A yDNA Haplogroup predictor

[Donald Cooley]

Whit Athey has developed a yDNA haplogroup predictor.
It is at http://home.comcast.net/~whitathey/hapest.htm .

It is real time - enter your DYS numbers and wait for the predictor to amend its prediction before entering the next number.

As a test, I entered my 12 DYS values (you can enter up to 37). I got these results (Scale of 1 to 100):

R1a = 66
N = 23
R1b = 12
I1b = 10
G = 10
J2 = 6
E3b = 3
I1c = 2
I1a = 1
E3a = 1

Note that values above 50 indicate a good fit with the haplogroup. Values between 20 and 50 indicate a fair fit. Values below 20 indicate an unlikely membership in that haplogroup.

The Genographic Project said I'm in haplogroup H1a, so it works for me.

I'd be interested if this works for other folks, plus any other comments you may have. Anyone know how it works?

[sterlingnotes]

It works...

I entered all 37 markers... and it correctly predicted my haplogroup of I1c with a value of 91.

[rroman]

I think it also worked for me in the sense that none of the values was 50 or above. I am Haplogroup K, and this group is not listed. Do you know why this group is not even listed?

[cobra]

Same thing with me. I do not fit in the model:

E3a: 2
E3b: 9
G: 6
I1a: 1
I1b: 2
I1c: 1
J2: 6
N: 3
R1a: 2
R1b 1

My results said I belong to undifferentiated "R" (there are a few belonging to the group from which R1a and R1b came from, but who do not have the markers of any of the 2 "daughter" groups). But I ordered an SNP test to be sure. My guess would be that only the haplogroups that are the most widespread in the British Isles and the US were included in that particular predictor. Since a basic excel spreadsheet is being used, it would have been incredibly complex to have dozens of haplogroups.

[Donald Cooley]

Try this if you get a strange answer

Is it possible you entered one (or more) of the STR values wrong? I did that the first time and got a strange answer also. (Nothing above 50.) Also note that the Genographic's Project (GP) gives the STR value at one locus (DYS 389ii) in a way that is different than FTDNA. You must use the FTDNA value for DYS 389ii. To be more precise:

FTDNA's 389ii STR value = GP's 389i + 389ii
Plugging in my STR values 30 = 13 + 17

Hence for 389ii, I'd use "30" as reported by FTDNA and not the "17" reported by GP.

Hope this helps.

-Donald

[rroman]

Same results:

E3a:7
E3b:22
G:17
I1a: 12
I1b: 14
I1c: 35
J2: 20
N: 7
R1a: 6
R1b 17

[cobra]

I also entered the 389ii the way it should. I entered 31 (value of 389i + value of 389ii) and not 17.

[Donald Cooley]

New yDNA Haplogroup Predictor

Whit Athey has a new yDNA haplogroup predictor at https://home.comcast.net/~whitathey/predictorinstr.htm .

Many interesting details can be found there -- including technical details on how it works. Here's the first few sentences from Whit's website:

NEW (6/22/05): The Predictor page has cells for inputting the repeat values of any arbitrary number of STR markers (only the 37 markers tested by FTDNA may be input to the “Original Predictor”, but up to 64 markers can be used in Version 2.0). ...

Whit desires feedback, so if you have questions, or if you think his predictor gives an answer you think is wrong, let him know.

I found the predictor on the following website that has many great links worth checking out:

http://www.worldfamilies.net/y-haplogroups.htm

[rroman]

Still, the new version doesn't tell me anything about Haplogroup K. Why not?

[dentate]

My maternal uncle is a haplogroup P according to FTDNA's prediction based on STR--I have ordered a SNP test that is now a month overdue. Like K, this P is a very rare haplogroup and there is probably not enough information on its STR haplotype variations to include it in the predictor. Notice also that J2 is listed on the predictor, but not J (or J1) even though these are common haplogroups. That is a little harder to understand, but it may be because there is enough overlap in STR haplotypes that the predictor cannot differentiate between them (I am a J2 but about half of my matches based on STR haplotype are J, and therefore not closely related to me--a concept that can be very confusing at first). STR haplotypes are really most useful for demonstrating family relationships. They have interesting correlations with haplogroups, but really are less appropriate than true SNP mutation testing for looking at deep ancestry. I am hopeful that the Genographic project will bring more information on all of this.

[wwgeorge]

Could SKS lookup the following STR values in Whit's Haplogroup Predictor.
13,24,14,11,11,14,12,12,12,14,11,30
Looks like R1b to me. Thanks in advance.

[Bill Hurst]

Could SKS lookup the following STR values in Whit's Haplogroup Predictor.
13,24,14,11,11,14,12,12,12,14,11,30
Looks like R1b to me. Thanks in advance.

It gives a 47 for R1a, 26 for R1b. Try it yourself at:

https://home.comcast.net/~whitathey/hapest/hapest.htm

Bill H.

[wwgeorge]

Thanks Bill H,
My WebTv browser doesn't work on his Haplogroup Predictor.