How deep could this X ancestry be?

[irenych]

I'm wondering if there is any way of calculating the age of a particularly sticky sretch around the centromere of my mother in law, my wife, and my stepdaughter. It is the same size in all three cases. It is securely "South Asian" as per Dr. McDonald's very cautious analysis. None of them have otherwise any significant autosomal levels of South Asian (all autosomal percentages being well in the "noise" range). They all have mtDNA HG U4. According to Gedmatch the stretch indicates at the very least 6 generations. But since it is the same in all it already seems to be 8 for my mother in law... Any ideas?

[irenych]

I'm wondering if there is any way of calculating the age of a particularly sticky sretch around the centromere of my mother in law, my wife, and my stepdaughter. It is the same size in all three cases. It is securely "South Asian" as per Dr. McDonald's very cautious analysis. None of them have otherwise any significant autosomal levels of South Asian (all autosomal percentages being well in the "noise" range). They all have mtDNA HG U4. According to Gedmatch the stretch indicates at the very least 6 generations. But since it is the same in all it already seems to be 8 for my stepdaughter... Any ideas?

I meant stepdaughter not mother in law. Corrected this. Sorry for the slip.

[thetick]

Cousin prediction beyond 3rd cousin is just to random and small to be reliable as far as which cousin. The X even more so due the fact father to daughter X are not recombined and the mother to daughter is recombined.

[mkdexter]

Cousin prediction beyond 3rd cousin is just to random and small to be reliable as far as which cousin. The X even more so due the fact father to daughter X are not recombined and the mother to daughter is recombined.

and ..

One chromosome all by itself is not a reliable method in determining shared cM equals a certain genetic distance such as first, second or third cousin. The X is just one chromosome. It takes a comparison of all 22 or 22 plus X chromosomes with another person to determine if they are a certain relation or not. This is true for any chromosome taken by itself, even the X.

Here is an example of why one chromosome alone can not be used to make a prediction. In this example it doesn't matter if it is 8, 14, or 16, just as it wouldn't matter if it was 1, 2, or X.

If I compare my three tested children to my wife's mother (mat grandmother):

Daughter 1 does not match on chromosome 14, yet the sum of all 22 chromosomes is 1614.40cM; a typical grandchild/grandparent shared sum. If I was trying to calculate the relation based on chromosome 14 I'd miss the prediction.

Daughter 2 does not match on chromosome 8, yet sum of all 22 chromosomes is 1548.68cM; a typical grandchild/grandparent shared sum. If I was trying to calculate the relation based on chromosome 8 I'd miss the prediction.

Son 1 does not match on chromosome 16, yet the sum of all 22 chromosomes is 1558.59cM; a typical grandchild/grandparent shared sum. If I was trying to calculate the relation based on chromosome 16 I'd miss the prediction.

On the flip side, my wife's father shares all of chromosome 14 with Daughter 1, all of chromosome 8 with Daughter 2 and all of chromosome 16 with Son 1. Again, impossible to use all by itself as a predictor.

No one can predict by the relationship based on quantity of shared cM on the X alone, no matter what they say they can do. They can say it is a parent/child/grandchild and that's about it accurately and if that was the case, the other 22 chromosomes would match too. Of course they can say maternal shared or not, but not what degree of relation (first, second, third cousin) the share is.

[irenych]

Thank you Tick and Matt. This is very useful and important information.
But my initial question was actually a little different. Let me rephrase (and I ask for your indulgence). Given the fact that this "South Asian" stretch appears at exactly the same spot in three continuous generations grandmother, mother, daughter) in the context of two indubitable recombinations, which do not seemingly affect IT (perhaps its location around the centromere is a factor?), and given the fact that there are no significant contributions of a "South Asian" nature anywhere else in the autosomal baggage, what is your best guess as to how old this "South Asian" connection could be? In other words, how many times could this stretch have been passed on as other "S.A." contributions were washed out? The stretch is about 20% of the X in the three tested cases.

[mihun]

http://forums.familytreedna.com/showthread.php?t=30475

I had a somewhat similar case with South Asian X.

[irenych]

http://forums.familytreedna.com/showthread.php?t=30475

I had a somewhat similar case with South Asian X.

I was wondering because I found a 1998 article by Mahtani and Willard which claimed that in the area straddling the X centromere the rate of recombination was extraordinarily slow (8 to 20 times slower than on the rest of the chromo). If their findings were still acceptable this might suggest a beginning of the process anytime between 1000 BCE and 800 CE. I wondered if more knowledgeable people had any ideas about this issue and the article. A part of the X functioning practically as a minor kind of deep ancestry marker.

[thetick]

There's not that much literature on X chromosome. It has sadly been ignored for the most part. I read before that sections of the X are believed to very old, but I'm not convinced there is enough evidence either way.

[mkdexter]

If their findings were still acceptable this might suggest a beginning of the process anytime between 1000 BCE and 800 CE.

Oh no that's not at all how recombination would work. Say the centromere never does become a point of cross-over as suggested, which lineage is passed down? It can't be both of her XX. It has a 50/50 chance of being one or the other.

Here are the choices from mother to child where her paternal X and maternal X must become only one X in her child:

1. paternal X is passed down, not maternal
2. maternal X is passed down, not paternal
3. maternal on one side of cross-over, paternal on the other is passed down
4. paternal on one side of cross-over, maternal on the other is passed down
5. maternal on the ends and paternal in the middle of two cross-over points is passed down
6. paternal on the ends and maternal in the middle of two cross-over points is passed down

Even if the centromere is never the point of cross-over, only one of the mother's lineages will be passed down in that area to her child, making it a 50/50 chance of even being the same information generation to generation.

There is no way the same information will always be passed down to any child of a mother over and over again in all of the descendants following for hundreds, if not thousands of years.

That said, BGA is SNP based not segment based. The only way for the SNPs to continue being of the same population is to keep both the maternal and paternal pools of the same group. Any varying admixtures will start to affect the outcome of the SNPs being passed down. That's the only way your sample population could run for hundreds of years and not be affected because as I just outlined, the maternal and paternal ancestry is constantly being mixed in each line descending.

Matt.

[irenych]

Thanks a mill for this excellent clarification! I'll muse over it and see how it applies to my case.

[irenych]

Thanks a mill for this excellent clarification! I'll muse over it and see how it applies to my case.

Just to complete things, here are the three McDonald X paintings (grandmother, mother, granddaughter, i.e. 3 generations)

It looks as though what occurred was possibility 6 (with a variation?). The granddaughter gets her ME from her father's mother (there is nothing in her mother or maternal grandmother along those lines), and she also gets one type of South Asian from the same source (for the same reason). Possibly she might also be be getting SA from her father's mother in the centromere connected stretch, because Gedmatch only finds S.A. identity between grandmother and granddaughter in the first part, and not the last part.

The Gedmatch genetic distance analysis suggests about 6-8 generations.

Would this sound about right?

BTW there is no S.A. in the granddaughter's father's autosomals, and practically none in the mother's.