PDA

View Full Version : Matching on X but not Autosomal?



isgaldo
07-22-2017, 05:39 PM
On Gedmatch I am matching with a lady with a 43cm chunk on the X Chromosome but absolutely no Autosomal DNA match? Is that possible?
This is quite a large segment on the X, is this a legitimate match?

Thanks for reading! ;)

Pylsteen
07-22-2017, 05:54 PM
I think it is possible due to the different inheritance pattern of the X-chromosome; for example, fathers give their X-chromosome unchanged to their daughters. Of course it may or may not recombine when a mother gives hers to her child. There are cases in which someone just inherited an unrecombined X-chromosome from their mother. It is therefore likely that the X-chromosome retains older segments for a longer time.

geebee
07-23-2017, 01:33 AM
On Gedmatch I am matching with a lady with a 43cm chunk on the X Chromosome but absolutely no Autosomal DNA match? Is that possible?
This is quite a large segment on the X, is this a legitimate match?

Thanks for reading! ;)

A match of this size is almost always going to be a legitimate match. But as Pylsteen points out, segments on the X chromosome can be handed down by very different paths.

Here are the actual paths from two different 6th great grandparents to me. Either path could be one by which I received a segment on my X chromosome.

The first path is from my 6th great grandmother, Marie Anne Catherine Berda dit Picard. Marie Anne Catherine has the distinction of being the most distant known ancestor in my mtDNA line.


1. 6th great grandmother. Marie Anne Catherine Berda dit Picard.
2. 5th great grandmother. Angelique Ladner.
3. 4th great grandmother. Rosalie Fayard.
4. 3rd great grandmother. Marie-Josephe Ladner.
5. 2nd great grandmother. Marie Eulalie Ryan.
6. Great grandmother. Marie Josephine Cannette.
7. Grandmother. Madeline Mildred Pons.
8. Mother. Marilyn Pearl Weaver.

Because these are all female transmissions, they all involved at least the possibility of recombination of the X chromosomes.

This second path begins with someone whose name I don't yet know, though I known he must have existed. He was the father of my 5th great grandmother, Polly Green. The reason I begin with him is that this line involved the least number of generations in which recombination of the X chromosomes could occur.


1. 6th great grandfather. Father of Polly Green.
2. 5th great grandmother. Polly Green.
3. 4th great grandfather. John E. Bowling.
4. 3rd great grandmother. Agnes Bowling.
5. 2nd great grandfather. William Sherman Muncy, Jr.
6. Great grandmother. Hannah Louise Muncy.
7. Grandfather. Milford Pearl Weaver.
8. Mother. Marilyn Pearl Weaver.

Instead of there being eight generations in which recombination is possible, for this path there are only four. Obviously, that makes it much more likely that a large segment could remain intact from an ancestor at the same distance.

A cousin whose most recent common ancestor is at the beginning of the first path is likely to share neither a X chromosome match nor an autosomal match -- though either is possible. A cousin whose most recent common ancestor is at the beginning of the second path has a rather good chance of sharing a match on the X chromosome -- but no greater a chance of sharing a match on any of the autosomes.

geebee
07-24-2017, 04:01 PM
On Gedmatch I am matching with a lady with a 43cm chunk on the X Chromosome but absolutely no Autosomal DNA match? Is that possible?
This is quite a large segment on the X, is this a legitimate match?

Thanks for reading! ;)

I believe it's possible to match someone only a single segment this large on an autosome, so it would certainly be possible to match on a single segment this large on the X chromosome.

I can't confirm this from my own results, but I do match someone on a 35 cM segment which cannot be on the X chromosome. This match is at Ancestry, so I can't actually see where it is, but based on my shared matches it has to be on my maternal grandmother's side. I don't know whether Ancestry includes the X chromosome in determining matching, but even if it does I only match my grandmother on a total of 20 cM on the X chromosome.

Therefore, to match someone on a single segment of 35 cM on my maternal grandmother's side, the shared segment would have to be on an autosome.

C J Wyatt III
07-24-2017, 05:08 PM
I believe it's possible to match someone only a single segment this large on an autosome, so it would certainly be possible to match on a single segment this large on the X chromosome.



Distant big segment matching, whether it is via autosome or X chromosome, is something that the genetic genealogy mainstream has got completely wrong. Some have said that those matching segments are almost always IBD. If you really examine what is going on from a logic and probability viewpoint, the opposite is true. For a large and distant matching segment to be IBD is a near impossibility.

Genetic genealogy frequently commits what I call the "chain letter fallacy". Chain letters work if you have an endless supply of people and can find people at each round who have not played before (and won't balk). It is hard to find people who do not have lines back to the same ancestor within the genealogical timeframe. In other words, these big segments are a reconstructed piece of a common ancestor's genome coming from have lines through each parent which go to that ancestor.

Jack Wyatt

Saetro
07-24-2017, 07:58 PM
Distant big segment matching, whether it is via autosome or X chromosome, is something that the genetic genealogy mainstream has got completely wrong. Some have said that those matching segments are almost always IBD. If you really examine what is going on from a logic and probability viewpoint, the opposite is true. For a large and distant matching segment to be IBD is a near impossibility.

Genetic genealogy frequently commits what I call the "chain letter fallacy". Chain letters work if you have an endless supply of people and can find people at each round who have not played before (and won't balk). It is hard to find people who do not have lines back to the same ancestor within the genealogical timeframe. In other words, these big segments are a reconstructed piece of a common ancestor's genome coming from have lines through each parent which go to that ancestor.
Jack Wyatt

Autosomal DNA involves crossovers. These are random events. People are generally not good at handling random events. They like a pattern, and tend to assume that the average applies. That's not too bad for autosomal over 22 chromosomes, especially since the observed variance is not too bad. Even then, some people have noticed that the smaller chromosomes sometimes have no crossovers in one generation.
The pattern of X inheritance is great for working out who you may and did not inherit that piece of X from.
Unless your relationship is close, the weird pattern of inheritance with X may cause great frustration.

Reading people's comments on their experience with the X chromosome suggests that this all or none happens more frequently with the X.
And then there is the anomalous way it is passed from generation to generation. A woman gets EXACTLY the same X chromosome that her father had.
A man receives his X chromosome usually as a result of one or more crossover from the two X chromosomes in his mother.
And then there is an expectation in people that somehow inheritance of the X chromosome is linked to inheritance of autosomal: that if you receive one significant segment on the other 22 and one significant piece on the X that they have been handed down by the same pathway.
The other thing is a sort of reverse-exceptionalism. We assume that whatever happens to us is the general case and vice versa.
That really does not work with random events. Sure, if you get enough of them, then the pattern is close to the average; but individual cases may be average or they may be extreme. And anyone who says that everyone they have ever talked to has the average may just never have met the extreme - but it may be the next case they encounter.

So weird stuff is going on with the X.
Some of it is already happening with other autosomal, but we often don't notice.
For both, we need to eventually go from observations rather than models. Models have to make assumptions. They are often based on observations from only limited data.
If Blaine Bettinger (or someone else) gets around to an X study, we may discover what is really going on.
We can only find out the average and the variance if we pool real data.

Or maybe I am misunderstanding your post and there is heaps of data backing your statements.

C J Wyatt III
07-25-2017, 02:00 AM
Or maybe I am misunderstanding your post and there is heaps of data backing your statements.

OK, let's start with my assertion: "For a large and distant matching segment to be IBD is a near impossibility."

The length of chromosomes in centiMorgans vary, but only five are greater than 200 cM's, so 200 cM will be a good number to use. If you go back six generations. the average decay rate would give you an expected 3.125 cM remaining from the original 200 cM. Of course you cannot expect that will be one segment, but why worry about that detail? if your match is on the same level as you, the person also ends up with an average of 3.125 cM remaining. The catch is, to be a matching segment, it has to line up perfectly with yours. Like I said, a big distant IBD matching segment is near impossible. Do these big segments we see come about from the parents sharing a common ancestor? What other explanations are there?

Jack

Pylsteen
07-25-2017, 04:16 PM
Yes, it is important to keep in mind that different matching segments with the same person, even if both legitimate, don't have to go back to the same ancestor, certainly not if you share ancestry in the same geographical region, in which case you would share multiple ancestors.

geebee
07-26-2017, 06:35 AM
OK, let's start with my assertion: "For a large and distant matching segment to be IBD is a near impossibility."

Sure, we can start with that. To begin with, you seem to have a fundamental misunderstanding of the word average.

If I were to tell you that the average height of adult residents of a certain country is 171 centimeters, does that mean I can assume that the next adult I meet from that country will be 171 centimeters? No, of course not. It's possible, but the person could be much taller or much shorter.

Not only that, but there's an important factor in height that means you aren't typically going to see average height reported this way. What you'll see is not the average height of an adult resident, but of an adult male resident or an adult female resident.

As it turns out, in the country I was referring to, the average height for a man is given as 179 centimeters, and the average height for a woman is 163 centimeters. There's a 16 centimeter difference in these two averages -- or about 6 inches. So if I guess that the next person I see will be 171 centimeters and the person is female, I'm likely to have guessed too high -- although since it's an average, that particular woman might well be much taller. Or if the person is a man, the guess of 171 centimeters will likely be a bit low. But again, because it's an average, that particular man might be even shorter; or he could be much, much taller.

The same thing applies to recombination. In fact, here the gender difference is even more significant than in height. The average crossover rate for men is much lower than it is for women. What does this mean? Well, for one thing, it means that when we're told the length of chromosomes in centimorgans, that number represents an average -- but it's an average that doesn't exist in the real world. To obtain it, you have to average the crossover rate for men and women together. But no one is the offspring of two generic parents. Everyone is the offspring of a male and a female, and these two will probably have very different crossover rates. (One study found an average of 42 crossovers per recombination event in women, versus 28 in men.)

If you have 42 crossovers spread over 23 chromosomes, it would be possible for a crossover to occur on every chromosome, and for a second crossover to occur on almost every chromosome. But, 28 crossovers only allows for a second crossover on a few chromosomes.

In reality, of course, in both cases a few chromosomes may escape recombination altogether (that is, have no crossovers). Other chromosomes may have more than two crossovers.

This isn't just hypothetical, it's what people have actually been able to observe. For example, I can tell by comparing my father's genome to my daughter's that I passed on to her:

8 chromosomes with no crossovers (7, 9, 15, 17, 18, 21, 22, X)
10 chromosomes with 1 crossover each (1, 4, 8, 10, 11, 12, 13, 16, 19, 20)
3 chromosomes with 2 crossovers each (5, 6, 14)
1 chromosome with 3 crossovers (2)
1 chromosome with 4 crossovers (3)


This is a total of 25 crossovers -- so not far from the average of 28.

But the important thing to realize about the gender difference in crossover rates is that it doesn't simply have implications for a single recombination event. It also affects entire series of recombination events.

Consider your example of an ancestor 6 generations ago. If that ancestor and all five of the descendants who are also your ancestors happen to all be male, then that makes for many fewer total crossovers in the six generations. The number in this case -- assuming each recombination event involves the average number of crossovers -- is 168. That actually yields an average segment size not of 3.125 cM, but a little over 21 cM.

At the other extreme, with the line of all females you get 6*47. That yields a total of 282 crossovers, or something more like 12.5 cM. (Still quite a bit more than 3.125 cM.)

But once again, these are averages. Why do you believe that in every transmission, the crossovers will manage to arrange themselves to get averaged-sized segments? There will be an average segment size only because no matter what size each segment is, the total length of all the segments can be summed and averaged. That doesn't mean that any, or even most, segments will actually be this average.

Here's another list for you. It shows what percentage of each chromosome my daughter inherited from my father.

chromosome 1: 82%
chromosome 2: 29%
chromosome 3: 39%
chromosome 4: 55%
chromosome 5: 33%
chromosome 6: 21%
chromosome 7: 0%
chromosome 8: 16%
chromosome 9: 100%
chromosome 10: 96%
chromosome 11: 78%
chromosome 12: 35%
chromosome 13: 87%
chromosome 14: 53%
chromosome 15: 0%
chromosome 16: 71%
chromosome 17: 0%
chromosome 18: 0%
chromosome 19: 98%
chromosome 20: 79%
chromosome 21: 100%
chromosome 22: 0%
X chromosome: 0%

The point of the above list is mainly to show how rarely crossovers resulted in an even division of segments. Actually, only twice -- on chromosomes 4 and 14 -- did my daughter inherited about half of the DNA on the chromosome from her grandfather. In all other cases, the division was more lopsided.

Now this is just one example, and I have no way of knowing how typical it is. But it should serve to show that the mechanism of recombination can result in both very large segments and very small segments. Because crossovers occur more-or-less randomly*, they can't be relied on to come up with "average" sized segments. And while over a series of recombination events in a given line this mechanism may tend to even things out, it definitely does not have to do so.

To give one last example of just how large a segment can persist for a surprisingly long time, a 2nd cousin of mine recently uploaded his DNA file to GEDmatch. This has allowed me to compare not only the segments I share with my cousin, but the segments my siblings share. One of my brothers shares an 80.5 cM segment on chromosome 20 with this cousin. This represents 70% of the entire chromosome, when the closest common ancestors are great grandparents.

And, no, it doesn't reflect the relatedness of either my maternal grandparents or my parents. My grandfather's ancestry was primarily from Virginia and Kentucky; my grandmother had three immigrant grandparents, and her non-immigrant grandparent was from the Gulf Coast of Mississippi. My father's ancestry is all from Pennsylvania since colonial times. It simply shows that very large segments can be passed on over several generations.

And if in three generations (great grandparents, grandparents, parents) a segment that began as 108 cM-long chromosome only "decayed" to 80.5 cM, it doesn't seem likely that survival of a "big" distant segment in just a few more generations is "near impossible".

Saetro
07-26-2017, 08:53 PM
OK, let's start with my assertion: "For a large and distant matching segment to be IBD is a near impossibility."
The length of chromosomes in centiMorgans vary, but only five are greater than 200 cM's, so 200 cM will be a good number to use. If you go back six generations. the average decay rate would give you an expected 3.125 cM remaining from the original 200 cM. Of course you cannot expect that will be one segment, but why worry about that detail? if your match is on the same level as you, the person also ends up with an average of 3.125 cM remaining. The catch is, to be a matching segment, it has to line up perfectly with yours. Like I said, a big distant IBD matching segment is near impossible. Do these big segments we see come about from the parents sharing a common ancestor? What other explanations are there?
Jack

geebee in Post #9 above has demonstrated that averages DO NOT APPLY, necessarily, to single transfer events like the transfer from one person to another of a single chromosome. Even over 2 generations from his father to his daughter there are 4 chromosomes that have ZERO transmitted.
And if the average applied, then we would not have ancestors dropping out from our matches after a few generations. The matches would just get smaller.
So, if we get zero from some transfers, then we don't just have no chromosome. We get MORE from someone else, often as a longer segment.
Crossovers are random events. Sure the average describes what happens over large populations, but it is the spread of results that matter.
Blaine Bettinger's tables of accumulated actual events shows how the total DNA inherited from one individual can be zero with a relationship of 2C1R.
http://thegeneticgenealogist.com/wp-content/uploads/2016/06/Shared-cM-Project-Version-2-UPDATED-1.pdf

So if some segments have zero inheritance, others are bigger.
Yes, they will usually become trimmed or lost by later crossovers, but just occasionally they survive for several generations longer than expected.
A very few will come from what we would normally consider a long way back.
It's rare, but it happens.
It's not the average, alone, that describes behavior of segments, but the average plus variance - the spread.

geebee
07-27-2017, 04:36 PM
Well, consider this:

My niece Stephanie's results have just come back, and as expected she matches her father perfectly on her paternal X chromosome. Her father is the one who inherited our mother's paternal X chromosome without recombination.

What this means is that the chromosome has passed from great grandfather to great granddaughter with no changes -- no segments have "decayed" from three generations previous. But even in the three generations before that, a great deal depends on which sexes were involved.

Since my grandfather received his X chromosome from his mother, it could have been a recombination of the X chromosomes from both of his maternal grandparents. But it could as easily have included DNA from just one of them. In that case, it makes a difference whether it was from his grandmother or from his grandfather. If from his grandmother, then it could contain DNA from either or both of her parents; if from her father, then it could contain DNA only from her father's mother.

In that last scenario, the person we're considering is my grandfather's maternal grandfather's mother -- or, my brother's and my 3rd great grandmother, and my niece's 4th great grandmother. This is someone who would have been six generations before my niece -- exactly the distance from which someone said it was "almost impossible" for a large segment to remain intact. To the contrary, it's quite possible that my niece could have inherited 100% of her paternal X chromosome from that person.

Whether this is true for my niece or not, it is possible. And not just barely possible; it's actually likely that many people have a nearly intact X chromosome from that generational distance. Notice, I am not saying that a large percentage of people do, only that the number of people in this situation is probably large. (23andMe does, after all, have over a million customers.)

EDIT:

I'm still doing more conparisons involving my niece and other close relatives, but one thing I've noticed is that she does not match my father on any part of chromosome 17, but she matches both me and my daughter Kathryn across the full length of the chromosome.

This tells me that my brother and I appear to have inherited identical copies of our maternal chromosome 17, which may mean that we both not only received unrecombined copies from our mother, but passed those copies on without recombination.

Now if you're wondering how I figure the copies my brother and I received from our mother likely were not recombined, but came from just one grandparent; it's if we inherited recombined copies, they'd have to have exactly the same crossovers. It's more likely that they had no crossovers (or effectively, no crossovers).

dp
07-27-2017, 09:05 PM
OK, let's start with my assertion: "For a large and distant matching segment to be IBD is a near impossibility."

The length of chromosomes in centiMorgans vary, but only five are greater than 200 cM's, so 200 cM will be a good number to use. If you go back six generations. the average decay rate would give you an expected 3.125 cM remaining from the original 200 cM. Of course you cannot expect that will be one segment, but why worry about that detail? if your match is on the same level as you, the person also ends up with an average of 3.125 cM remaining. The catch is, to be a matching segment, it has to line up perfectly with yours. Like I said, a big distant IBD matching segment is near impossible. Do these big segments we see come about from the parents sharing a common ancestor? What other explanations are there?

Jack
males can pass more autosomal dna from one of their grandparents than one on the other side of their tree. Women tend to pass it as 50% from one of their maternal grandparents & 50% from one of their paternal grandparents.
this would have an effect on auDNA transmission.

I don't pay much attention to X-matches, unless it is a male that is being matched which would be on his mother's side.
BTW. did any one remember https://www.daysoftheyear.com/days/cousins-day/

C J Wyatt III
07-28-2017, 07:11 AM
Saetro, geebee and anyone interested:

If my hypothesis is correct, that big, distant matching segments are not IBD, but reconstructed from multiple paths to a common ancestor, then the arguments you used against me are flawed. You cannot disprove my hypothesis with an argument that is disproved by the hypothesis. We are at an impasse.

One way to get beyond this impasse is if I can show that most of us share a common ancestor, not in ancient times but within recent times. This can easily be done by triangulating X-matches. Many females have around 10,000 X-matches on the GEDmatch 'X one-to-many' using default parameters of 500 SNP's and 7.0 cM minimum segment. I have seen some females with around 25,000 matches and even one with over 50,000. I did an analysis on the 10,000 example and found that on average, any place on the X-chromosome (only one 'X' assuming male common ancestor) has around 400 or more triangulations with the matches from the 'X one-to-many' list. Intuitively I think the overlapping would mean that everything points to just one common ancestor, but I recently undertook another analysis that deals with that overlap issue. I used a female with about 10,000 matches, and skipped over the few close and known relationships at the top. Each additional match I would compare with the segments derived from all of the previous ones to add segments in untouched areas and extend the previous segments which overlapped with the ones from the new match. If the overlap between two segments is more than a handful of SNP's, the common ancestor from both segments is almost certainly the same. After less than fifty tries, I had the whole 196 cM X-chromsome divided into four segments with just three small gaps. I was able to very quickly find the matches that closed the gaps and show that anywhere on the whole X-chromosome points to one common ancestor, with about 9900 matches to spare. I believe almost anyone who has enough matches would get similar results.

At least one or two genetic genealogists will admit that there is a large source of common ancestry in the 18th Century that they have not been able to pinpoint. These X-DNA triangulations add tremendous support to that idea. The big problem is that the existence of this large amount of common ancestry undermines much of mainstream genetic genealogy thought, at least for distant matching.

Jack Wyatt

Pylsteen
07-28-2017, 09:03 AM
Very interesting; the X-chromosome seems to harbour more ancient segments; this thread increased my knowledge about crossing-overs in general; I am lucky to know all (paper) ancestors 9 generations back who could have contributed to my X-chomosome, being Dutch, Belgian and German people. Reading the above, that would generally not even be enough to establish a match, since it could be from medieval times.

geebee
07-28-2017, 02:47 PM
Saetro, geebee and anyone interested:

If my hypothesis is correct, that big, distant matching segments are not IBD, but reconstructed from multiple paths to a common ancestor, then the arguments you used against me are flawed. You cannot disprove my hypothesis with an argument that is disproved by the hypothesis. We are at an impasse.

One way to get beyond this impasse is if I can show that most of us share a common ancestor, not in ancient times but within recent times. This can easily be done by triangulating X-matches. Many females have around 10,000 X-matches on the GEDmatch 'X one-to-many' using default parameters of 500 SNP's and 7.0 cM minimum segment. I have seen some females with around 25,000 matches and even one with over 50,000. I did an analysis on the 10,000 example and found that on average, any place on the X-chromosome (only one 'X' assuming male common ancestor) has around 400 or more triangulations with the matches from the 'X one-to-many' list. Intuitively I think the overlapping would mean that everything points to just one common ancestor, but I recently undertook another analysis that deals with that overlap issue. I used a female with about 10,000 matches, and skipped over the few close and known relationships at the top. Each additional match I would compare with the segments derived from all of the previous ones to add segments in untouched areas and extend the previous segments which overlapped with the ones from the new match. If the overlap between two segments is more than a handful of SNP's, the common ancestor from both segments is almost certainly the same. After less than fifty tries, I had the whole 196 cM X-chromsome divided into four segments with just three small gaps. I was able to very quickly find the matches that closed the gaps and show that anywhere on the whole X-chromosome points to one common ancestor, with about 9900 matches to spare. I believe almost anyone who has enough matches would get similar results.

At least one or two genetic genealogists will admit that there is a large source of common ancestry in the 18th Century that they have not been able to pinpoint. These X-DNA triangulations add tremendous support to that idea. The big problem is that the existence of this large amount of common ancestry undermines much of mainstream genetic genealogy thought, at least for distant matching.

Jack Wyatt

Jack,

Did you not read the part when I showed that one of my brothers and a 2nd cousin share an 80 cM segment? Another sibling and this 2nd cousin share 57 cM on the same segment, and yet another sibling shares 67 cM on an entirely different chromosome. These siblings and our 2nd cousin do not have any relatives more recent than two 2nd great grandparents.

Why, if a segment of this size can survive for three successive transmissions on each side is it so hard for you to believe that large segments could survive for three more transmissions, or longer? The number of pairs of is vast, so that there are many, many opportunities for wide variances from the "average". The average is merely the result of summing all individual segments and dividing by the number of individuals.

In virtually every recombination event, multiple chromosomes are transmitted which have entirely escaped recombination. Most of the time, you might not expect the same chromosome to do so twice in a row, yet whether it does or not is believed to be essentially random.

With enough transmissions, it is inevitable; with even more transmissions, there is no reason it can't happen a third time, or a fourth. This is very similar to tossing a coin and getting ten "heads" in a row. It does happen. In fact, once you have managed to toss "heads" ten times in a row, the odds of doing so an eleventh time are undiminished. It's still 50% for the next toss.

C J Wyatt III
07-28-2017, 04:02 PM
Jack,

Did you not read the part when I showed that one of my brothers and a 2nd cousin share an 80 cM segment? Another sibling and this 2nd cousin share 57 cM on the same segment, and yet another sibling shares 67 cM on an entirely different chromosome. These siblings and our 2nd cousin do not have any relatives more recent than two 2nd great grandparents.

Why, if a segment of this size can survive for three successive transmissions on each side is it so hard for you to believe that large segments could survive for three more transmissions, or longer? The number of pairs of is vast, so that there are many, many opportunities for wide variances from the "average". The average is merely the result of summing all individual segments and dividing by the number of individuals.

In virtually every recombination event, multiple chromosomes are transmitted which have entirely escaped recombination. Most of the time, you might not expect the same chromosome to do so twice in a row, yet whether it does or not is believed to be essentially random.

With enough transmissions, it is inevitable; with even more transmissions, there is no reason it can't happen a third time, or a fourth. This is very similar to tossing a coin and getting ten "heads" in a row. It does happen. In fact, once you have managed to toss "heads" ten times in a row, the odds of doing so an eleventh time are undiminished. It's still 50% for the next toss.

and you completely brush off the X-chromosome analysis?

I am not questioning your big segments. I am questioning whether they are IBD or reconstructed by multiple lines back to the common ancestor. Without being able to examine your data with the high resolution chromosme browser, I cannot make that determination. However I have seen plenty of big segments pop up that are not IBD.

I thought second cousins share great grandparents.

Jack

geebee
07-28-2017, 04:24 PM
I should mention that my brother Curt -- the one who shares an 80 cM segment on chromosome 20 with our 2nd cousin -- shares a total of 238.5 cM on the autosomes and 38.5 cM on the X chromosome. This is a pretty typical matching percentage -- more or less "average", if you will -- with a 2nd cousin.

The 80 cM, however, is hardly an averaged-sized segment for 2nd cousins. But that's because shared segments are not uniform. There can be immensely large segments -- potentially, as large as the entire chromosome -- or immensely small segments, perhaps of a smaller size than is detectable. Together, these may well "average" out, without any particularly segment having to be at the average.

As I've mentioned in previous posts, I have five full siblings with results at 23andMe. My father has also tested there, as has my daughter; and now one my nieces -- my eldest brother's daughter -- also has results.

Among six siblings, there is very a large number of pairs: 5+4+3+2+1, as a matter of fact, or 15. There are multiple instances in which pairs of us have not only inherited "unrecombined" autosomes, but even the same unrecombined autosome.

For example, my eldest brother and I fully match on both copies of chromosome 16. We also match on one copy of chromosome 17, and nearly match on all of the second copy. Our daughters only match on 94.27 cM of chromosome 16, although this is still a great deal more than half of the chromosome.

On chromosome 17, those their matching is essentially across the entire chromosome. Further, neither one of the girls matches her paternal grandfather at all on chromosome 17, which means this must be my brother's and my maternal copy of the chromosome.

So my brother and I each inherited an identical maternal copy of chromosome 17, meaning that it either has identically-placed crossovers for both of us, or is a chromosome which (essentially) avoided recombination altogether. The latter is actually the more likely of the two possibilities.

Now, if both girls received a copy of chromosome 17 that came from the same great grandparent -- one of my my brother's and my grandparents; I haven't yet tracked down which one, but I may be able to -- what are the odds of the same chromosome being passed on in full to a child in the next generation?

Well, for each daughter the odds aren't any different than they were in any of the previous generations. This is how recombination works. It is reasonably likely that the chromosome will have one or more crossovers in the next generation, but is far from impossible that it won't. My daughter has no children, so obviously she hasn't passed on any chromosomes yet. However, my niece has five children, so she's had five different opportunities to pass on part or all of this chromosome.

The point, though, is to make it very clear just how far an entire chromosome can be passed on intact. And if the chromosome has been passed on intact, then every segment on the chromosome has also been passed on without "decay".

So, five times my niece passed a set of 23 chromosomes on to a child. This set obviously included chromosome 17. What are the odds that she failed to pass on any portion of her paternal copy? What are the odds that in at least one transmission, she didn't pass on a fairly substantial portion of that copy? If she did, it means that the child who received it would share a rather large chunk of DNA -- much larger than average -- with a 2nd great grandparent.

Now, my siblings and I collectively only have ten children, and a slightly larger number of grandchildren. But if we'd all been as prolific as my parents -- who had six children by themselves -- or my father's parents, who had twelve; there would be very many cousins by now.

My father's parents had over 30 grandchildren. But I have a 4th great grandfather who had 20 children himself -- although it was by four different mothers. The point, though, is potentially have many 5th cousins, and 4th, and 3rd. I don't know precisely how many 2nd cousins I have, but my maternal grandfather's parents had at least seven children in addition to him, and most of these had children of their own.

Now, CJ Wyatt III proposes that very large segments cannot be from distant relatives, even though it's clear that even whole chromosomes are sometimes passed through multiple generations -- which means the "average decay rate" idea is misleading when applied to individual situations. Instead, he claims that what we're seeing is just the "piling up" of smaller segments from multiple shared ancestors.

This would require that both cousins inherited essentially the same small segments from these more distant ancestors, and that those segments would happen to line up in such a way as to give the appearance of a larger match. This might be true for someone with very recent membership in a very endogamous population, but most of us aren't -- or at least, not recently.

Like most people (all, eventually) my tree has experienced a certain amount of pedigree collapse. However, the closest ancestors I have who were at all related were my paternal grandmother's parents. They were 2nd cousins twice removed, and I have a pair of ancestors who are both 4th and 6th grandparents to me. Neither of these is a particularly close relationship, and while I have some cousins who appear to be related to me through these ancestors, none of them shares any especially large segments with me.

At any rate, ancestry on my father's side has no bearing on my mother's side, since she is unlikely to be related to my father within a genealogical time frame. They are from Pennsylvania and Mississippi, respectively, and met when my father was in basic training at Keesler Air Force Base. They certainly do not share any DNA.

My maternal grandparents aren't related, either. My grandmother's ancestry consists of three immigrants -- two from Minorca and one from Alsace-Lorraine -- and one grandparent whose ancestry was from southern Mississippi for many generations. Now, that grandparent had a fairly large number of ancestors who were related to each other; but the only relevance of this would be to cousins in this part of my family. And, yes, here I have cousins who are related to me by multiple paths. But guess what? None of the more distant cousins on this side shares any unexpectedly large segments with me. A few may share more than one smaller segment, but they don't seem to have all managed to "pile up" anywhere. Why not? Probably because to have large pile ups -- especially on both sides -- you still need recent multiple paths, not distant multiple paths.

In my previous posts, I've made reference to a 2nd cousin. Here we are talking about recent common ancestor -- our great grandparents. But we don't have any more recent connection. Nor do our great parents themselves seem to be related, so whatever pedigree collapse might exist between the two of us, it's beyond that. Eventually, there is some. Many of my Sizemore, Asher, and Bowling cousins are related by several paths.

But it's still irrelevant for me, because I don't have all those multiple paths in my recent ancestry. Having multiple distant paths can't result in large segments today, because they still need some mechanism for "lining up". The only thing multiple distant paths gets you is more chances for that ancestor to pass on any DNA at all -- but for distant ancestors, the odds of any DNA at all being shared with you is small to begin with. So they end up with several small chances.

Jack is wanting to find an alternate mechanism for explaining large distant matches, because he's already concluded that it's virtually impossible for a large segment to remain intact for so long. But I've already showed how even a whole chromosome can be passed on intact over multiple generations. An alternate mechanism is not needed. The most likely mechanism is obvious, and it isn't what Jack suggests. It's simply the random working of recombination -- operating much like the coin tosses I mentioned.

On average, the coin will be heads half the time and tails half the time, but this doesn't mean that in any given set of 100 tosses, it will always be 50 of each. It can be anywhere from 0 heads and 100 tails, to 100 heads and 0 tails. Given enough coin tosses, you will see all possible combinations eventually.

geebee
09-07-2017, 04:34 AM
and you completely brush off the X-chromosome analysis?

I am not questioning your big segments. I am questioning whether they are IBD or reconstructed by multiple lines back to the common ancestor. Without being able to examine your data with the high resolution chromosme browser, I cannot make that determination. However I have seen plenty of big segments pop up that are not IBD.

I thought second cousins share great grandparents.

Jack

Yeah, you're right about 2nd cousins and great grandparents. I hate it when my fingers type without any input from my brain.

There's really no opportunity for the kind of "reconstruction" you're talking about. Most people -- including me -- do not have pedigree collapse sooner than great grandparents. For that you'd have to have either parents or grandparents marrying 1st cousins.

This actually is the case for some of my relatives, but smaller segments that somehow manage to "bunch up" on just one side aren't going to have a corresponding match unless the other person actually managed to inherit a very large segment from the same source.

Take a look at this chart from "The Shared cM Project" managed by by Blaine Bettinger. It certainly doesn't disprove your hypothesis, but it shows how variable the amount of DNA passed on by ancestors is. This variability is not just the product of receiving more or fewer segments, but also receiving longer or shorter segments.

Recombination is essentially random, and because of that you don't get nice and neat segment sizes. You talked about an "average decay rate" at some point, but averages aren't prescriptive. You can speak of average family size, and some families will indeed be average-sized. But others will be very small -- just one person, really -- while still others will be huge (into double digits).

In genealogy, as you know, it's common to treat the average length of a generation as 30 years. But of course there will be times when someone is descending from a series of very young parents, or a series of very old parents. The same thing is true of recombination. You can have very large segments which stay very large over multiple generations. In an essentially random process, it is inevitable that this will happen sometimes -- just as it's inevitable that when you do lots of coin tosses, you'll sometimes get very long strings of all heads or all tails.

The average may 50-50, but that has absolutely no effect on whatever happens on the next toss.

EDIT:

Here's a comparison between the genomes of my father Bern and two of his granddaughters. They are my daughter Kathryn, shown in purple; and my brother Bernie's daughter Stephanie, shown in orange.

18590

The main thing I want to call attention to is just how rarely recombination has resulted in a chromosome being evenly split between the two grandparents. Some chromosomes are passed on with virtually no recombination, while others have just small amounts of DNA from one grandparent and large amounts from the other.

The overall result may be that children show close to 25% of their DNA with each grandparent -- though that can vary pretty widely, too -- but they hardly do this through sharing half of each chromosome.

And, just as with coin tosses, while it may be that most chromosomes will see an "evening out" over a long period of time, there are going to be instances in which the same chromosome ends up being passed down over several generations without having all that many crossovers, or crossovers which are simply near one end or the other.

The net result is that any segments on such a chromosome will not experience any sort of "average decay rate", at least over a sufficient number of generations that some cousins will share large segments with each other.

So the idea of large segments always being formed by some sort of "piling up" process is completely unnecessary to explain why people sometimes share larger-than-expected segments. It is possible that it could happen. I do not, however, believe it is the most parsimonious explanation.

kikkk
11-14-2017, 12:03 PM
Hi!
I have 4 X matches with American persons all of them women, does that mean we have a common female ancestor?
Knowing that I got my X chromosome from my mother, does that mean that my maternal grandmother is related to those 4 Americans?!
I am really confused, any help is welcomed, thanks!

Angoliga
11-14-2017, 01:21 PM
Hi!
I have 4 X matches with American persons all of them women, does that mean we have a common female ancestor?
Knowing that I got my X chromosome from my mother, does that mean that my maternal grandmother is related to those 4 Americans?!
I am really confused, any help is welcomed, thanks!

What are the sizes of the segments? I also have American/Latin American X matches on Gedmatch but they're all <10cM i.e. most likely IBS instead of really IBD.

kikkk
11-14-2017, 02:02 PM
Thanks!
Yes they seem to be IBS, I have just looked in depth and the segments are smaller than 5cM.

geebee
11-15-2017, 10:01 PM
Hi!
I have 4 X matches with American persons all of them women, does that mean we have a common female ancestor?

Not necessarily. When it comes to determining whether two people have a common ancestor, size definitely matters. On any chromosome -- not just the X -- if a match is small, there's a greater chance that it might be identical more by coincidence than by common descent.

On the X chromosome, the only real difference is that when you're comparing two men, you have a greater likelihood that the match is "real" (which means, identical by descent and not just by state). But all four of your matches are female, so you can't tell which of their two copies is involved in the match -- of whether even bits of both.


Knowing that I got my X chromosome from my mother, does that mean that my maternal grandmother is related to those 4 Americans?!

Since you're male, you do know that your X chromosome came from your mother. But don't automatically presume that it therefore came from her mother. Remember, a woman inherits an X chromosome from both parents. Therefore, an X chromosome she passes on to a child can consist of: a recombination of both of her X chromosomes; an exact copy of her paternal X chromosome; or an exact copy of her maternal chromosome.

I've seen this in my own family. Out of my five siblings, all three of my sisters inherited an identical X chromosome from our father. However, two of my siblings -- one of my sisters plus one of my two brothers -- also match each other over the entire span of the copy each received from our mother.

I match both of these siblings on all but 20 cM of my single X chromosome. Based on matches we share with more distant relatives (including a 2nd cousin, and multiple 3rd cousins and half 2nd cousins), it's clear that these two siblings inherited an unrecombined X chromosome traceable to our maternal grandfather.

So, don't assume that the most recent common ancestor for a match on the X chromosome has to be female. (Even though, yes, if the MRCA is male, you can go one more generation back to his mother.)