Amount of DNA shared

[Pauline]

I maybe missing something somewhere, and know that different companies do things differently, but I am puzzled as to why I am getting different amounts of matching DNA with the same people on different sites. I guess they may have done more than one test so the two sites may be not be comparing like with like, or is something else going on?

I did my autosomal test with Ancestry and have uploaded my results to FamilyTreeDNA and GEDmatch, and some of my matches at Ancestry also show up one or other of these sites.

For example, I have a match with person A on Ancestry and this shows our shared DNA to be 29.1cM across 2 DNA segments. At FamilyTreeDNA the match information says we share 74cM, with the largest block being 34cM. (As yet I can't work out where my relationship with A might be.)

As another example, at Ancestry I share 18.9cM across 2 DNA segments with person B, while at GEDmatch it is 31.9cM with the largest being 10.5. However, when I look at the details at GEDmatch it shows two segments of 9.6 and 8.7cM, and a more distant estimated relationship than on the one to many matches page. (B and I already knew we were 4th cousins.)

Can anyone help me understand the differences, please?

 

[peter]

Ancestry use a phasing algorithm to reduce the chance of false matches - that's one of the reasons I now recommend Ancestry over FTDNA. If the results aren't phased then a segment of your DNA can appear to match your cousin's DNA even though you inherited some parts of that segment from your mother, and some from your father.

GEDmatch recommend that you always carry out a one-to-one comparison before contacting a possible DNA cousin. I don't know why the one-to-many search sometimes comes up with different results, but I suspect it's because they make small compromises to dramatically reduce the time it takes to compare 300,000 sets of results each with around 700,000 pairs of data points.

The amount of shared DNA is only a very vague pointer to the possible relationship once you get beyond 3rd cousin. In almost every case the relationship will be considerably more distant than you might expect from the averages shown in the ISOGG table - see the table in my DNA Special Newsletter for the reason why this is.

 

[Pauline]

Thank you, Peter - that all makes sense.

 

[Pauline]

The amount of shared DNA is only a very vague pointer to the possible relationship once you get beyond 3rd cousin

Does this mean that if an Ancestry match is estimated as a 3rd cousin match then there is a reasonable likelihood, give or take a generation, that they actually are a 3rd cousin?

I have a supposed 3rd cousin match with whom I share 137cM over 8 DNA segments but our trees do not seem to overlap at all, except in a broad geographical sense. Which possible explanations should we be considering?

 

[peter]

Does this mean that if an Ancestry match is estimated as a 3rd cousin match then there is a reasonable likelihood, give or take a generation, that they actually are a 3rd cousin?

I have a supposed 3rd cousin match with whom I share 137cM over 8 DNA segments but our trees do not seem to overlap at all, except in a broad geographical sense. Which possible explanations should we be considering?

Start by looking at shared matches. If there aren't any, or they don't provide any clues look at your 'brick walls' - assuming you're not both so fortunate as to have no 'brick walls' in the last 5 generations.

You should both upload your results to GEDmatch so that you can take a closer look at how you match.

 

[Pauline]

No shared matches, unfortunately. My DNA is already at GEDmatch, but I will have to check if hers is.

My assumption is, from looking at the geographical distribution of the other person's ancestry, that this is probably a match on my father's side. Also, I seem to have sufficient confirmed matches on my mother's side to make it unlikely that there would be no shared matches with a 3rd cousin on her side.

This narrows down the possibilities but still doesn't really get me anywhere.

I guess this is why I am interested to know how reliable a 3rd cousin match might be, and at what stage we need to start considering that for one or both of us, all is not as the paper trail might suggest.

 

[jorghes]

Well, I have 5 people tested on Ancestry (including myself) and as for the 3rd cousin listings, they are as follows:
Test 1 (me): 1 correct 3rd cousin, 1 maternal 3rd cousin - not enough information (illegitimate g-g-grandfather in that line).
Test 2: 1 correct 2nd cousin, correct 3rd cousin
Test 3: not enough information for either listed 3rd cousins - attempting to discover.
Test 4: 1 correct 3rd cousin, one without enough information.
Test 5: No 3rd cousins identified.

So I don't know how well you can define that from those results for me, most of the not enough information in the lists are all on my maternal side, and unfortunately that is the side where it seems those who are listed as my 3rd cousins don't have particularly much information at this point. I keep hoping that they are from the illegitimate section of the tree!

I don't think this necessarily proves how accurate or inaccurate the 3rd cousin tag is. So far, considering I have quite a few proven matches they seem reasonably accurate to me. I haven't check every single listed cousin, but those listed as 3rds I've tried - or I at least know in which area of the tree they belong due to other information (and other matches!)

I don't know if I've helped now, or just muddied the issue more.

 

[peter]

No shared matches, unfortunately. My DNA is already at GEDmatch, but I will have to check if hers is.

I think I'm right in saying that when you click the shared matches tab at Ancestry it doesn't necessarily show all of the matches you share with the other person. I'm struggling to think what the criteria are (possibly it excludes remote cousins).

My assumption is, from looking at the geographical distribution of the other person's ancestry, that this is probably a match on my father's side. Also, I seem to have sufficient confirmed matches on my mother's side to make it unlikely that there would be no shared matches with a 3rd cousin on her side.

I don't think you can make that assumption if you've only looked for shared matches at Ancestry.

I guess this is why I am interested to know how reliable a 3rd cousin match might be, and at what stage we need to start considering that for one or both of us, all is not as the paper trail might suggest.

It depends what you mean by reliable. Someone who shows up as a likely 3rd cousin is definitely a cousin, but whether they're actually a third cousin is impossible to know - whether for me, you, or even Ancestry - because of the randomness in the pattern of inheritance.

The big question is, do you both know who all of your 3G grandparents were? I certainly don't.

 

[Pauline]

I take the point about shared matches at Ancestry, though this observation was just an after thought - my assumption was primarily based on where the geographical distribution of our ancestors overlaps.

As for my 3G grandparents, I know all but two of them, both unknown fathers of illegitimate offspring. I guess though that if either of these unknown fathers is the link (which for various reasons seems rather less likely than other possibilities), then we would be looking at a half blood cousinship.