferred Matching Segments
The entirely uncolored regions are where the tester does not match her blue grandmother and where she would match her grandfather, who has not tested, instead of her blue grandmother.
The testers father only received his DNA from his mother and father, and if his daughter does not match his mother, then she must match his untested father on that segment.
Looking at the Big Inheritance Picture
The tester’s full autosomal match between the blue grandmother, red great-grandmother and green great-grandfather is shown below.
In light of the discussion that follows, it’s worth noting that chromosomes 4 and 20 (orange arrows) were passed intact from the blue grandmother to the tester through two meiosis (inheritance) events. We know this because the tester matches the green great-grandfather’s DNA entirely on these two chromosomes that he passed to his blue daughter, her son and then the tester.
Let’s track this for chromosomes 4 and 20:
- Meiosis 1 –The tester matches her blue grandmother, so we know that there was no crossover on that segment between the father and the tester.
- Meiosis 2 – The tester matches her green great-grandfather along the entire chromosome, proving that it was passed intact from the grandmother to the tester’s father, her son.
- What we don’t know is whether there were any crossovers between the green great-grandfather when he passed his parent or parents DNA to the blue grandmother, his daughter. In order to determine that, we would need at least one of the green great-grandfather’s parents, which we don’t have. We don’t know if the green great-grandfather passed on his maternal or paternal copy of his chromosome, or parts of each to the blue great-grandmother, his daughter.
Meiosis Events and the Tree
So let’s look at these meiosis or inheritance events in a different way, beginning at the bottom with the pink tester and counting backwards, or up the tree.
By inference, we know that chromosomes 11, 16 and 22 (purple arrows) were also passed intact, but not from the blue grandmother. The tester’s father passed his father’s chromosome intact to his daughter. That’s the untested grandfather again. We know this because the tester does not match her blue grandmother at all on either of these three chromosomes, so the tester must match her untested grandfather instead, because those are the only two sources of DNA for the tester’s father.
A Blip, or Not?
If you’ve noticed that chromosome 14 looks unusual, in that the tester matches her grandmother’s blue segment, but not either of her great-grandparents, which is impossible, give yourself extra points for your good eye.
In this case, the green great-grandfather’s kit was a transfer kit in which that portion of chromosome 14 was not included or did not read accurately. Given that the red great-grandmother’s kit DID read in that region and does not match the tester, we know that chromosome 14 would actually have a matching green segment exactly the size of the blue segment.
However, in another situation where we didn’t know of an issue with the transfer kit, it is also possible that the granddaughter matched a small segment of the blue grandmother’s DNA where they were identical by chance. In that case, chromosome 14 would actually have been passed to the tester intact from her father’s father, who is untested.
Every Segment has a Story
Looking at this matching pattern and our ability to determine the source of the DNA back several generations, originating from great-grandparents, I hope you’re beginning to get a sense of why understanding crossovers better is important to genealogists.
Every single segment has a story and that story is comprised of crossovers where the DNA of our ancestors is combined in their offspring. Today, we see the evidence of these historical genetic meiosis or division/recombination events in the start and end points of matches to our genetic cousins. Every start and end point represents a crossover sometime in the past.
What else can we tell about these events and how often they occur?
Of the 22 autosomes, not counting the X chromosome which has a unique inheritance pattern, 17 chromosomes experienced at least one crossover.
What does this mean to me as a genealogist and how can I interpret this type of information?
Philip Gammon
You may remember our statistician friend Philip Gammon. Philip and I have collaborated before authoring the following articles where Philip did the heavy lifting.
I discussed crossovers in the article Concepts – DNA Recombination and Crossovers, also in collaboration with Philip, and showed several examples in a Four Generation Inheritance Study.
If you haven’t read those articles, now might be a good time to do so, as they set the stage for understanding the rest of this article.
The frequency of chromosome segment divisions and their resulting crossovers are key to understanding how recombination occurs, which is key to understanding how far back in time a common ancestor between you and a match can expect to be found.
In other words, everything we think we know about relationships, especially more distant relationships, is predicated on the rate that crossovers occur.
The Concepts article references the Chowdhury paper and revealed that females average about 42 crossovers per child and males average about 27 but these quantities refer to the total number of crossovers on all 22 autosomes and reveal nothing about the distribution of the number of crossovers at the individual chromosome level.
Philip Gammon has been taking a closer look at this particular issue and has done some very interesting crossover simulations by chromosome, which are different sizes, as he reports beginning here.
Crossover Statistics by Philip Gammon
For chromosomes there is surprisingly little information available regarding the variation in the number of crossovers experienced during meiosis, the process of cell division that results in the production of ova and sperm cells. In the scientific literature I have been able to find only one reference that provides a table showing a frequency distribution for the number of crossovers by chromosome.
The paper Broad-Scale Recombination Patterns Underlying Proper Disjunction in Humans by Fledel-Alon et al in 2009 contains this information tucked away at the back of the “Supplementary methods, figures, and tables” section. It was likely not produced with genetic genealogists in mind but could be of great interest to some. The columns X0 to X8 refer to the number of crossovers on each chromosome that were measured in parental transmissions. Separate tables are shown for male and female transmissions because the rates between the two sexes differ significantly. Note that it’s the gender of the parent that matters, not the child. The sample size is quite small, containing only 288 occurrences for each gender.
A few years ago I stumbled across a paper titled Escape from crossover interference increases with maternal age by Campbell et al 2015. This study investigated the properties of crossover placement utilising family groups contained within the database of the direct-to-consumer genetic testing company 23andMe. In total more than 645,000 well-supported crossover events were able to be identified. Although this study didn’t directly report the observed frequency distribution of crossovers per chromosome, it did produce a table of parameters that accurately described the distribution of inter-crossover distances for each chromosome.
By introducing these parameters into a model that I had developed to implement the equations described by Housworth and Stahl in their 2003 paper Crossover Interference in Humans I was able to derive tables depicting the frequency of crossovers. The following results were produced for each chromosome by running 100,000 simulations in my crossover model:
Transmissions from female parent to child, above.
Transmissions from male parent to child.
To be sure that we understand what these tables are revealing let’s look at the first row of the female table. The most frequent outcome for chromosome #1 is that there will be three crossovers and this occurs 27% of the time. There were instances when up to 10 crossovers were observed in a single meiosis but these were extremely rare. Cells that are blank recorded no observations in the 100,000 simulations. On average there are 3.36 crossovers observed on chromosome #1 in female to child transmissions i.e. the female chromosome #1 is 3.36 Morgans (336 centimorgans) in genetic length.
Blaine Bettinger has since examined crossover statistics using crowdsourced data in The Recombination Project: Analyzing Recombination Frequencies Using Crowdsourced Data, but only for females. His sample size was 250 maternal transmissions and Table 2 in the report presents the results in the same format as the tables above. There is a remarkable degree of conformity between Blaine’s measurements and the output from my simulation model and also to the earlier Fledel-Alon et al study.
The diagrams below are a typical representation of the chromosomes inherited by a child.
The red and orange (above) are the set of chromosomes inherited from the mother and the aqua and green (below) from the father. The locations where the colours change identify the crossover points.
It’s worth noting that all chromosomes have a chance of being passed from parent to child without recombination. These probabilities are found in the column for zero crossovers.
In the picture above the mother has passed on two red chromosomes (#14 and #20) without recombination from one of the maternal grandparents. No yellow chromosomes were passed intact.
Similarly, below, the father has passed on a total of five chromosomes that have no crossover points. Blue chromosomes #15, #18 and #21 were passed on intact from one paternal grandparent and green chromosomes #4 and #20 from the other.
It’s quite a rare event for one of the larger chromosomes to be passed on without recombination (only a 1.4% probability for chromosome #1 in female transmissions) but occurs far more frequently in the smaller chromosomes. In fact, the male chromosome #21 is passed on intact more often (50.6% of the time) than containing DNA from both of the father’s parents.
However, there is nothing especially significant about chromosome #21.
The same could be said for any region of similar genetic length on any of the autosomes i.e. the first 52 cM of chromosome #1 or the middle 52 cM of chromosome #10 etc. From my simulations I have observed that on average 2.8 autosomes are passed down from a mother to child without a crossover and an average of 5.1 autosomes from a father to child.
In total (from both parents), 94% of offspring will inherit between 4 and 12 chromosomes containing DNA exclusively from a single grandparent. In the 100,000 simulations the child always inherited at least one chromosome without recombination.
Back to Roberta
If you have 3 generations who have tested, you can view the crossovers in the grandchild as compared to either one or two grandparents.
If the child doesn’t match one grandparent, even if their other grandparent through that parent hasn’t tested, you can certainly infer that any DNA where the grandchild doesn’t match the available grandparent comes from the non-tested “other” grandparent on that side.
Let’s Look at Real-Life Examples
Using the example of my 2 granddaughters, both of their parents and 3 of their 4 grandparents have tested, so I was able to measure the crossovers that my granddaughters experienced from all 4 of their grandparents.
| Maternal Crossovers | Granddaughter 1 | Granddaughter 2 | Average |
| Chromosome 1 | 6 | 2 | 3.36 |
| Chromosome 2 | 4 | 2 | 3.17 |
| Chromosome 3 | 3 | 2 | 2.71 |
| Chromosome 4 | 2 | 2 | 2.59 |
| Chromosome 5 | 2 | 1 | 2.49 |
| Chromosome 6 | 4 | 2 | 2.36 |
| Chromosome 7 | 3 | 1 | 2.23 |
| Chromosome 8 | 2 | 2 | 2.11 |
| Chromosome 9 | 3 | 1 | 1.95 |
| Chromosome 10 | 4 | 2 | 2.08 |
| Chromosome 11 | 3 | 0 | 1.93 |
| Chromosome 12 | 3 | 3 | 2.00 |
| Chromosome 13 | 1 | 1 | 1.52 |
| Chromosome 14 | 3 | 1 | 1.38 |
| Chromosome 15 | 4 | 1 | 1.44 |
| Chromosome 16 | 2 | 2 | 1.58 |
| Chromosome 17 | 2 | 2 | 1.53 |
| Chromosome 18 | 2 | 0 | 1.40 |
| Chromosome 19 | 2 | 1 | 1.18 |
| Chromosome 20 | 0 | 1 | 1.19 |
| Chromosome 21 | 0 | 1 | 0.74 |
| Chromosome 22 | 1 | 0 | 0.78 |
| Total | 56 | 30 | 41.71 |
Looking at these results, it’s easy to see just how different inheritance between two full siblings can be. Granddaughter 1 has 56 crossovers through her mother, significantly more than the average of 41.71. Granddaughter 2 has 30, significantly less than average.
The average of the 2 girls is 43, very close to the total average of 41.71.
Note that one child received 2 chromosomes intact from her mother, and the other received 3.
| Paternal Crossovers | Granddaughter 1 | Granddaughter 2 | Average |
| Chromosome 1 | 2 | 2 | 1.98 |
| Chromosome 2 | 3 | 2 | 1.85 |
| Chromosome 3 | 2 | 2 | 1.64 |
| Chromosome 4 | 0 | 1 | 1.46 |
| Chromosome 5 | 1 | 2 | 1.46 |
| Chromosome 6 | 2 | 1 | 1.41 |
| Chromosome 7 | 1 | 2 | 1.36 |
| Chromosome 8 | 1 | 1 | 1.23 |
| Chromosome 9 | 1 | 3 | 1.26 |
| Chromosome 10 | 3 | 2 | 1.30 |
| Chromosome 11 | 0 | 1 | 1.20 |
| Chromosome 12 | 1 | 1 | 1.32 |
| Chromosome 13 | 2 | 1 | 1.02 |
| Chromosome 14 | 1 | 0 | 0.97 |
| Chromosome 15 | 1 | 2 | 1.01 |
| Chromosome 16 | 0 | 1 | 1.02 |
| Chromosome 17 | 0 | 0 | 1.06 |
| Chromosome 18 | 1 | 1 | 0.98 |
| Chromosome 19 | 1 | 1 | 1.00 |
| Chromosome 20 | 0 | 0 | 0.99 |
| Chromosome 21 | 0 | 0 | 0.52 |
| Chromosome 22 | 0 | 0 | 0.63 |
| Total | 23 | 26 | 26.65 |
Granddaughter 2 had slightly more paternal crossovers than did granddaughter 1.
One child received 7 chromosomes intact from her father, and the other received 5.
| Chromosome | Granddaughter 1 Maternal | Granddaughter 1 Paternal |
| Chromosome 1 | 6 | 2 |
| Chromosome 2 | 4 | 3 |
| Chromosome 3 | 3 | 2 |
| Chromosome 4 | 2 | 0 |
| Chromosome 5 | 2 | 1 |
| Chromosome 6 | 4 | 2 |
| Chromosome 7 | 3 | 1 |
| Chromosome 8 | 2 | 1 |
| Chromosome 9 | 3 | 1 |
| Chromosome 10 | 4 | 3 |
| Chromosome 11 | 3 | 0 |
| Chromosome 12 | 3 | 1 |
| Chromosome 13 | 1 | 2 |
| Chromosome 14 | 3 | 1 |
| Chromosome 15 | 4 | 1 |
| Chromosome 16 | 2 | 0 |
| Chromosome 17 | 2 | 0 |
| Chromosome 18 | 2 | 1 |
| Chromosome 19 | 2 | 1 |
| Chromosome 20 | 0 | 0 |
| Chromosome 21 | 0 | 0 |
| Chromosome 22 | 1 | 0 |
| Total | 56 | 23 |
Comparing each child’s maternal and paternal crossovers side by side, we can see that Granddaughter 1 has more than double the number of maternal as compared to paternal crossovers, while Granddaughter 2 only had slightly more.
| Chromosome | Granddaughter 2 Maternal | Granddaughter 2 Paternal |
| Chromosome 1 | 2 | 2 |
| Chromosome 2 | 2 | 2 |
| Chromosome 3 | 2 | 2 |
| Chromosome 4 | 2 | 1 |
| Chromosome 5 | 1 | 2 |
| Chromosome 6 | 2 | 1 |
| Chromosome 7 | 1 | 2 |
| Chromosome 8 | 2 | 1 |
| Chromosome 9 | 1 | 3 |
| Chromosome 10 | 2 | 2 |
| Chromosome 11 | 0 | 1 |
| Chromosome 12 | 3 | 1 |
| Chromosome 13 | 1 | 1 |
| Chromosome 14 | 1 | 0 |
| Chromosome 15 | 1 | 2 |
| Chromosome 16 | 2 | 1 |
| Chromosome 17 | 2 | 0 |
| Chromosome 18 | 0 | 1 |
| Chromosome 19 | 1 | 1 |
| Chromosome 20 | 1 | 0 |
| Chromosome 21 | 1 | 0 |
| Chromosome 22 | 0 | 0 |
| Total | 30 | 26 |
Granddaughter 2 has closer to the same number of maternal and paternal of crossovers, but about 8% more maternal.
Comparing Maternal and Paternal Crossover Rates
Given that males clearly have a much, much lower crossover rate, according to the Philip’s chart as well as the evidence in just these two individual cases, over time, we would expect to see the DNA segments significantly LESS broken up in male to male transmissions, especially an entire line of male to male transmissions, as compared to female to female linear transmissions. This means we can expect to see larger intact shared segments in a male to male transmission line as compared to a female to female transmission line.
| G1 Mat | G2 Mat | Mat Avg | G1 Pat | G2 Pat | Pat Avg | |
| Gen 1 | 56 | 30 | 41.71 | 23 | 26 | 26.65 |
| Gen 2 | 112 | 60 | 83.42 | 46 | 52 | 53.30 |
| Gen 3 | 168 | 90 | 125.13 | 69 | 78 | 79.95 |
| Gen 4 | 224 | 120 | 166.84 | 92 | 104 | 106.60 |
Using the Transmission rates for Granddaughter 1, Granddaughter 2, and the average calculated by Philip, it’s easy to see the cumulative expected average number of crossovers vary dramatically in every generation.
By the 4th generation, the maternal crossovers seen in someone entirely maternally descended at the rate of Grandchild 1 would equal 224 crossovers meaning that the descendant’s DNA would be divided that many times, while the same number of paternal linear divisions at 4 generations would only equal 92.
Yet today, we would never look at 2 people’s DNA, one with 224 crossovers compared to one with 92 crossovers and even consider the possibility that they are both only three generations descended from an ancestor, counting the parents as generation 1.
What Does This Mean?
The number of males and females in a specific line clearly has a direct influence on the number of crossovers experienced, and what we can expect to see as a result in terms of average segment size of inherited segments in a specific number of generations.
Using Granddaughter 1’s maternal crossover rate as an example, in 4 generations, chromosome 1 would have incurred a total of 24 crossovers, so the DNA would be divided into in 25 pieces. At the paternal rate, only 8 crossovers so the DNA would be in 9 pieces.
Chromosome 1 is a total of 267 centimorgans in length, so dividing 267 cM by 25 would mean the average segment would only be 10.68 cM for the maternal transmission, while the average segment divided by 9 would be 29.67 cM in length for the paternal transmission.
Given that the longest matching segment is a portion of the estimated relationship calculation, the difference between a 10.68 cM maternal linear segment match and a 29.67 paternal linear cM segment match is significant.
While I used the highest and lowest maternal and paternal rates of the granddaughters, the average would be 19 and 29, respectively – still a significant difference.
Maternal and Paternal Crossover Average Segment Size
Each person has an autosomal total of 3374 cM on chromosomes 1-22, excluding the X chromosome, that is being compared to other testers. Applying these calculations to all 22 autosomes using the maternal and paternal averages for 4 generations, dividing into the 3374 total we find the following average segment centiMorgan matches:
Keep in mind, of course, that the chart above represents 3 generations in a row of either maternal or paternal crossovers, but even one generation is significant.
The average size segment of a grandparent’s DNA that a child receives from their mother is 80.89 cM where the average segment of a grandparent’s DNA inherited from their father is 1.57 times larger at 126.6 cM.
Keep the maternal versus paternal inheritance path in mind as you evaluate matches to cousins with identified common ancestors, especially if the path is entirely or mostly maternal or paternal.
For unknown matches, just keep in mind that the average that vendors calculate and use to predict relationships, because they can’t and don’t have “inside knowledge” about the inheritance path, may or may not be either accurate or average. They do the best they can do with the information they have at hand.
Back to Philip again who provides us with additional information.
Maternal Versus Paternal Descent
Along a predominantly maternal path the DNA is likely to be inherited in more numerous smaller segments while along a predominantly paternal path it will likely be in fewer but larger segments. So matches who descend paternally from a common ancestor and carry the surname are not likely to carry more DNA from that common male ancestor than someone who descends from a mixed or directly maternal line. In fact, someone descending from an ancestor down an all-male path is more likely to inherit no DNA at all from that ancestor than someone descending down an all-female path. This is because the fewer segments there are the higher the risk is that a person won’t pass on any of them. Of course, there’s also a greater chance that all of the segments could be passed on. Fewer segments leads to more variation in the amount of DNA inherited but not a higher average amount of DNA inherited.
The chart above shows the spread in the amount of DNA inherited from a 3xgreat-grandparent, down all-maternal, all-paternal and down all possible paths. The average in each case is 3.125% i.e. 1 part in 32 but as expected the all-paternal path shows much more variation. Compared to the all-maternal path, on the all-paternal you are more likely to inherit either less than 2.0% or more than 5.0%. In 50,000 simulations there were 14 instances where a 3xgreat-grandchild did not inherit any DNA down the all-paternal path. There were no cases of zero DNA inherited down the all-maternal path.
One way to think about this is to consider a single chromosome. If at least one crossovers occurs in the meiosis some DNA from each grandparent will be passed down to the grandchild but when it is passed on without recombination, as occurs more frequently in paternal than maternal meiosis, all of the DNA from one grandparent is passed on but none at all from the other. When this happens, there is no bias toward either the grandfather’s or the grandmother’s chromosome being passed on. It’s just as likely that the segment coming down the all-paternal path will be lost entirely as it is that it will be passed on in full.
______________________________________________________________
Disclosure
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 Transfers
- FamilyTreeDNA
- MyHeritage DNA only
- MyHeritage DNA plus Health
- MyHeritage FREE DNA file upload
- AncestryDNA
- 23andMe Ancestry
- 23andMe Ancestry Plus Health
- LivingDNA
Genealogy Services
Genealogy Research
- Legacy Tree Genealogists for genealogy research
Thank you, Roberta 😚
To me, one consequence is that segments will tend to be smaller or more likely absent for the same number of generations back along the all-female line compared with, say, the all-male line. Including a sample form that line in the most distant generation becomes a little more important.
Is that right?
I believe that would be the outcome. Smaller for sure if there are more females in the line. Eventually, the segment would either be present or absent, not divided, so I think we would need simulations to prove this for sure.
Thank you for covering this in such depth Roberta. A lot for me to digest and consider!
Be careful with “all-female generations in a row”. The crossovers in one generation result in a mix of alternating DNA from male and female grandparents. When the next generation of female occurs, you are working on about half the segments from a male, which already have crossovers in them. Because each generation includes both male and female segments, I think as the number of generations increases the number and distribution of smaller and smaller segments will tend toward the average between the sexes.
I would like to see this at 10 generations for both all male and all female which represents both extremes.
Roberta, you can see the results for 3rd-cousins through to 6th-cousins in the “Crossover interference and sex-specific genetic maps shape identical by descent sharing in close relatives” paper from the team at Cornell University and University of Texas. Figure 3 shows that 6th-cousins related down all female lines of descent have a 13.1% chance of sharing DNA whereas 6th-cousins related down all-male lines of descent have only a 9.0% chance of sharing DNA. The more generations you go the wider the gap becomes.
https://www.biorxiv.org/content/10.1101/527655v2.full
Thank you. That’s exactly what I needed.
One of the most interesting observations from Figure 3 is the variation at 3rd-cousins in the number of shared segments. Around 60% of 3rd-cousins related along all-female lines will share more than 4 segments whereas only 30% of 3rd-cousins related along all-male lines will share more than 4 segments. But the average amount of shared DNA will be the same in both cases. It’s just a case of more shared segments but smaller in size down the all-female lines versus fewer shared segments but larger in size down the all-male lines. On average!
All transmissions of DNA from one generation to the next involve a male and a female but that’s not what is being discussed here. The key to understanding the DNA shared between cousins is that you only need to consider the path that each cousin has back to the shared ancestral couple. If that path is along an all-male line there will likely be fewer crossovers resulting in a comparatively smaller number of larger segments inherited than along an all-female line where the higher number of crossovers results in a greater number of smaller segments inherited. But the average amount of DNA inherited along any path is the same, regardless of the number of male or female transmissions in the path. More segments (all-female lines) provide a greater chance of cousins having some segments in common hence the higher match rates. A smaller number of larger segments (all-male lines) mean that there is less probability of cousins matching – but when they do they can share some comparatively larger segments. The more generations involved the larger the difference becomes between cousins related along all-female lines and cousins related on all-male lines.
All of my crossover points were fixed in my DNA 9 months before I was born. Same for each Match. A Match and I almost always get a different segment from our Common Ancestor. What we “see” in a shared segment is only the overlap – rarely the full size of the segment either of us got. From one shared segment we cannot determine either of the crossover points. The length of individual shared segments doesn’t tell us much – they may, or may not, identify one of my two crossover points (from my Ancestor). When we form Triangulated Groups, we tend to see both crossover points (if even a little fuzzy). With 98% of my DNA mapped with 371 TG segments, I have 181 paternal and 190 maternal TGs. However, they are not all from one generation – I just followed the natural break points.
