GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2009-07 > 1247881631


From: "Alister John Marsh" <>
Subject: Re: [DNA] For Age Estimates beware of tossing STRs away
Date: Sat, 18 Jul 2009 13:47:11 +1200
References: <516801.35250.qm@web86610.mail.ird.yahoo.com><4A610DC6.4040607@san.rr.com><004f01ca0740$f6b0be40$6400a8c0@Ken1>
In-Reply-To: <004f01ca0740$f6b0be40$6400a8c0@Ken1>


Ken,

You said...
>>>>>>>
People have a tendency to select the STRS which give them ages they are most

comfortable with or like for some reason.
The slow markers could actually be mis-measured, as they are based on fewer
observations.
<<<<<<<

This has been a thought in my mind also, I agree.

I know from my own DNA research that when you are looking for DNA evidence
to support a hypothesis, there is a tendency to want to selectively focus on
the supporting evidence, and conveniently discard non supporting evidence.

Although my notorious "intuition" (right "almost" 50% of the time predicting
a coin toss) is that archaeology suits older dates better, I think we have
to be cautious about ditching either "fast" or "slow" markers simply on the
basis of hunches.

I think that the out of Africa dates do seem an approximate archaeological
reference point which imply some of the younger dates from faster markers
for Haplotree branch points are improbable. But the underlying problem may
not be "fast" or "slow", but something else which more often or coincidently
manifests itself in some faster markers. The context of mutations is very
complex, and there are many variable factors superimposed in different ways.

Something I have wondered... if we took the 10 fastest mutating markers, and
calculated some of the older nodes (ie speculated to be around 70,000 years
plus in age) using "just one of these fast markers at a time", what might
that show? Perhaps it might reveal if one or two fast markers produced more
erratic results, or perhaps some fast markers might produce more consistent
results.

A single fast marker can have perhaps 100 times faster mutation rates than
some of the slowest markers. So using a single fast marker should.... if it
was mutating randomly.... have 10 times more power of resolution that using
10 of the very slowest markers together.

I would suggest that DYS449 has a fairly fast mutation rate, yet compared to
some other faster markers has a moderately even and wide spreading
distribution of allele values, moderately following a normal distribution
curve. Would it be possible to as a test, estimate a few of the nodes using
just this marker "by itself", to see if it performs more consistently than
all of the faster markers grouped together?

John.





This thread: