To: pike-dna-l@rootsweb.com
Date: Fri,  8 Dec 2006 15:55:13 -0330 (NST)
From: dapike@math.mun.ca (David Pike)
Subject: [PIKE-DNA] Some Recent News


Hi again.  In my last email I mentioned that there have been a
few new results that have come out lately, so I want now to
discuss a few of them in more detail.

First, anybody who might have been paying close attention to the
list of pending test results that is at the bottom of our "Results"
page might have noticed that we had markers 26-37 re-tested for
Bryan (kit 28606).  Why?  Well... because at 25 markers Bryan was
a close match to both Fred (kit N31746) and Drake (70672), but
at 37 markers Bryan's DNA showed several mutations.  Said another
way, the 25 marker data made it look like Bryan might be kin to Fred
and Drake, and given also that both Drake and Bryan have Pike
ancestry from the 1700s in Maine, it appeared suspicious that
Bryan's markers 26-37 were so far off when his first 25 were close.

The re-test (from the second vial that Bryan originally sent in)
is now complete and has come back exactly the same as the first test.
The conclusion is that Bryan is not of the same Pike line as
Drake and Fred.

We have seen examples previously in which low-resolution matches
fell apart when additional markers were considered.  For example,
Stuart and I are only 2 markers apart when looking at only 12
markers, but at 25 markers, our difference is 10.  Not until now
though have we witnessed in our project an example of a close
25-marker match that was shown to be a non-match at 37 markers.
In particular, comparing Drake and Bryan:  they match on 11 out of
the first 12 markers, 23 out of 25, but only 25 out of 37.

The statistical tool that FTDNA let's me use to come up with
probabilities of relationships said this of Bryan and Drake:
based on 12 markers there was a 46% chance that they were
related within the past 400 years, a 70% chance based on
25 markers, but only a 5% chance when all 37 markers are
considered.  I asked FTDNA to reflect on this... they suggested
thinking of things this way:  "his closest match gave a FTDNATip
score of 46% in the first 12. In other words, at that point we had a 54%
chance of them being related past 16 generations. The additional 12 markers
gave us the hope that perhaps they could be related 16 generations or less,
but there was still a 30% probability that they were been related past 16
generations. The additional 13 markers (to 37) made the initial assessment
prevail: they were not related in the last 16 generations."

One lesson that's to be learned here is that when 12 (or even 25) marker
comparisons yield uncertain results, then having the benefit of additional
information can help to provide clarification.  While writing on the
nature of 12, 25, and 37-marker results, let me convey one analogy that
I've heard used:  the first 12 markers can usually figure out which
treetrunk (in a big forest of family trees) is yours, 25 markers splits
you off onto one particular branch of the tree, and 37 markers moves you
onto a twig where your recent relatives would be found (where "recent"
refers to a genealogical time frame).

Okay, now let me say a few words about some of the other recent results,
the last two of which have matched up with our project's "Group 2" which
is for Pikes that hail from the Newfoundland town of Carbonear [who so
far have not genetically matched with Pikes from anywhere else].  For
those not familiar with Newfoundland, Pike is the 22nd most common surname
here, and many Newfoundland Pikes trace their ancestry back to Carbonear.
However, we almost all get genealogically stuck in the early 1800s, before
which there really aren't any good historical records.  Scattered early
references to Pikes indicate that we've been in Carbonear since at least
1681.  The result is that we have a plethora of Pike families trees, all
"starting" about 1800, but with no way to tell how any of them are connected
to one another.  But thanks to DNA we are now able to figure out which ones
at least belong together, even if we can't yet figure out how.

Among these latest results is one for kit 72548, for which we now have
37 markers available.  Comparing 72548's results with mine, there are
two differences.  One is a small difference on marker DYS-456, but the
other one looks interesting:  it's a difference of 3 on marker CDY(b),
where I have a value of 40 while 72548 seems to have a relatively rare
value of 43.  It may be too early to derive any meaning from this, but
it's something to consider as more results become available.  Incidentally,
the statistical tool that I mentioned earlier (when comparing Bryan
and Drake) says that 72548 and I have a 96.7% chance of being related
in the past 400 years, based on our 37-marker results.

- David.