It appears in our surname project anyone with big y has disappeared from the spreadsheets and results pages but they appear to be their in member subgrouping

开云体育

>How reliable is the tree in FTDNA?

?

That depends on what you mean by "reliable". It is highly accurate, but incomplete. If you read point (1) on my recently posted message #8352, you'll see why.

?

In short, any tree is only as good as the tests that make it up. FTDNA's tree is the most complete and is kept up-to-date, but it can only contain a portion of the whole story because we can only test a portion of the men who have ever lived. The tests themselves also only test a slightly different portion of each Y chromosome, so the tree does not represent all the mutations that exist, and a few individual mutations may not be in their exact positions if they are not covered in all of a haplogroup's tests.

?

- Iain.

Hi Herman,

?

Thanks for uploading your data to the Warehouse. The Warehouse tree is not currently loading fully for me, so I'm struggling to check the exact details. However, there is some generic information I can give you.

?

TMRCA estimates depend on two main factors:

?

(1) The tree. A TMRCA is literally a time to most-recent common ancestor. Any dates you are given for R-Z159 will not be the TMRCA of all Z159+ men, but the TMRCA of all the Z159+ men whose test results go into the tree.

?

Let's imagine we have the test results of every living man who is descended from the Z159 founder. Their most-recent common ancestor may be the Z159 founder himself, but it could easily be several generations later since most of his descendant lineages will have died out. Let's say this true TMRCA is in 2000 BC.

?

But we know that any tree only contains a small proportion of all of these testers. The Family Tree DNA tree is the most complete, so it might miss a few of these earlier branches. Every man in R-Z159 at Family Tree DNA has the that they use to define R-Z159: Z159 itself, and Z157, Z158, Z350 and BY41065. It's possible someone will come along from an earlier branch that is positive for only some of these SNPs. This will create a new haplogroup that branches off between the current R-Z160 and R-Z159, and these five mutations will be split between these two new haplogroups. In this way, R-Z159 will be defined. For now, let's imagine that those branches exist, but are untested, so the TMRCA is a more recent 1900 BC.

?

Now let's consider what happens to the TMRCA if we take results away from the Family Tree DNA tree. R-Z159 is a large and well-tested group with five branches and . Let's look at the : it has only 153 members, but still all five branches are represented. It often doesn't take many testers to establish an accurate TMRCA. But what happens if we keep cutting back testers so that the branches fall away? If we compare the and trees for R-A6535, we can see some very big differences. Many of the branches are not represented in the YFull tree, meaning several of the haplogroups in the Family Tree DNA tree are only represented by a single, much younger haplogroup in the YFull tree. YFull's definition of R-A6535 includes the SNP Y17443, which Family Tree DNA places . If you ask YFull what the TMRCA of R-Y17443 is, it is not the same as asking Family Tree DNA what the TMRCA of R-Y17443 is, it's the same as asking Family Tree DNA what the TMRCA of R-A6535 is. So a TMRCA depends on the completeness of a tree in question.

?

Taking this to its logical conclusion, you can imagine creating a tree where the only Z159+ tests were of two brothers. In this case, that tree would give a TMRCA for R-Z159 in the 20th Century. This is the correct answer for the tree, but that tree has a fundamentally different definition of R-Z159 to the one you are used to, since that tree's R-Z159 will include all the SNPs that those brothers share in common, not just those that are truly shared among all men.

?

Every tree is incomplete and subject to continual change but, as more people test within a branch, it quickly becomes much more stable and robust. This also mean that TMRCAs can quickly go out of date and need refreshed as new branches are found. The TMRCAs I computed back in 2016 are very wrong, often not because the calculations were wrong or inaccurate, but because the tree has changed underneath them. This is the largest cause of TMRCA discrepancies in smaller haplogroups these days.

?

(2) The methodology. There are many different methodologies for treating TMRCAs, but we can probably ignore those that deal solely with STR results for now (since all the trees we are discussing are based on sequencing tests with SNP results) and focus on the SNP and SNP+STR-based methods. There are several factors in the methodology that can change the computed TMRCA. These include:

• Whether the methodology uses only SNPs or both SNPs and STRs.
• The mutation rates used for SNPs and STRs.
• Whether any STR mutation rates are computed with individual mutation rates or averaged mutation rates.
• Whether and how back mutations and convergent mutations in STRs are accounted for.
• If multi-copy and null STRs are included and, if so, how order changes are allowed.
• If recLOHs are recognised and how they are treated.
• Which regions of the chromosome are searched for SNPs (FTDNA uses typically 25% of the Y chromosome, YFull only uses 15%).
• Quality criteria for determining the number of callable loci and calling SNPs within them.
• Whether tests are reprocessed (e.g. to a new reference sequence to extract more callable loci and SNPs).
• How causality is treated (sub-clades cannot be older than their parent haplogroups).
• How branch normalisation is computed (for brother haplogroups with different average numbers of SNPs).
• How branches are combined to form parents (e.g., are two brother haplogroups of very unequal sizes treated equally or not?).
• How ancient DNA is used to constrain the ages of haplogroups (e.g. the U106 mutation must be older than the oldest U106+ ancient sample).
• How genealogies are used to constrain the ages of haplogroups.
• Whether changes in population size need to be accounted for in the specific calculation, and how this is done.

These decisions individually have very different effects on the TMRCAs. Some (like mutation rates) scale the tree in its entirety. Some, like causality and normalisation treatments and ancient DNA, often affect large branches. Some, like treatment of null returns typically affect much smaller branches. These can combine to give significant changes in the TMRCAs, and represent the largest cause of TMRCA discrepancies in larger haplogroups between trees.

?

Quite which of these is true for the Y-DNA Warehouse, I can't currently tell you. However, the best TMRCA will be the one that comes from the largest testing database that best takes all these factors into account as best they can. The Family Tree DNA database is by far the largest testing database, so their TMRCAs for small haplogroups are almost always going to be more accurate. By and large, their Discover methodology is also highly accurate as it correctly takes into account most of these factors. They didn't take into full consideration all the factors in my - I think they could do better. However, the quibbles I have are minor and their methodology is the most advanced that currently automatically explores the whole Y-DNA tree. Consequently, I would generally advise people to take the TMRCAs on Family Tree DNA as their default unless they have very good reason to use another (e.g., if I've spent painstaking hours computing a better one based on the same data!).

?

Best wishes,

?

Iain.

Hi Shane,

?

It looks to me like the update at YFull is not complete.

?

It's impossible to have a haplogroup with one person. That rather defies the idea of a group! It's certainly not possible to define the common ancestor of one person except as themselves, so to establish a TMRCA for the haplogroup as they have done means something is amiss. That something is having other people in the haplogroup. Which other people those are, we don't yet know, but it may be you and Bertram. If that is the correct interpretation, then hold tight and they'll make further changes. Then we can see what they're up to.

?

Cheers,

?

Iain.