Origin, Y-DNA of Sandzak Muslims

[eastara]

There actually are methods to predict subclades even from 17 markers, used by DNA project admins.
It's not Nevgen, but other methods which were developed on old mailing lists 15 years ago. Nobody is willing to share them anymore.

Yes, it is possible to predict the branch with very few markers when samples form a very close STR clusters. Then you need to SNP test only one and the rest are clear. This happens with Albanians and even Serbs, but Bulgarians usually don't form clusters and don't have any close matches elsewhere.

[Zhivet]

@Slavomir, no, Nevgen is used and you do not need 67 markers. Were compared literally hundreds of R1a haplotypes with terminsl SNP & 111 vs 23 vs 17 markers. Prediction is very accurate and consistent. In case of South Slavs and most of East Slavs you only need to ignore Scandinavian-British (Y2395>L448) and CTS4385(>L664) as it means Z280+. You need to understand context and how Nevgen functions.

[slavomir]

@Slavomir, no, Nevgen is used and you do not need 67 markers. Were compared literally hundreds of R1a haplotypes with terminsl SNP & 111 vs 23 vs 17 markers. Prediction is very accurate and consistent. In case of South Slavs and most of East Slavs you only need to ignore Scandinavian-British (Y2395>L448) and CTS4385(>L664) as it means Z280+. You need to understand context and how Nevgen functions.

in other words, it does give wrong predictions.
the Z93 is false as well, because in thousands of commercial testees it's super rare, while Nevgen finds a lot of it.

check out this SNP study as well:
https://www.nature.com/articles/ejhg201450

Bulgaria - 0.7% Z93
Macedonia - 0%

[slavomir]

Yes, it is possible to predict the branch with very few markers when samples form a very close STR clusters. Then you need to SNP test only one and the rest are clear. This happens with Albanians and even Serbs, but Bulgarians usually don't form clusters and don't have any close matches elsewhere.

I'm pretty sure some relatively deep haplogorups like M458>L1029, M458>L260, Z280>L1280, Z280>Y2902 can be predicted as well without those close matches.

They had names like "Carpathian cluster" and similar on those old forums.

[vasil]

8000 Serbs who did an actual DNA test, not a single one is Z93.

Wasn't there some STR tested R1a from the SE that could maybe be Z93 or am I having a brain fart?

[slavomir]

Wasn't there some STR tested R1a from the SE that could maybe be Z93 or am I having a brain fart?

There were some undetermined R1a who could be either Z284 or Z93, or some very basal branch. No certain ones.
There are two R2 in Kosovo (or SE, not sure) and Macedonia, maybe you mean that.

[vasil]

There were some undetermined R1a who could be either Z284 or Z93, or some very basal branch. No certain ones.
There are two R2 in Kosovo (or SE, not sure) and Macedonia, maybe you mean that.

I think I saw it on the new tested samples part of the forum, must have been from somewhere in the east if I made a mental mark of it. It's probably one of those undetermined ones. BTW R2 in Kosovo and Macedonia could be interesting there is one R2 sample from Moldova.

[Zhivet]

in other words, it does give wrong predictions.
the Z93 is false as well, because in thousands of commercial testees it's super rare, while Nevgen finds a lot of it.

check out this SNP study as well:
https://www.nature.com/articles/ejhg201450

Bulgaria - 0.7% Z93
Macedonia - 0%

Nevgen does not find a lot of it either, it literally predicts around 0-1% in Bulgaria, Romania and Macedonia. Given it is a STR prediction the SNP calibration is minimal, doesn't change anything relevant for the general population showing Nevgen overall is very accurate.

[eastara]

Nevgen does not find a lot of it either, it literally predicts around 0-1% in Bulgaria, Romania and Macedonia. Given it is a STR prediction the SNP calibration is minimal, doesn't change anything relevant for the general population showing Nevgen overall is very accurate.

Nevgen is accurate predicting the main branches at 37 markers like L1029, etc. Further than that it is right only if the sample falls in some known cluster, which you can see by his matches anyway. However I don't think the branches could be predicted with certainty on 12 and even 17 markers.
Nevgen is also helpless predicting Z93 branches, Sometimes honestly says "100% unknown subclade".

[eastara]

WE have 2 more samples in the Bulgarian project who are most likely Z93. Is there somebody willing to sponsor them?
MI57657 is predicted by Nevgen as 94.3% Unsupported subclade, but with the small percentage can't make its mind if it is R1a Z93>Z94> Z2123>Y934>Y7094 or R1a Z282>Z280> CTS3402>Y2613
IN87435 on the other hand is given confidently as 99% R1a Z93>Z94> Z2122>Y57. This usually means he has a close match, but on FTDNA he has inly 1 at 25 markers with Arabic name and no SNP test, while at 12 marker he has a mixture of Z280 and the only Z93 is R-YP1532.
The B6024 111 STR markers are ready, but Nevgen still predicts 100% Unsupported Subclade.