Surname Genetic Distance Puzzle

[wcole67]

We have a chap in our COLE surname group who has taken the 37 marker test. His results at 25 markers show a 1 GD mutation (with the rest of us) at fast marker 464c.
Continuing on....He then has a 1 GD mutation at fast marker 456, a 2 GD at fast marker 570, another 2 GD at fast marker CDYa, and a 1 GD at marker 442. So, a 1 GD at 25 markers and a 7 GD (total) at 37 markers.
So he's quite closely "related" at 25 markers, but, apparently totally "unrelated" at 37 markers, according to the FTDNA tutorials. And remember... the same surname! Anyone have any reasonable anwers to this vexing situation?
Thanks...

[T E Peterman]

This just goes to show why it is important to ramp up to higher resolutions as they become available. You should probably go for 67 & see how much, if any, the genetic distance increases.

Timothy Peterman

[GhostX]

I completely agree with Timothy that an upgrade to 67 markers might be very useful in your case (or for anybody, for that matter). For all you know, your relatedness might fall back into the other direction (increase) after you've tested more markers.

In your case, it appears to just be the luck of the draw that you happened to get a good match on 25 markers. This is a perfect example of how testing a small number of markers can present rather misleading results. The fact that you do share a surname may indeed provide evidence that you are related, but the genetic distance seems to indicate that it is a very distant relation at best (many hundreds of years). The 67-marker test will help answer that question.

Alternatively, the shared surname may just be a complete coincidence (especially if it's a relatively common surname). In my own surname project, which includes 20 members sharing a relatively uncommon surname, there appear to be at least a dozen completely unrelated (or very loosely related) lineages! There's not a single perfect match at more than 12 markers, and very few at 12 markers. This tells me that my surname may have arisen independently several different times throughout history, or else that there were a number of non-paternal events that caused people's previous surnames to change to mine (in other words, their surname isn't really supposed to be the same as mine, biologically speaking... their ancestor may have simply been adopted by one of my ancestors).

[vinnie]

We have a chap in our COLE surname group who has taken the 37 marker test. His results at 25 markers show a 1 GD mutation (with the rest of us) at fast marker 464c.
Continuing on....He then has a 1 GD mutation at fast marker 456, a 2 GD at fast marker 570, another 2 GD at fast marker CDYa, and a 1 GD at marker 442. So, a 1 GD at 25 markers and a 7 GD (total) at 37 markers.
So he's quite closely "related" at 25 markers, but, apparently totally "unrelated" at 37 markers, according to the FTDNA tutorials. And remember... the same surname! Anyone have any reasonable anwers to this vexing situation?
Thanks...

As with other postings today, it all depends on what your goals are. How are you defining "related" for the purposes of your project? How far back in time do you want to go to showing some degree of "relatedness"? It's notable that all the differences you cite between this person and the others are on fast markers. Do these markers have any particular significance in your haplogroup/subclade? Some people disregard CDY because it's so susceptable to mutation; some people count every step difference on 464 a-d, but others don't. There are studies/projects (Kercher's?) that show mutations can occur much more frequently in some family lines than in others of the same haplogroup. Upgrading to 67 markers and doing all snp tests possible is the best thing you can do, but remember that the FTDNA tutorials are based on averages, and there's still significant discussion on the mutation rates of many markers.

[Clochaire]

Or alternatively . . .

Or alternatively, the labs are just rushing through the motions and producing a garbage quality product.

I've been for 4 months now trying to get three labs to reconcile 2 discrepancies on a 23-locus haplotype. But they can't or won't.

So naturally I have to wonder how far the testing failures really go.

I did get one explanation that seemed plausible on the surface--that there may be mutations in non-repeat primer binding regions, and that different labs may use different primers. But then I discovered that, at least for 1 of the discrepancies, the labs are using the SAME PRIMER.

I did get some tracings from one lab, which was amusing enough. One of them actually had automated messages from the analysis software stating that the default base call had been overridden manually.

Oh, that and the allele ladders presented don't match the observed range of alleles in the literature.

But those aren't big problems, right? After all, the labs collected their money from me, and that's what really matters, right?

Ah, what am I going on about? 21 of 23 ain't bad. And it's not like they convict people of crimes based on this technology, right?

[GhostX]

Ah, what am I going on about? 21 of 23 ain't bad. And it's not like they convict people of crimes based on this technology, right?

Hehe, there's food for thought... and I'm getting hungry.

[wmccown]

Surname Genetic Distance Puzzle

All of the answers have been interesting. All I have to suggest is that if the match shows an FTDNATip icon to the right of his name, click on it and see if the probablitly of a Most Recent Common Ancestor (MRCA) looks promising. What is promising? I was only looking at the highest percentage at the most distant number of generations. Barry McCain, founder of FTDNA Ulster Heritage Group, www.ulsterheritage.com, says that percentages ranging from a low of 50% to 75% can be worth further study, although he personally uses 70-75%. I am confident that none of my matches could have been as recent as 9 generations ago. Using that to recalculate the MRCA, all of my 67 marker matches ranging from 60/67 to 62/67 showed that they all scored from 69.8 to 82.4% 17 generations ago and upwards of 99.7% 33 generations ago. It is worth the effort to see if your MRCA calculations tun up anything interesting.

[wcole67]

Surname GD Puzzle

As the fellow who started this thread, I have found all of your answers really thoughtful (in the technical sense).
I agree that a 38 to 67 marker upgrade would be a wise move for this testee, however I doubt he will make the effort. It was a bit of a struggle to get him to take the 13 to 37 marker upgrade in the first place.
As an early believer (in this technology), these results really disturb me. I am starting to think that all tests of 25 markers or below are just one big waste of time and money (and yes, I am familiar with some of the good things you can find out from these tests). But now, even with a good match and a common surname, the 25 marker results cannot be trusted. Let's carry this to extremes...what now if someone has a good match with his fellow testees at 37 markers and the results fall apart from 38 to 67 markers? What then??
Incidently, this situation happened to me. (2 GD at 37 markers, 21 GD at 67 markers!). Luckily, (from a genealogical standpoint) it was an error on FTDNA's part. A series of complaints by me finally resolved it. I think FTDNA got someone elses's test results mixed up with mine, but never gave me a definitive answer as to how and why the mistake occurred. FTDNA did give me a mild apology for their grevious error after several months of total confusion.
Excuse me for the diversion....
Getting back to the 37 marker mismatch...My gut feeling is that all of us testees are related to this surnamed Cole but far into the distant past, probably somewhere in England. All of us other matching Coles (almost a dozen) trace our MRCA to John Cole 1605-1685 of Hartford, CT, an immigrant from England. I do have a problem with the possibility that his Cole surname is "just a coincidence". I keep thinking of all those matching markers (32 matches out of 37)and a 1 GD in the first 29 markers.
If this is so, one interesting tid-bit emerges..the Cole surname has been around for a heckava long time!
Thanks....

[lsmc1022]

I have a general question regarding surnames, how long have they been in use in western Europe?

[FredSpringer]

Why Upgrade Further?

I thought the reason to upgrade to higher number markers was to filter out the false matches at the lower number marker tests. Once those matches are eliminated, you are left with matches which are more likely related to you. The reason to upgrade even further to a higher number of markers would be to filter out even more possible false matches, or substantiate the matches you previously found.

Once you have eliminated a match by upgrading to more markers, I can't think of what information you would garner by seeing just how unrelated you become.

[wcole67]

why upgrade further?

I started this thread.
I've been re-thinking my options too and I believe you (Fred Springer) are quite right. The 37 marker test has already proven either a no-relationship or very distant relationship between this Cole and the rest of us who have a good match. If there were no further mismatches at 67 markers, nothing would be gained (the GD is still too great). If there were more mismatches at 67 markers, it would be information that we already had determined at 37 markers. My conclusion (like yours) would be that a 67 marker test would be a waste of time and money. We cannot "force" more testing to improve a non (or extremely) distant relationship.
I wish FTDNA would better explain these problems.
Thanks

[GhostX]

I started this thread.
I've been re-thinking my options too and I believe you (Fred Springer) are quite right. The 37 marker test has already proven either a no-relationship or very distant relationship between this Cole and the rest of us who have a good match. If there were no further mismatches at 67 markers, nothing would be gained (the GD is still too great). If there were more mismatches at 67 markers, it would be information that we already had determined at 37 markers. My conclusion (like yours) would be that a 67 marker test would be a waste of time and money. We cannot "force" more testing to improve a non (or extremely) distant relationship.
I wish FTDNA would better explain these problems.
Thanks

Upgrading to a 67 marker test is never a waste of money. In the future, it is likely that you will get more matches, as more people get tested (the field of genetic genealogy is a rapidly-growing one), and you'd have much greater power to evaluate those matches as they arise if you have a 67-marker test under your belt.

Considering your particular case, you are sort of correct that you can't force a match from a previously-tested mismatch, but you can reveal with greater accuracy whether he was actually a mismatch in the first place. For example, if your mismatch was a 30/37 result, that would seem to indicate that the person is not related to you. But let's say that you upgrade to 67 markers, and none of the additional markers are a mismatch (unlikely, but it's possible). Then your results would be 60/67. That completely changes your relatedness from "not related" to "related," according to FTDNA's definitions, and shows that the 37-marker test wasn't accurate enough to give you the correct result.

Again, this is pretty unlikely, and I probably wouldn't do the upgrade if your only goal was to double-check this one particular person. There are other databases out there than FTDNA's though (notably Ysearch, and SMGF), where you might find that a 67-marker test could come in very handy. It's pretty frustrating to discover somebody who might be a good match, only to think to yourself, "If I (or they) had only tested 67 markers, I'd have a much better answer!" (I speak from personal experience).

[wcole67]

Thanks for all the input...

Our group of about 10 Cole males who trace our MRCA back to John of Hartford, CT about 11 generations back, are interested in close GD matches, not distant GD's (non-matches) based on FTDNA testing results. The 37 marker test has told us that this Cole is "off the chart", as far as we are concerned. There is no sense in trying to "force" him into becoming part of our "Hartford" group by telling him to take a 67 marker test. All of our group has taken the 37 marker test since this is our so-called unofficial minimum standard to be part of our "Hartford" group, and about half of us have upgraded to the 67 marker test. As you know, the curve flattens out considerably after 37 markers, so not much additional information is gained in those 30 extra markers, so there were no surprises in any of our 67 marker results (a couple of us picked up an additional 1GD).

Again my original concern, which has not been solidly answered, is we have a "closely related" Cole (1GD) at 25 markers who turns into an "unrelated" Cole (7GD) at 37 markers. Shouldn't the 37 marker test at least maintain a "related" situation? (say, a 2 or possibly 3GD?)

Thanks...

[GhostX]

Again my original concern, which has not been solidly answered, is we have a "closely related" Cole (1GD) at 25 markers who turns into an "unrelated" Cole (7GD) at 37 markers. Shouldn't the 37 marker test at least maintain a "related" situation? (say, a 2 or possibly 3GD?)

I can see why you might be wondering about that question, but unfortunately for statistical reasons, it doesn't always work out that relatedness will be preserved when upgrading to more markers. It's just the luck of the draw as to which panel of markers will show the greatest number of mutations between any given two people. Sometimes they're evenly distributed among all markers, and sometimes, by pure chance, they tend to be clustered in one end or another of the series. It's as if you threw a handful of change on the floor--occasionally a disproportionate fraction of the coins will end up getting clustered in on area, rather than evenly spread across the floor.

Furthermore, you really need to consider the difference between "GD" and "relatedness." GD will usually increase, the more markers you have tested, simply because you'll be more likely to discover mutations, the more markers you test. GD is simply a count of how many markers mismatched, so the more markers you test, the higher will be the mismatch count, generally speaking. Relatedness (not actual relatedness, but calculated relatedness--there's a big difference) however, can either increase or decrease as you test more markers (and only really becomes reliable once you've tested a lot of markers). Having a GD=3 for a 12-marker test means the people compared are unrelated, but having that exact same GD at 67 markers would mean that those people are probably tightly related.

The main limitation that I have found for testing to 67 markers is that most people don't test that many markers (unfortunately), so markers #38-67 are just never compared against most other people. In my surname project, I am the only person who has tested 67 markers. I think (or at least hope) that this trend is going to change in the upcoming few years. Once more people upgrade to 67 markers, a lot of people are going to discover that they weren't as closely related to people as they thought they were (and conversely, a lot of others are going to find better support that their previously-discovered match is accurate).

[hdw]

I can see why you might be wondering about that question, but unfortunately for statistical reasons, it doesn't always work out that relatedness will be preserved when upgrading to more markers. It's just the luck of the draw as to which panel of markers will show the greatest number of mutations between any given two people. Sometimes they're evenly distributed among all markers, and sometimes, by pure chance, they tend to be clustered in one end or another of the series. It's as if you threw a handful of change on the floor--occasionally a disproportionate fraction of the coins will end up getting clustered in on area, rather than evenly spread across the floor.

Furthermore, you really need to consider the difference between "GD" and "relatedness." GD will usually increase, the more markers you have tested, simply because you'll be more likely to discover mutations, the more markers you test. GD is simply a count of how many markers mismatched, so the more markers you test, the higher will be the mismatch count, generally speaking. Relatedness (not actual relatedness, but calculated relatedness--there's a big difference) however, can either increase or decrease as you test more markers (and only really becomes reliable once you've tested a lot of markers). Having a GD=3 for a 12-marker test means the people compared are unrelated, but having that exact same GD at 67 markers would mean that those people are probably tightly related.

The main limitation that I have found for testing to 67 markers is that most people don't test that many markers (unfortunately), so markers #38-67 are just never compared against most other people. In my surname project, I am the only person who has tested 67 markers. I think (or at least hope) that this trend is going to change in the upcoming few years. Once more people upgrade to 67 markers, a lot of people are going to discover that they weren't as closely related to people as they thought they were (and conversely, a lot of others are going to find better support that their previously-discovered match is accurate).

Even a seemingly close match at 60+ out of 67 is no great shakes. I have had people say they envy me my four matches at 61/67, but the calculator on my FTDNA personal page says there is roughly a 97% likelihood that these people and I had a common ancestor 24 generations ago, which by my calculations is about 800 years, so we're not talking about kissing cousins here.

Harry