Y-DNA project member subgrouping by STR or by SNP results?

[DougInAmbler]

It seems to me there are two fundamentally different ways to subgroup, but I don't see the question addressed anywhere.

by STR (example Raymond project)
-members subgrouped into arbitrary subgroup names or subgroup numbers (e.g. 1 to 100) according to Genetic Distance between their STR results, i.e. STR haplotypes
-equivalent to looking for Y-DNA matches for each member and grouping them together where GD is less than 3, leaving the rest ungrouped
-advantage: simple; disadvantage: haplogroups only predicted not confirmed

By SNP (example Pratt project)
-subgroup names are meaningful names of SNP Haplotree branches (subclades, lineages..)
-Terminal SNP is very important, but only available for Big-Y members?
-advantage: haplogroup is confirmed (but only for those with many SNP tests?) disadvantage: results not clear for dummies

Questions:
1. Is it true, there are these 2 distinct ways?
2. Is SNP subgrouping the way to go, especially as more members test Big-Y, and "private" SNPs/variants are being revealed?

[Stevo]

I prefer using SNPs to determine subgroups. Of course, for those who haven't done SNP testing, one can use STRs (if they have enough of them to make it meaningful) to place them in likely SNP categories. For example, if you have a few members who have done the Big Y and have the same terminal SNP, you can place close STR matches (especially exact 111-marker matches) in that same category, as if they had done the Big Y themselves.

Of course, known relatives who are close matches can be placed in a category without a Big Y test.

In the old days, we pretty much had to look for STR haplotype clusters around which to organize things, because the Y-DNA haplogroups were so broad as to be nearly meaningless.

[DougInAmbler]

I guess most surname projects began with few members having confirmed Haplogroups, so of course sub-grouping by Genetic Distance of the STR results was natural. It conveniently groups them with possible STR matches, though many of them may be false.

Once a significant number get confirmed Haplogroups via SNP test, Y-111 or Big Y, it is probably better to sub-group by major Haplogroups, and to sub-sub-group by lineages (unique terminal SNP sequences).

The previous STR matches (both true and false) will likely disappear, to the annoyance of some. And the Haplogroup-grouping, though very accurate, may not offer immediate clues to find male relatives. But this will be improving steadily, as more take the Bigy Y test. It's the way of the future, I am told.

[Stevo]

That's true, and even close 111-marker matches, while sharing a Y-DNA ancestor, can be on separate branches of the same Y-chromosome line, as indicated by terminal SNPs.

For example, all the kit numbers in this R1b-BY168 graphic, except those on the far left under FT61490, and Selfe (53479, far right), are 111-marker matches of mine, yet Big Y testing has revealed quite a bit of branching.

R1b-BY168 subclades.jpg

[dbeezley]

As far as I can see, the "grouping" charts aren't very useful for SNPs. Even for STRs, some surnames come from so many different places and are so complicated with NPE's that groupings can look pretty messy. Many I've seen look rather random and not very useful. In the Beasley study I administer, the surname has a rather tight genealogical and genetic history, so the groupings look more meaningful. Of course there are those with not Y-STR matches and I have a hunch that even SNP groupings in that bunch wouldn't be genealogically meaningful. I could be wrong, but that' how I see it. In case you are interested, here is the grouping we have: https://www.familytreedna.com/public...ame=ycolorized. As you can see, we have very few Big Y tests be we are getting more, particularly in the Blue Group. Even as we get more, however, this grouping chart would be of little use. The sort that Stevo shows above is much more useful and that must be custom made.

[Stevo]

I just use Paint to make my Y-DNA tree graphics, but it does take time. For me it's fun though. It helps me to get a handle on the results.

[mattn]

As far as I can see, the "grouping" charts aren't very useful for SNPs. Even for STRs, some surnames come from so many different places and are so complicated with NPE's that groupings can look pretty messy. Many I've seen look rather random and not very useful. In the Beasley study I administer, the surname has a rather tight genealogical and genetic history, so the groupings look more meaningful. Of course there are those with not Y-STR matches and I have a hunch that even SNP groupings in that bunch wouldn't be genealogically meaningful. I could be wrong, but that' how I see it. In case you are interested, here is the grouping we have: https://www.familytreedna.com/public...ame=ycolorized. As you can see, we have very few Big Y tests be we are getting more, particularly in the Blue Group. Even as we get more, however, this grouping chart would be of little use. The sort that Stevo shows above is much more useful and that must be custom made.

I was curious as to why numbers 65 and 66 are marked as unmatched. They match exactly, are they too closely related to one another to make them into their own grouping?