PDA

View Full Version : Nevgen prediction from STRs



rwtodd
04-25-2019, 06:49 PM
We have an unknown 2xgreat grandfather in our family tree. We tested my cousin, since he carries the paternal line from that unknown ancestor and we need every clue possible.

I entered my cousin's 111 marker results from FTDNA into the Nevgen calculator. It returned, with 100% probability, this: R1b=>U106=>Z381=>Z301=>L48=>BY12508.

Two questions come to mind. Is this path to be trusted, especially given 100% probability? And, is it at 100 % because R1b is so extensively sampled that the relationships between STRs and SNPs are well-established?

GoldenHind
04-25-2019, 07:43 PM
I don't know about the 100%, but I would at least say it is highly probable, but needs verification by actual testing.

R1b and U106 and L48 are all very numerous in the FTDNA database, and you are correct that the connection between R1b STRs and SNPs is well established. But since all STRs are subject to random mutation, a surprise result is always possible and one can never be 100% certain without confirmation by actual testing.

swid
04-25-2019, 07:58 PM
I don't remember offhand where below U106 it begins, but much of U106 has very distinctive STR marker combinations, so I'm inclined to agree with the Nevgen prediction, and that your cousin is likely L48+.

Also, does your cousin have 111- or 67-marker matches? Any matches with SNP results will have their furthest-tested result listed.

TigerMW
06-10-2019, 05:12 PM
... but needs verification by actual testing.

... since all STRs are subject to random mutation, a surprise result is always possible and one can never be 100% certain without confirmation by actual testing.
Agreed, it is important to confirm a predicted subclade with SNP testing.

Another aspect of this is the age of the subclade predicted as well as the subclade itself. Some subclades don't have very distinguished STR signature patterns and are therefore hard to predict. Subclades that are mid-aged, or 1200 to 2500 years old are often the easiest to predict. Very young subclades, or those in the genealogical timeframe, are much less distinguished than their peers so surprises happen.

Wing Genealogist
06-11-2019, 11:43 AM
The L48 clade has a subgroup which has a characteristic STR mutation of DYS636=11. As far as I am aware, no SNP has been discovered (to date) to correspond with this STR mutation, but there are a small cluster of SNP/clades below this "636=11" including BY12508. It is possible the Nevgen predictor is using this STR result (in combination with others) to predict its placement.

I would not make a "100%" assignment based on STR markers. But it is likely your cousin falls under the "636=11" cluster below L48.