Even if I knew the path to my Match’s Ancestor, the shared segment wouldn’t be the same as her real ancestral segment.
I think Roberta’s original post laid out a good case for the total segments we are likely to see with TGs or Clusters or DNA Painting; but with so much random variation, I wouldn’t try to pin it down in any specific subset.
I am so envious. 98%. I’m still stuck at 78 or 79. It’s the recent German and Dutch immigration.
It is mind-boggling to realize that there is such a big difference between male and female inheritance of autosomal DNA. I have not considered this at all – I was just starting to use cMs to get a ballpark estimate of genetic distance – now I have a whole different sense of it. Thanks for sharing these insights.
I had always thought there was a difference based on so many observations. Now we know there is and can quantify it, thanks to Philip.
For me, the two most important outcomes of this analysis are: 1) there are relatively few crossovers per generation, and a manageable number after say 8 or so generations (for me, each one is a Triangulated Group); and 2) the number of crossovers each generation is relatively stable; while the number of accumulated segments going back increases, which means more and more segments are NOT subdivided and are passed on intact. Dare I say it: a growing number of segments look like sticky segments – being passed along for multiple generations;>j
Yes, and it looks like we can expect more preserved segments, meaning more marches for a longer period of time, in direct male, meaning surname, lines. That’s very helpful.
Very interesting Roberta – thank you.
I think what is the best use of this difference in number of recombinations between the sexes is in Visual Phasing when you have no cousins to help. Andy Lee has made a great tool for that and explains how to use it here. https://www.youtube.com/watch?v=XNgqDPG3AYE
Excellent article, very thought provoking and very useful. I just had a quick look at my DNA painter chromosome map in light of what was revealed in this article and whilst I haven’t carried out a detailed analysis, it does indeed seem that I have some chromosomes that I have received exclusively from my paternal grandfather.
I have also made significantly more progress on finding DNA matches related to me through my paternal grandfather than any of my other grandparents. I had just thought this was coincidence but not I’m thinking there could be a reason for that as I may have inherited larger segments from that particular grandparent due to fewer crossovers. And if the segments are larger then there is a better chance of finding matches who share them. Is this a fair assumption based on what you and Philip have said?
Yes, I would think so. Also, if others also descend from males in that line.
Really interesting, Roberta.
I was wondering if the actual crossover points themselves are specific to maternal or paternal inheritance. On DNA Painter I’ve been trying to use the common crossover points (beginning and/or ending) to determine which unknown autosomal matches are coming from which parent. It’s unfortunate that my mother has Metis and Irish while my father is predominantly Scottish descent and I have very few matches closer than 3rd cousins. I have noticed that segments that I’ve initially called Maternal will occasionally match segments that should be Paternal. Am I following a triangulation strategy that is doomed to failure?
No, you just are aware of what you need to watch for. We all have foolers. I don’t think the location itself is relevant.
Roberta, this is FASCINATING and so many points to ponder, well worth many re-reads.
IMHO, I think Blaine Bettinger on his DNA Painter should post a link to this educational article on crossovers and re-combination so the public can be made aware of the potential variables in his Painter.
Blaine is certainly welcome to do that.
Thanks Roberta. This helps explain the sticky bits that persist for generations. I find it amazing how among 45+ ancestors that are 3C to 4C or even 4C2R I can see a shared centimorgan range between 120cM down to 7cM.
I don’t get to travel to search for documents as much as I’d like. I’m relying heavily on matching sticky blocks to identify which family my dna matches come from. Sonce ancestry doesn’t offer a chromosome browser which means I don’t always know how/where a paper trail confirmed dna match actually matches me. But sometimes I have someone that only matches certain people that only match a certain dna block and they form a dna cluster with other people. But that’s guess work and why it’s so important to get your dna matches and cousins to transfer to something with a chromosome browser.
But what’s cool is that I’ve isolated 4 distinct dna blocks that can only come from my 3G grandparents. One block has to be the grandfather. The other has to be the grandmother (and there are dna matches that supports this). My biggest sticky block on my dad’s side corresponds to my direct male line through him to my 3G grandfather . The block that corresponds to his wife and my 3G grandmother is smaller and less prevalent.
Your article was so timely. It’s the rational explanation of what I just discovered. Until last weekend I kept hoping for a dna match with a large sticky block that matches the dna block of my direct male line 3G grandfather. Someone who also traces to the surname and was born *before* him. This weekend I got that. I have a dna match with whom I’ve maintained a collaborative relationship with for two years who has a large, 33cM match on the sticky block that corresponds to my direct male line 3G grandfather. This weekend she contacted me and said “I think I know who we match through”. Someone with the same surname who lived 1750ish-1784.
She traces to someone with the same last name who lived 75 years earlier than my MDKA on my dad’s ydna line. Same state. Just a few counties south of where family legend says my dad’s ydna line first settled. And the mans wife? She was from the right county.
I just don’t know if I’m looking at a 5G grandfather or a 6G uncle. He had one son and I don’t know where he went afterwards. I hope he has direct male 5G or 6G grandsons for ydna testing .
Can there be another explanation for why someone matches the big sticky block that corresponds to my 3G grandfather? What are the odds that a 33cM block survived intact since a common ancestor sometime in the 1700s? We all want to do good (perfect) research. Anyone – Let me know if you think I’m going down the rabbit hole. Btw- the family I’m researching is from counties with really severe loss of courthouse artifacts like deeds, marriage books, order books, etc. The documentation just isn’t enough in those counties.
– John
If you have the right people, Y DNA would be helpful too. Form triangulation groups. That’s a huge segment yo be identical by chance. You’re doing the right things.
Roberta, and does this indicate, generally speaking, that we tend to pick up more matches from our paternal side because the rate of crossover/re-combo is lower, and the segments larger, this helping meet the threshold for matching and reporting by the companies?
Also, factoring in “matches from a surname paternally more likely to carry more dna………”
Because, I definitely have more matches from my father’s side.
I would think so, but that’s an intuitive answer, not a scientific one. No one has looked at that yet and I’m not sure how the comparison could be made “equal.”
In the range of 3rd to 6th cousins you will have more matches to cousins related along predominantly female lines of descent than to cousins related along predominantly male lines of descent. See Figure 3 in https://www.biorxiv.org/content/10.1101/527655v2.full. Note that the average amount of DNA inherited from ancestors of a particular generation is the same regardless of the number of male or female transitions along the path. The difference is that along all-male paths of descent you will inherit a smaller number of large segments compared to a larger number of small segments along all-female paths. More segments translates into a greater chance of matching.
But the interesting aspect is that you must reach a point where many of the smaller segments will be below the reporting threshold. From Figure 3, 6th-cousins related along all-female paths share DNA 13.1% of the time compared to only 9.0% along all-male paths. In both cases it’s nearly always just a single segment. What if the majority of those 13.1% of matches are small segments, below 7 cM? Matches along the all-male path are fewer in number but larger in size. I think that it’s likely that you will reach a situation after a number of generations where the number of “reportable” matches becomes higher along the all-male paths than the all-female paths. I wonder what generation that will be? I’ll explore this in some future simulations!
I am a novice at this and I am thoroughly confused?? It will take a lot of studying to understand. Are there clinics for beginners?
DNAadoption offers educational classes for everyone. I wrote this article: https://dna-explained.com/2019/08/06/first-steps-when-your-dna-results-are-ready-sticking-your-toe-in-the-genealogy-water/
Roberta,
Thanks for this link. I to still have issues trying to wrap my head on all of this DNA business and figuring out what is what…lol. Thanks for writing all your blog articles when it comes to DNA…I have learned alot and still learning. My head is swimming around this article on crossovers definitely…probably have to read the article again several times but definitely be worth it and learning more about my own ancestry especially on my Father’s biological Father side definitely…thanks again for all you write about on DNA.
Regards,
Cindy Carrasco
You’re most welcome.
This article might help: https://dna-explained.com/2019/07/10/dna-results-first-glances-at-ethnicity-and-matching/
For me this has to be the most useful information I have ever got so many times I have looked at those lines and thought how do I know how this really works and understand it and now I actually do thank you you are a dazzling jewel in the crown of DNA support.
Ahhh, thank you so much.
I think that it is worth pointing out that the figures of 26.6 crossovers in male transmissions and 41.7 crossovers in female transmissions are just averages. No individual is going to be exactly on the average. There is considerable variation in these numbers. The range in male meiosis events (with the top and bottom 1% excluded) is between 17 and 36 crossovers and the range in female meiosis events is between 30 and 54 crossovers. There is some overlap between these two ranges so there are people around who received a set of chromosomes from their father that had experienced more crossovers than the set of chromosomes received from their mother.
Oh my goodness, this is an entire course! Thank you for providing so much to think about and digest. As I try and get my head around this topic, one observation you referenced is the difference in crossover based on maternal age, at least I think that was what the Campbell article addressed. Since the essence of your post is that the rate of crossover affects predicted relationships, would that also imply that descendants of siblings separated by many years would be affected?
Bonnie, the Campbell et al. paper found that the average number of crossovers increases by 3 per year for mothers over the age of 39 years. The main noticeable effect would be that children of much older mothers, where there has been more crossovers, would likely inherit a more even split of DNA from the mother’s two parents. This would mean that siblings separated by many years, where the younger one was born to a mother considerably older than 39 years, would also be more likely to share closer to the average amount of DNA that siblings share.
Philip, thank you so much for that explanation. It’s the years between sibilings as well as the age of the mother that is significant. So siblings born 10 years apart when their mother was 18 and 28 would not have the same effect as siblings born 10 years apart when their mother was 31 and 41.
Indeed, I have quite a number of “large” matches on my paternal grandfather’s side, but still not a lot on my maternal grandmother’s side. Extremes are on the smaller chromosomes; I share on chrom 16 almost 60cM with someone; our shared ancestor is his 3x-gg-grandfather from his direct paternal line, but for me this person is a 7x-gg-grandfather (involving 6 males). Initially, I thought this was too far back for such a large match, but I know I got this chromosome intact from my grandfather, so effectively, I am comparing this match not with me, but with my grandfather (this “saves” 2 generations). Chromosome 22 is almost entirely from a 1/32 grandparent.
For my paternal side, I am lucky to have my grandmother tested, for my maternal side, I’d love to know what I got from each of my grandparents, unfortunately they are not alive anymore. I might have a shortcut, which is testing my sibling – who is still cautious about it – see here: http://geneticgenealogygirl.com/en/2017/06/visual-phasing-of-chromosome-1/
This is a great article. Thank you for writing about this – and I also appreciate Phillip’s work. I think you may be interested in work done by Dr. Amy Williams at Cornell University and others that I have found to be very impressive. In their work, they use sex specific recombination signatures to infer the gender of a match (See “Distinguishing pedigree relationships using multi-way identical by descent sharing and sex-specific genetic maps” : https://www.biorxiv.org/content/10.1101/753343v1). She has a number of interesting projects that build on an understanding how recombination differs in males and females. I think Phillip will also appreciate her work – he’s probably already familiar with ped-sim. You can find her publications at http://williamslab.bscb.cornell.edu/.
Alan
Hi Alan,
Thanks for your comment. Roberta first started posting results from my GAT-C simulation model in September 2017 (First Cousin Match Simulations) but I asked her to delay any further articles as I was planning to publish a paper in a scientific journal. In January of this year Roberta pointed out to me the preprint article “Surprising impacts of crossover interference and sex-specific genetic maps on identical by descent distributions” (https://www.biorxiv.org/content/10.1101/527655v1) of which Amy Williams was one of the authors. Upon reading this paper I discovered that the team from Cornell and University of Texas were using the exact same model (two-pathway interference-escape model described by Housworth and Stahl) and the exact same parameters (obtained from “Escape from crossover interference increases with maternal age” by Campbell et al 2015) that I had used in my GAT-C model. This would be no coincidence as the two-pathway model is the only one that closely resembles the observed distribution of crossovers in humans and the Campbell study is the only comprehensive source of crossover data that I have been able to find. Comparisons between my results and those published in the “Surprising impacts…” paper suggest that our simulation results are identical.
Clearly I had procrastinated for too long and there seemed little value in trying to publish in a journal what would essentially be the same as the “Surprising impacts…” paper. Instead, Roberta has kindly offered to collaborate and to release more of my simulation results on her DNAeXplained blog. We are aiming to concentrate on aspects that will be of particular interest to the genealogical community rather than to scientists. Hopefully this article on the difference between male and female crossover rates is just the first of many more to come. If there are any specific simulation results that you would like to see and think that they would be of general interest to genealogists please let Roberta or I know.
I haven’t looked at Ped-sim as I had already built my own GAT-C simulation model a few years prior to Ped-sim being referenced in publications. When I get around to tidying up my GAT-C model I might make it available to the public.
A revised version of the “Surprising impacts…” paper was posted in June under a new title “Crossover interference and sex-specific genetic maps shape identical by descent sharing in close relatives” (https://www.biorxiv.org/content/10.1101/527655v2). Disappointingly for genealogists some of the most useful information in the original (Table 1: IBD sharing fraction summary statistics for relatives) has been stripped from the revised version. The original version of the paper is still available however.
Philip
I am just now starting to figure all of this out. It would have been helpful to know if the grandparents and greats were maternal or paternal.
Or does it matter? Thank you
Yes, assuming you want to know if your matches are maternal or paternal. Or maybe I don’t understand the question.
I was referring to the beginning of your post that mentioned your granddaughter, her grandparents and great grandparents.
I was wondering if they were maternal or paternal.
It doesn’t matter if the grandparents are maternal or paternal, except for the X chromosome which has a different inheritance path. Men receive an X only from their mother. Woman receive one from both parents. In this case, the three grandparents consisted of 2 females and a male, and the great-grandparent is a female.
What I love about genetics is that ultimately we only have about 120 people’s genes in us due to crossover frequency and we pass on 60 to our children. There are probably only 2 people who have Charlemagne’s genes (1 in 4.3 billion or .5^32) because it becomes a flip of the coin after 10 generations and he is 42 generations back.