Archiver > GENEALOGY-DNA > 2010-02 > 1267148151

From: (John Chandler)
Subject: [DNA] Clerical mutations and lab errors
Date: Thu, 25 Feb 2010 20:35:51 -0500

I have a statistic that may interest some of you. In 2008, I visited
a large number of Y DNA project web pages and collected a substantial
number of 67-marker haplotypes -- about 10,000 in all -- in my effort
to get an improved set of relative mutation rates. Alas, I found that
number to be still too few. Recently, I began revisiting most of
those same projects to expand my collection, and, because it is
easiest to collect everything during a visit, I now have two snapshots
separated by almost two years. About 9,000 haplotypes appear in both
snapshots, and nearly all of them are the same both times -- but not
quite all. 25 have changed in the meantime, and 10 of those 25 are in
projects with pages that provide a direct window
into the FTDNA results database. Indeed, about 8,000 of the
haplotypes in play were found in such projects.

Note: the duplicate haplotypes found for testees who belong to
multiple projects have been weeded out. These are 8,000 distinct

Thus, I have a reading on the discovery rate for lab errors, including
clerical errors at the lab: 1 per 800 haplotypes. You can divide by
67 if you want the per-marker error rate, but I don't have enough
events to say how the error probability is distributed among the
various markers. It's surely not distributed evenly.

Note the much higher error rate (including clerical errors by project
webmasters) for the results found only on privately maintained web
sites: 1 per 67 haplotypes.

RECOMMENDATION: even though your project may have members who tested
at other labs besides FTDNA, go ahead and set up a "public" web site
at FTDNA as well.

John Chandler

This thread: