GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2010-11 > 1288941813


From: "Clay Daniels" <>
Subject: Re: [DNA] FTDNA haplogroup prediction more sophisticated thanbefore?
Date: Fri, 5 Nov 2010 02:23:33 -0500
References: <0df101cb7c2e$6544ab10$2fce0130$@org>


I have wondered about this for some time, and disappointed at times in
FTDNA's hapglgroup predictions. This thread is a welcome relief.

I've always guessed that they just looked at the first twelve markers, but
it does make sense that they might pick DYS448 and DYS438 as a check values.

The main haplogroup missed in FTDNA's prediction scheme is I or any of its
subclades. Several folks have pointed out that haplo I is easy, but they
always refer to markers beyond the first 12.

I have my own reservations about haplo prediction from STR's. As an
R1b1b2a1b5 (at this point), I'm a big fan of SNP testing, but realize it's
only necessary for one member of a particular family group. I also have some
reservations about the SNP "backbone" test. Marleen Van Horne has ordered it
for all our un-predicted members of the White project, but Kevin Daniel
seems to ignore it with our Daniel project where we still have some haplo
"-" results.

Trouble with "backbone" tests is that they only tell part of the story. Same
can be said for the full deep clade SNP tests if those taking the tests
don't realize we (thanks to Thomas Krahn) are still in the middle of a big
explosion of new SNP's. For instance, I'm L21+ and Thomas K. has recently
added two new sub-SNP tests under L21 in the new draft tree, L302 & L380,
that are not available yet for testing.

Then there's the issue of "private" SNP's which I really don't understand.
Nora ( ) mentions L319 below L21 but I can't see it
on the draft tree.

Guess we just need to keep digging,

Clay Daniels
----- Original Message -----
From: "Lawrence Mayka" <>
To: <>
Sent: Thursday, November 04, 2010 9:41 AM
Subject: [DNA] FTDNA haplogroup prediction more sophisticated than before?


> Many years ago, FTDNA explained its haplogroup prediction algorithm in an
> issue of Facts & Genes. Since then, the algorithm has been tweaked, but I
> presumed that it still:
>
> - Looked at only the first 12 markers, perhaps because 12-marker tests are
> still available
>
> - Ignored geographical and ethnic information, perhaps because
> undocumented
> adoptions (NPEs) are always a possibility
>
> - Did not weight one marker over another, perhaps because such weighting
> is
> itself potentially haplogroup-dependent.
>
> Months ago, however, someone pointed out to me that the same 12-marker
> haplotype occasionally produced a slightly different prediction (e.g., R1a
> vs. R1a1). I verified that, but presumed that we had merely caught the
> database in the middle of a minor algorithm tweak.
>
> However, it is now obvious that FTDNA is doing _something_ more
> sophisticated. Consider this actual 12-marker haplotype:
>
> 12 22 14 10 13-14 11 15 11 13 11 29
>
> This individual's Y-DNA Haplotree/My Matches page shows no exact matches
> (with a confirmed haplogroup); but at 1 step away are:
>
> I1 : 5
> J2 : 6
> J2ab1 : 3
>
> At 2 steps away are:
>
> I : 1
> I1 : 117
> J1e : 1
> J2 : 20
> J2a : 1
> J2a4b : 4
> J2a4b1 : 10
>
> As far as I know, until recently this case would definitely have merited a
> SNP Assurance test. But instead, FTDNA's haplogroup prediction is J2.
>
> The individual's next two marker panels (markers 13-37) make clear that
> the
> prediction is correct, but the question is: How did FTDNA arrive at that?
> Is FTDNA now using more than 12 markers for its haplogroup predictions?
> Or
> is FTDNA giving extra weight to DYS388? Or has the algorithm been altered
> in some other way?
>
>
> -------------------------------
> To unsubscribe from the list, please send an email to
> with the word 'unsubscribe' without the
> quotes in the subject and the body of the message


This thread: