I thought I would share my experience with a pile-up because it was instructive for me. I was building a triangulation group and started getting suspicious because too many people matched the segment. I apologize for forgetting the details here. By too many I mean something like 25% of the people on gedmatch when I did 1-1 comparisons with kits I suspected of being matches. I was looking at a relatively short segment (3-4cM?) so all these matches didn’t show up in the standard one-to-many search. I remembered the “pile-up” phenomenon and though I didn’t actually know what it was, it sounded like it might be an explanation of what was going on. I started checking kits totally at random and found a similar high frequency of matches. Researching Pile-ups I found that it refers to the phenominum where the percentage of people that have a particular segment is unreasonable based on any possible relatedness. The plausible explanation is that this genetic combination is favored by natural selection. Next I just looked for articles on pile-ups, and sure enough, the area where my too-frequent matches occurred was one that had been identified in several studies as one where these “pile-ups” are found. The point here is that pile-ups are not an aspect of ones particular group of matches but occur at the same places for everyone. I don’t know of a central compendium of pile-up areas that have been identified, but it would be nice if there was one. Maybe one quick and dirty way to see if a segment that appears to match for too many relatives is the result of a pile up, would be to see if it occurs in a similar frequency in both our relatives and kits selected at random…. Thats what tipped me off. In terms of the argument that it is caused by natural selection, what makes sense to me is that people for whom this area gets recombined due to crossovers are at a disadvantage (less likely to survive or reproduce) compared to people for home this region is intact. After all our DNA does have a job to do! David PS Of course just ignoring short segments in my case, I would have avoided the issue. I think what happened was I started with a few matches for a segment that included the pile-up area, and started looking for kits with related surnames who I thought might have a short piece of the same segment. And too many did….. > On Dec 13, 2015, at 8:26 AM, genealogy-dna-request@rootsweb.com wrote: > > Message: 3 > Date: Sun, 13 Dec 2015 10:56:24 -0000 > From: Andreas West <ahnen@awest.de> > Subject: [DNA] At what number of matches (at the same loci) are we > talking about a pileup? > To: genealogy-dna <genealogy-dna@rootsweb.com> > Message-ID: > <hJSp0t3pb-zqwGRIfEOeuWFWvSJGGA9D-CxJbhscXGdALUpl0@smtp.1und1.de> > Content-Type: text/plain; charset=us-ascii; format=flowed > > Hi everyone, > > I hope we can find some consensus here and maybe some of you know even what > number (of matches at a certain loci) is used by AncestryDNA to identify > pileups. > > We're obviously not talking about 1000 here as that would give us 499500 > 1-to-1 comparisons to run between the 1000 matches. That's the main reason why > DTC DNA testing companies (and also GEDmatch) are interested to identify > pileups to limit useless calculations (which in the end will still not find a > single triangulated group (maybe) or it's too far back anyway, see the Timber > algorithm used by AncestrDNA to cut matches. > > a) I have 97 matches (at the same loci) for one of my kits (at the "X" > chromosome interestingly, it's a female person), which means 4656 > combinations. Is that number already a pileup? > > How about: > > b) 52 matches = 1326 combinations > > c) 36 matches = 630 combinations > > d) 23 matches = 253 combinations > > e) 18 matches = 153 combinations > > Where is the line to draw? At a, b, c, d, e or where? > > > What is the largest number of matches that you have in your triangulated > groups? > > We obviously don't want to miss out a large TG as it also means a lot of > people can "crowdsource" together and identify the CA much quicker than a > group of 3 can (usually means also more family trees to compare with). > > Thanks for your answers! > > Andreas (WEST) born BASSO > > My ancestors: [http://www.wikitree.com/genealogy/Basso-Family- > Tree-23](http://www.wikitree.com/genealogy/Basso-Family-Tree-23)