GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2006-12 > 1165008734


From: (John Chandler)
Subject: Re: [DNA] Question on testing for y haplogroups
Date: Fri, 1 Dec 2006 16:32:14 -0500 (EST)
References: <000601c6ebaf$220e1430$1bc9e804@DW><004801c710ef$e5e5a890$650fa8c0@Villandra2><008301c71549$02a2d450$640fa8c0@Villandra2>
In-Reply-To: <008301c71549$02a2d450$640fa8c0@Villandra2>


Dora wrote:
> I thought that STR's were used in testing because they are relatively easy
> to test for.

No, the reason is that they are very informative for genealogical
purposes.

> I thought that single nucleotide substitutions were NOT used
> because, especially on the long Y chromosome, it is difficult, time
> consuming and expensive to test for them.

No, the reason is that they are essentially useless for genealogy,
compared to STRs. Consequently, the testing volume is much lower in
the genetic genealogy market, and some of the economies of scale
available for STRs are missing for SNPs, and *that* is why SNP testing
appears to be more time consuming and expensive.

> Further, I had gotten the idea
> that the numbers for specific Y DNA markers that I see all the time are
> specifically counts of STR's.

Yes.

> But in one of the articles in the Journal of Genetic Genealogy, it appears
> to say that actually, single nucleotide polymorphisms - single letter
> substitutions in the genetic code - are used, but lately there is interest
> in using STR's to "predict" the Y haplogroup because that is easier to do.

Yes, SNPs are used for defining haplogroups, but see above. Indeed,
in some ways, the STR-predicted haplogroups are actually *preferable*
to the "real thing" determined from SNP testing -- if the lab makes an
error on a SNP test, the result is *completely* wrong because it's an
all-or-nothing test, whereas an error in measuring STRs will not
necessarily change the predicted haplogroup. The lower volume of SNP
testing probably means that the lab error rate is higher for them than
for STRs. We generally avoid thinking about lab errors, but they should
not be ignored.

John Chandler


This thread: