Search for content in message boards

Multiple long matches from putatively unrelated families

Multiple long matches from putatively unrelated families

Posted: 30 Dec 2013 4:01PM GMT
Classification: Query
Edited: 30 Dec 2013 6:15PM GMT
I have found half a dozen people who have almost exact matches of 19.85 cM (4,600 SNP's) on Chromosome 18 and few other matches above 5 or 6 cM elsewhere. I have looked at what I can find of the family trees of these people (the people I have contacted have been very helpful). The best I can do is to put the six or so into three quite different tribes. All are Ashkenazi but their known roots (locations) are far distant. Has anyone else found so many multiple matches on a single chromosome with far distant....too far to be really related....individuals? Statistically, I would think that a longest block of this size MUST indicate a common ancestry not that many generations back.
Can anyone help with this puzzzle?
(I am helping two adoptees to find their roots. Both are among the half dozen with this particular near-exact match. Their birth dates are over forty years apart!)

wilf

Re: Multiple long matches from putatively unrelated families

Posted: 30 Dec 2013 7:45PM GMT
Classification: Query
This has been discussed a little bit on the gedmatch message board (also under DNA research) where people are encountering puzzling results.

Apparently, there are certain places on specific chromosomes that don't change much over generations so in doing genealogy they are less useful as they may reflect common ancestors too back in time to be useful.

Some experts call these hot and cold spots and I found this wonderful color graph on Kitty Cooper's website...it does show a big block on 18 in blue...(she color-coded those "cold" spots blue and the warmer colors show the places on the chromosomes that change more frequently and thus are more useful to researching common ancestors in a genealogical framework).

Others have posted specific hot spots on the gedmatch message board.

http://blog.kittycooper.com/2013/08/areas-of-faster-and-slow...



Re: Multiple long matches from putatively unrelated families

Posted: 30 Dec 2013 8:01PM GMT
Classification: Query
wonderful
this might be just what I needed

many thanks

I will check further tomorrow

wilf

Re: Multiple long matches from putatively unrelated families

Posted: 31 Dec 2013 6:45AM GMT
Classification: Query
Thinking more about this, I feel that the problem of these long matches with putatively unrelated people is not one of low mutation rates. Rather it is a problem of low recombination rates.
The mutations (SNP's) are there but one would have thought that they would have separated from each other by recombination during meiosis in a generation or two.
I am talking about a 20 cM stretch of 4600 SNP's. They are SNP's so they have already, long ago, mutated into that particular SNP, and will never (almost) mutate away from what they are. But they should separate from each other during a recombination event. Yet they seem not to do so.
So do you know about hot and cold spots for recombination events?

Re: Multiple long matches from putatively unrelated families

Posted: 31 Dec 2013 7:02AM GMT
Classification: Query
Thinking more about this, I think the issue is high or low recombination - not mutation.
I have 4600 SNP's on Chr 18 that match exactly between putatively unrelated families.
The SNP's have already mutated, long ago. That's why they are SNP's. But the long stretch should have split into segments by recombination in a very few generations. Yet it does not. So this long stretch is in a low (cold) recombination region. This is NOT what Kitty's message, or the paper she refers to, is dealing with. There it is high or low mutation rates. So I am still very puzzled.

wilfo

Re: Multiple long matches from putatively unrelated families

Posted: 31 Dec 2013 7:27AM GMT
Classification: Query
Wilfred those SNPs do change, that is wy they have been selected to sample. See my answer on my blog
http://blog.kittycooper.com/2013/08/areas-of-faster-and-slow...

So yes, blue areas will be matches further back in time.

Those 4600 SNPs may be several segments glommed together ... have you looked at the raw data to make sure it is not very hereozygous?

Ashkenazi DNA is very very difficult. I am still trying to figure it out. I will blog about it again soon when I have more to say. double the cousin predictions for ashkenazi ... too many shared ancestors a few hundred years back

Re: Multiple long matches from putatively unrelated families

Posted: 31 Dec 2013 7:32AM GMT
Classification: Query
The other problem with ashkenzi DNA is almost nobody has a family tree reaching much before 1800. So many of the new relatives I have found on my Norwegian line have shared ancestors much further back in time than that

This blog post of mine talks about that
http://blog.kittycooper.com/2012/12/common-ancestors-can-be-...

Re: Multiple long matches from putatively unrelated families

Posted: 31 Dec 2013 9:44AM GMT
Classification: Query
I have indeed seen your blog. I was referred to it on this message board and found it very relevant. But the issue is not one of high or low mutation rate. All SNP's are mutations.
What we have here are low recombination rates. This is quite different and I know of no explanation for this.
The 4600 SNP's are of course string together. That's why they match over this long distance.
Am I missing something here? Please correct me if I am wrong...I am only learning this stuff.

I think that I do understand the question of cousin cousin intermarriage in the Ashkenazim and I have my own way of dealing with this - I make histograms of a person's matches and look only at those matches that fall outside the range of the Gaussian distribution of the spread. I can explain this in detail if it is not clear.

wilf

Re: Multiple long matches from putatively unrelated families

Posted: 1 Jan 2014 9:46AM GMT
Classification: Query
I went to a wonderful seminar this morning by Giora Simchen, an expert on recombination in yeast. He shows that hot spots of recombination are likely to map well with hot spots of mutation rates since much mutation (90% perhaps) occurs during the recombination event itself and at that site. So the mutation hot and cold spots mapping suggestion of Kitty in a previous thread may well apply also to what we need - which is the recombination map. I have tried to find an up to date recombination map across the human chromosomes but have failed so far. If it isn't yet available it will be soon as a result of the ongoing genome analyses of human sperm.
If anyone can find a direct link to a modern recombination map across the human chromosomes, please add it to this thread wilf

Re: Multiple long matches from putatively unrelated families

Posted: 1 Jan 2014 1:16PM GMT
Classification: Query
I use the Rutgers site, which has several tools

http://compgen.rutgers.edu/RutgersMap/

This link is old, but it gives a visual representation of how the rate varies over each chromosome

http://web.archive.org/web/20070113005025/http://compgen.rut...

Note however that the cM unit is specifically designed to normalize the difference between hotspots and coldspots -- it's the *probability* of recombination. That is, for a 20cM segment, there is a 20% chance that a crossover point will occur somewhere within it, and on the flip side, there is an 80% chance that it will be passed on intact (if it survives the 50-50 chance of being passed on at all). There are some quirky regions of the genome where this doesn't work as well as expected. For instance, a chromosomal inversion, which might be found in some percentage of the population, will reduce the number of crossovers, since it won't align well with a normal chromosome.
per page

Find a board about a specific topic

  • Visit our other sites:

© 1997-2014 Ancestry.com | Corporate Information | New Privacy | New Terms and Conditions