GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2005-03 > 1110418498


From: (John Chandler)
Subject: Re: [DNA] DNAPrint, Native American
Date: Wed, 9 Mar 2005 20:34:58 -0500 (EST)
References: <20050309232833.26990.qmail@web41509.mail.yahoo.com>
In-Reply-To: <20050309232833.26990.qmail@web41509.mail.yahoo.com> (messagefrom Crystal on Wed, 9 Mar 2005 15:28:33 -0800 (PST))


Crystal wrote:
> I was just curious if say one has 14% NA .. would the
> MLE's be the same for everyone that has 14%? Or (what
> I'm thinking) is everyones different based on the
> their "criteria" meaning if marker 1 is found in 99%
> of NA and only 1% of IE then that would tighten the
> probability contours?

Not much. The contours are tightened by every marker's contribution
to the analysis. If your results showed several markers dropping out
(i.e., marked as FL), then your contours would be a little wider than
they should be, but only a little. If your question is, "Does
everyone get a test result exactly equal to the true ethnicity?"
then the answer is emphatically NO. You have a roughly 1/100 chance
of getting exactly the right answer, to the nearest percent.

> My triangle chart is
> http://www.geocities.com/mywebtree/abydna/plot.jpg

Ok, it's clear from your plot that the MLE is actually off scale. In
other words, the best guess mathematically at your Asian fraction is
slightly negative. But don't worry about that because it's all within
the statistical uncertainty.

John Chandler

This thread: