Archiver > GENEALOGY-DNA > 2010-03 > 1269473431

Subject: Re: [DNA] Comparing FGS and 23andMe mtDNA Tests - Results Don'tAgree
Date: Wed, 24 Mar 2010 19:32:15 EDT

In a message dated 3/24/2010 9:56:46 AM Pacific Daylight Time,

> Thank you for the response. I understand that you are saying that
> for the mtDNA SNPs for which I have the results that don't agree from
> the FTDNA FGS and 23andMe tests, that I should consider the FTDNA
> results to likely be more accurate. This is because of certain
> conditions that exist in "chip technology" (for instance "poly-C
> tracts") that make it more difficult to "call" the results for
> certain SNPs - is this correct?


> Are these "conditions," that you mention, limited to mtDNA testing,
> or do the same "conditions" exist in X, Y, and Autosomal tests done
> with "chip technology"? Would we expect the same error rate to be
> revealed if we were able to compare the "chip technology" tests in
> these other areas with "actual sequencing" tests?

The situation is pretty much limited to mtDNA. The mutation rate is much
higher and everything is packed very close together in the mtDNA molecule.

> What about the "multiple" tests results, that I mentioned, that I
> have in the 23andMe results that relate to a specific rCRS
> position? Are these actually SNP "repeats" or are these multiple
> tests that were done for the one SNP at the rCRS position?

I believe these are multiple tests, dangling slightly different forms of
"bait" to lure a specific mutation

> Also what about the 23andMe results, that I mentioned, at 23andMe
> positions 310, 311, and 16184 that don't have a corresponding rCRS
> position? Apparently 23andMe thinks that they measured SNPs at these
> locations - why don't they show in the FGS results?

Partly a poly-C problem and partly a problem of basing things on the Yoruba
Reference Sequence.

> Ann, from a Google search I did find an Excel file that you developed
> - "CRS_lookup_with_23andMe_SNPS.xls, 11/14/2008." I thought that
> this is a great tool, however, I experienced some problems trying to
> us it. Is there a newer version of this file?

No, I decided not to update it due to the sorts of anomalies you listed in
your initial message. These only cropped up last October when 23andMe began
to "call" more of the SNPs on the chip, but it appears some of them were not
very robust.

Ann Turner

This thread: