Hi Dado, all,

?

It's not so much a mistake, but it is misleading. If you take an individual person to whom you are related and no mutations have occurred in your first 25 markers to separate you, then you are most likely related within the last 600 years, so a range of 1400 AD to 1950 AD is correct.

?

If we go back 250 years ago (about fifth cousins), about* 50% of your cousins should be exact matches to you. Go back about 600 years, and only about* 2.5% of your 15th cousins should be exact matches to you. (*Exact numbers will vary depending on the mutation rates you adopt.) However, you might find you have more matches to you 600 years ago than 250 years ago because 2.5% of your 15th cousins might be more people than 50% of your fifth cousins. And there are back mutations and convergent mutations on top of this.

?

I've tried to produce my own version of the table (this was going to go into the new version of the textbook, but it got axed), which I'll try to reproduce here. This is my estimate of the conversion between genetic distance and years before present. The ranges are 95% confidence intervals:

?

0/12 = 150 – 3600
1/12 = 260 – 4800

?

0/25 = 100 – 1560
1/25 = 160 – 2100
2/25 = 230 – 2800

?

0/37 = 60 – 840
1/37 = 110 – 1080
2/37 = 160 – 1380
3/37 = 200 – 1630
4/37 = 250 – 2000

?

0/67 = 40 – 680
1/67 = 80 – 800
2/67 = 120 – 970
3/67 = 160 – 1140
4/67 = 190 – 1340
5/67 = 240 – 1550
6/67 = 280 – 1750
7/67 = 320 – 2000

?

0/111 = 20 – 450
1/111 = 50 – 510
2/111 = 70 – 610
3/111 = 110 – 700
4/111 = 140 – 780
5/111 = 160 – 870
6/111 = 190 – 990
7/111 = 220 – 1080
8/111 = 240 – 1200
9/111 = 270 – 1300
10/111 = 290 – 1430

?

You can see from this how going to larger numbers of markers really starts to narrow down the ranges of how closely you are related to someone, but also how broad the possible range of dates is for the 12 and 25-marker datasets in particular. It's important to caveat this: this table is only true for the average person. If you have few mutations, common mutations or back mutations in your line, your ranges will be older; if you have many mutations, or rare mutations and few back mutations in your line, your ranges will be younger.

?

Since this is no longer going to go into the textbook, I'll try to write it up properly and post it on my website, but that may take me some time.

?

Cheers,

?

Iain.

Actually, this is really why BigY (and sequencing tests in general) are so important - because they remove this ambiguity of descent: as long as you have at least one SNP mutation that defines a branch in your family tree, you know exactly how the other branches relate to it.

?

I would prefer to say Y-STR tests can be ambiguous rather than misleading, but their main problems (as I recently highlighted) are that STRs can mutate back to their original value ("back mutation") and that two different families can experience the same STR mutation ("convergent mutations"). In either of these cases, instead of adding to the genetic distance, a mutation can subtract from it.

?

Rather than relying on genetic distance itself, it is often best to get hold of the original Y-STR results. In the days before BigY, we had to group people based on their Y-STR values, and we would look for key combinations of values that formed tight clusters of closely related people, e.g. DYS492=13 predicts R-U106 with 97% accuracy, while a high probability of being R-DF98 comes from a DYF395S1=16-16 + DYS557=15 combination. We could often form smaller groups than this, and I have spend days of my life trying to merge people's STR results within R-DF98 into sub-groups before BigY testing really took off. But these groups weren't foolproof, and BigY testing has proved some and disproved others. This is a real driver behind our encouragement for anyone joining our project to upgrade to BigY - without it we can only guess approximately at where people fit in with others.

?

- Iain

Hi folks,

?

A few replies/comments to messages...

?

The idea behind the Y-STR genetic-distance-based matching criteria at FTDNA are that you should have about a 50% probability of matching any one person within the last 1000 years. If you look at your Y-DNA matches page, there is a little blue calender icon at the top-right of each match. Clicking on this brings up a "Time Predictor" table, which shows you how a match at a specific genetic distance should nominally(!) correspond to a period in time.

?

You might therefore expect the number of matches to have at each level to be constant. However, the number of people you will find at each matching level will depend on three things: (1) the mutations your family has in that set of 12, 25, 37, 67 or 111 Y-STRs; (2) the number of cousins you have in successive generations throughout the last 1000 years and beyond; and (3) how many of those cousins have tested with FTDNA.

?

The key thing that more STRs gets you is accuracy. The usefulness of your match list at any level relies on your family having had mutations on those STRs within a genealogical timeframe, and also relies on no-one else having had those same mutations. For example, my family has had two mutations in the first 12 Y-STRs, and these occurred many centuries ago. A couple of dozen people have tested at FTDNA from my broader family, which is about 600 years old. However, many people across R-U106 and R-P312 have also had these mutations, so I have thousands of irrelevant matches in my Y-12 match list. Most other people have similar stories.

?

It's very rare to find someone whose Y-12 matches are generally useful, since they normally consist of people who also match them at higher levels. So if you see someone who is a Y-12 match, but they have also taken a Y-111 test and you don't match them there, chances are you are not related within the last 1000 years or so, and may share no connection for thousands of years in the past. The fact that you share these first few Y-STR mutations but not the remainder of your mutations is likely due to a simple chance co-incidence of having the same mutations, not because you share mutations by descent.

?

As you increase the number of Y-STRs tested, the chances of someone from a completely different haplogroup sharing your Y-STR mutations by chance drops, so the number of irrelevant matches drops. How fast this happens depends on your haplogroup, and the three factors I mentioned above. If (like me) you still have hundreds of Y-25 matches, chances are these STRs aren't being very selective, and these people aren't relevant to you. If you only have a handful of Y-25 matches, or if (like Robert) you have more matches that mostly share the same surname, then your mutations are rarer or more numerous and these 25 STRs are being much more selective for your closer relations.

?

Again, the same check is valuable: if people have tested beyond Y-25 and don't match you at higher levels, probably they are not related to you within the last 1000 years or thereabouts.

?

For most people, Y-37 will at least give you matches within an historical timeframe, but many people will find from later testing that list this includes many of the matches that are 2000 or 3000 years ago. A small fraction will find matches that are even older, for R-U106 perhaps even including some in R-P312, so Y-37 still doesn't guarantee a close relationship for everyone. Then again, there are some people (like me) who don't have any matches in these timescales between about 1000 and 4000 years ago, so Y-37 provides a very clean separation!

?

I have yet to come across anyone whose 67-marker matches do not show their historical era ancestors. So the bottom line is that I would treat every Y-67 and every Y-111 match as relevant and, for most people, Y-37 matches are relevant too. This is now FTDNA's minimum testing level for a reason! Equally, while Y-25 and maybe even Y-12 matches are useful for some people, these people are fairly few and far between, especially at the Y-12 level. Their principle purpose is to identify really old tests where people of interest have not tested to the Y-37 level.

?

If you don't have any matches beyond Y-12 and Y-25, then it's likely that either no-one in the FTDNA Y-DNA database is related to you via your male line within the last 1000 years or so, or you have an unusual number of mutations in your Y-DNA. This normally happens if your family comes from somewhere that's severely under-tested (for R-U106, that might include places such as France or Eastern Europe) but can rarely happen if you have had unusual mutations in your Y-DNA (e.g. recent large-scale deletions, recLOHs, etc.). In this case, the SNP results and resulting haplogroup from a BigY test will give you an indication of where you fit into the overall structure of European history, but will not tell you much about your family's exploits in the genealogical era.

?

Cheers,

?

Iain.

I'll take the bait.

Genetic distances can't be "misleading" or "mistaken" (unless there is an arithmetic error). ?They are an objective measurement, whether calculated using Y-STRs or Y-SNPs.

It is true that researchers can make mistakes or be misled by data, but that's true no matter what form the data take. When it comes to genetic genealogy it always pays to use multiple lines of evidence.

Vince Vizachero

Our Project’s recent experience illustrates how Y-STRs can be misleading in predicting the branch of a family to which a new participant belongs, even at the 700 marker level.

?

???????????????????????????????????????????????????????????????????????????????????????????

The 18-year-old Acree DNA Project that I administer tested in its early years at Ancestry.com, but now tests at FTDNA and YSEQ. We were fortunate to discover during our first year that the ancestor of most Acrees was William Acree (c1710-c1767) of Hanover Co., Virginia, and that his descendants (61 of us currently in the Project) possess the rare, distinguishing microallele 13.2 at DYS385b. With the advent of Big-Y testing, we found that we also possess the slightly more widespread Y-SNP A2156, and that descendants of William’s son John uniquely possess the distinguishing Y-SNP A2155.

?

FTDNA’s Y-STR matching of our latest Project participant, a descendant of William’s son William Jr., at the 111-marker level, shows incorrectly that his first four matches are not with other descendants of William Jr., but rather with descendants of John (including me), at a genetic distance of 3 steps His first match with a descendant of William Jr .is shown fifth, with a GD of 4. His other matches, including non-Acrees, show GDs of 5-8 steps.? A known Acree fourth cousin once-removed of?his (also a descendant of William Jr.), who appears among his autosomal (Family Finder) matches, comes in tenth on the list, with a GD of 8. The Big-Y’s 700-marker comparisons are also misleading: He has 4 differences with me and 7 with this fourth cousin.

?

That’s certainly a misleading performance for Y-STRs. Fortunately, the Big-Y Block Tree (not the Big-Y match list) correctly groups our newcomer among William Jr.’s descendants, clearly apart from John’s descendants. This experience testifies once again to a consistently dependable performance for hierarchical Y-SNPs, in contrast to ambiguous Y-STR strings.

开云体育

开云体育

开云体育