How Many Ancestors Share Our DNA? Inheritance probability

Discussions concerning Autosomal DNA.

Moderator: CeCeMoore

User avatar
Posts: 269
Joined: Mon Apr 23, 2012 6:15 pm
PostPosted: Wed Aug 22, 2012 10:33 pm
While every ancestor is important, because by him you exist, by DNA you share with them, in my opinion an important question is not discussed currently AFAIK.
Luke Jostins had a blog entry in Nov. 2009 about inheritance probability with the result that the DNA Ancestry fades out at 125. This means by his calculations normally you don't get pieces of DNA from more then this number of ancestors; at 10 generations away you share DNA only with about 10-12% of your ancestors!
Image

I've found no similar or newer calculations (or papers) on this topic and wonder why so many speak about admixture etc. but none looks at this interesting fact? Many have paper trail to some etnicity many generations ago, but wonder why admixture tools can't show that ancestry. I also think this is crucial for auDNA population studies etc.

Please spread this in the hope some insider can give a current opinion on Luke Jostings findings:
http://www.genetic-inference.co.uk/blog ... e-our-dna/
DNA/Admixture Central Europe (Alps, Tyrol, Dolomites, Raetia); Y-DNA J2a-L1064, J2a-L210, R1a-M17, R1b-U106 (L48-); mtDNA J1b1b, J1c1d, U5a2b2, U5b1b1. Projects : J2-M172, J2a-PF5197, Alpine DNA, ISOGG Wiki

Posts: 542
Joined: Wed Mar 14, 2012 2:16 pm
PostPosted: Thu Aug 23, 2012 9:53 am
I have myself pointed to this calculation before, in other forums. It is unpopular and politically incorrect, because of its implications:

- Genealogy is not the same as genetic ancestry. An enormous family tree with thousands of ancestors is wonderful historical research, but most of those distant ancestors are no more genetically related to the descendant in question than adoptive ancestors would be.

- In particular, the pride of descending autosomally from some medieval royalty is rather hollow, since the probability that one has actually inherited any of that royal's DNA is vanishingly small.

- Pop-science writers of a certain political stripe like to claim that all humans, or at least all those of European ancestry, descend from one individual who lived only 1000 or 1500 or 2000 years ago. Such a claim is not only highly insulting (because it glosses over oppressive social stratifications like serfdom) but also utterly meaningless (because only a very small number of Europeans would have actually inherited that individual's DNA even if the claim were true).

On the positive side, this calculation underlines the importance of yDNA and mtDNA studies, because they investigate specific genetic ancestors rather than a collective of mostly unrelated (genetically) people.
User avatar
Posts: 202
Joined: Sun Apr 01, 2012 5:38 am

YDNA:
R-L21+, L226+
PostPosted: Thu Aug 23, 2012 9:51 pm
The chart implies that individuals were not very mobile, therefore, they intermarried within the limited genetic pool that was available. If your ancestors lived in a remote smaller village in Italy for 20 generations, this chart may be accurate. If your ancestor migrated to America in the early 1600s and kept moving westward for lower cost land, this estimate would not be accurate. Mobility is a key assumption in this curve and it assumes limited mobility which is not always the case. I have no proven ancestors outside of the US yet have over 100 proven ancestors - with only one intermarriage around 1870 where they slowed down their southward and westward migration for a couple of generations in Alabama. My pedigree is probably an exception to most but my ancestors were apparently always on the move for lower cost land that became available.

For Y-SNPs - who cares what the total DNA is shared. It is not relevant as long as we have the .01 % of the Y-DNA of Y-SNPs needed to track our ancestors. atDNA has a limit between 1800 and 1850 before the process of recombination leaves little common atDNA left - this is 99 % of our DNA - but with Y-SNPs and mtDNA, this is not relevant.
Last edited by Robert Casey on Thu Aug 23, 2012 10:18 pm, edited 1 time in total.

Posts: 542
Joined: Wed Mar 14, 2012 2:16 pm
PostPosted: Thu Aug 23, 2012 10:07 pm
Robert Casey wrote:Mobility is a key assumption in this curve and it assumes limited mobility which is not always the case. I have no proven ancestors outside of the US yet have over 100 proven ancestors - with only one intermarriage around 1870 where they slowed down their southward and westward migration for a couple of generations in Alabama.

No, you miss the point. The curve has nothing whatsoever to do with mobility--it is based entirely on the biology of chromosome recombination.

The author first takes the simplistic case in which no chromosome recombination ever occurs. In that case, a human has only 46 chromosomes and so can have only 46 genetic ancestors at any particular generation distance.

The author then took into account the latest and best research on chromosome recombination rates:
---
To do this, I put together a computer simulation of recombination, using data from Chowdhury et al‘s large population study of recombination; they found that recombination rates vary from person to person, and especially between genders, with significantly more recombination in women than in men (see this graph for the data I used). I simulation individuals with the 22 non-sex chromosomes, and each person had their own recombination rate, chosen at random, dependent on their gender. I simulated sexual reproduction with unrelated individuals, and checked whether their DNA was present in their descendants N generations in the future.
...
The number of genetic ancestors starts off growing exponentially, but eventually flattens out to around 125 (at 10 generations, 120 of your 1024 genealogical ancestors are genetic ancestors).
---

Please understand that the author is not referring to paper genealogical ancestors, but actual genetic ancestors.

Here is the study whose recombination rates he used: http://www.plosgenetics.org/article/inf ... en.1000648 .
User avatar
Posts: 202
Joined: Sun Apr 01, 2012 5:38 am

YDNA:
R-L21+, L226+
PostPosted: Thu Aug 23, 2012 10:22 pm
But who cares about only looking at atDNA that recombines ? With Y-SNPs and mtDNA we can trace back 10,000 of years with no problem - just because there is little atDNA passed on after a couple of hundred years - atDNA is not the only part of DNA we look at.
User avatar
Posts: 269
Joined: Mon Apr 23, 2012 6:15 pm
PostPosted: Thu Aug 23, 2012 10:58 pm
Robert Casey wrote:But who cares about only looking at atDNA that recombines ? With Y-SNPs and mtDNA we can trace back 10,000 of years with no problem - just because there is little atDNA passed on after a couple of hundred years - atDNA is not the only part of DNA we look at.

I agree as lgmayka already said, Y-DNA and mtDNA are powerful tools.
But for the correct use of also useful and interesting atDNA Admixture tools we need evidence for recombination rates and genetic ancestors numbers.
lgmayka wrote: I have myself pointed to this calculation before, in other forums. It is unpopular and politically incorrect, because of its implications
I understand that it is unpopular, but not why it is politically incorrect.
It is simply crazy how so many paid and citizen scientists do PCA, IBD, admixture and never mention this topic or try to estimate how this affects admixture.

Here a long quote from ISOGG Facebook discussion about this topic (I hope it is OK fro Matthew):
Matthеw H. wrote: The overwhelming majority of autosomal SNPS, particularly the notable ones (RS etc) used for "fingerprinting", are from thousands of years ago. A given SNP could have been common among lots of people, for example, Neanderthal SNPS, but if (say, for instance) the only people who survive today that have them are from eg. Silesia. Those neanderthal SNPs could have been present in Celts, Slavs, maybe Mongols- but if the only surviving MRCA is from Silesia, and those SNPS aren't sequenced in any representative sample from earlier in the 3rd century, I would expect that the SNP would be dated back to that period. It would be useless to tell if the Slav, Celt, or Mongol forebear was involved except by extremely careful examination of pure reference samples, but if you're from a few middle-age melting pots, all bets are off.

There's a lot of reasonable work done with SNPs that through a lot of testing, you find people who have it and don't, and correlate that with other snps to find the branching point and a rough time period. But the gain or loss of a SNP is not tell-tale in autosomal DNA- in Y or mtDNA, you're extremely unlikely to lose a SNP a few generations later. But in autosomal DNA, the branch points where someone loses a SNP can be as early as one generation later. That's the seat of the paradox- the event where you gain a SNP, and where your family loses it, do not provide information about each other. It's not falsifiable- not having a SNP doesn't mean anything.

And as such, PCA charts cannot be "sorted" in chronological order the way that histories tell who begot who when. Useless by themselves for near genealogy timeframes, useless without reference genomes of people you "dug up" for historical timeframes.

It's the luck of the draw which ones get carried forward from a much larger pool. Unfortunately, the main reason for spending the money on the test is going to be justified by how much info you can get out of it- and that requires a very *socially promiscuous* attitude to your raw data, because only data that is public can be of any help in making connections. If you're spending beaucoup bucks on testing and don't share, you've thrown that money away, and with the ridiculous focus on privacy in our culture, we are never going to emotionally be able to let go.

I think if people were made more aware of the fact that their DNA, while telling everything about the circumstances of their birth and "fingerprint", tell nothing about what kind of person they are or what their future holds- that's a good message that we need to consider. Until we purge Genetic Genealogy of its often-racist past and our social mores, and encourage data sharing, we won't actually unlock the genuine inter-relatedness of all people. The leap from tribalism to globalism is one small step- you know how that saying goes.
Last edited by ChrisR on Wed Sep 12, 2012 7:00 am, edited 1 time in total.
DNA/Admixture Central Europe (Alps, Tyrol, Dolomites, Raetia); Y-DNA J2a-L1064, J2a-L210, R1a-M17, R1b-U106 (L48-); mtDNA J1b1b, J1c1d, U5a2b2, U5b1b1. Projects : J2-M172, J2a-PF5197, Alpine DNA, ISOGG Wiki

Posts: 542
Joined: Wed Mar 14, 2012 2:16 pm
PostPosted: Fri Aug 24, 2012 12:31 am
Robert Casey wrote:With Y-SNPs and mtDNA we can trace back 10,000 of years with no problem - just because there is little atDNA passed on after a couple of hundred years - atDNA is not the only part of DNA we look at.

We agree, and that is exactly the point.

If we consider my ancestors of 30 generations ago, I know for a fact that my Y chromosome came from exactly one man of that era, and my mtDNA from exactly one woman of that era. If we had sufficient and correct historical records, I could determine the name of each!

In contrast, my autosomal DNA probably descends from about 120 people of that era, but I have no way of knowing which people are my genetic ancestors, nor which chromosome segment came from which person.

This is why many of us focus on Y-DNA and mtDNA. Y-DNA and mtDNA can give us specific historical knowledge, theoretically all the way back to "Adam" and "Eve"; whereas atDNA cannot tell us anything specific beyond the last few generations.
User avatar
Posts: 61
Joined: Fri Jun 15, 2012 5:42 pm
Location: Hong Kong
YDNA:
R1b, L21, DF21, L720
MtDNA:
J2b1a2
PostPosted: Fri Aug 24, 2012 12:03 pm
What I'd like to know is with my fairly well-documented family tree, where did all of the people on 23andme and FF come from that I share matches with but no documentation? Does recombination cause all sorts of false matches? Or are there very old chunks that just keep getting passed down without shrinking for many generations?
Y DNA lineage ancestor: Isaac Farris (b. 1787 in Virginia, USA)
mtDNA lineage ancestress: Ruth Stovall (b. 1731 in Virginia, USA)
Oldest known of any line: Ralfe Hurte (b. ca. 1410, Ashbourne, Derbyshire, England)

Known Surnames of 3rd gr.-grands:
Farris, Ramey, Montgomery, Summers, Tupman, Reynolds, Griffin, Atkinson, Powell, Witt, Dunaway, Woolery, White, Nickerson, Griest, Humbert, Mullin, Peacock, Heeter, Peters, Kern, Law, Meade, Conley, Estep, Lemaster, May, Caudill, Cooper, Salyer
User avatar
Posts: 269
Joined: Mon Apr 23, 2012 6:15 pm
PostPosted: Wed Sep 12, 2012 7:16 am
At DNA in Forensics 2012 I've spoken to C. Tyler-Smith about this topic and he thinks the inheritance probability is a major task for medical research and in the next time probably some researcher will pickup this for the medical field as more and more high coverage sequencing data gets available. For example in Netherlands they have a new project sequencing families (250 Dutch trios) - look at http://www.nlgenome.nl/

@JAFarris: I think like for Y-DNA (surname timeframe) FTDNA is over optimistic with FF matches. When Luke Jostins calculations are correct, this means exactly that auDNA will not split further at a certain point. But we need evidence from family sequencing where the boarder is.
DNA/Admixture Central Europe (Alps, Tyrol, Dolomites, Raetia); Y-DNA J2a-L1064, J2a-L210, R1a-M17, R1b-U106 (L48-); mtDNA J1b1b, J1c1d, U5a2b2, U5b1b1. Projects : J2-M172, J2a-PF5197, Alpine DNA, ISOGG Wiki

Posts: 129
Joined: Sun Apr 15, 2012 1:03 am

YDNA:
R1a1a1a R-L664
MtDNA:
U5b2b2 PhyloTree
PostPosted: Wed Sep 12, 2012 8:39 am
As for me, I'm more interested in a historically accurate family tree than with my autosomal DNA (blended European ancestry). The problem is that word "accurate."
:lol:
Mitosearch: G986T
Next

Return to Autosomal DNA

Who is online

Users browsing this forum: No registered users and 0 guests