Archiver > GENEALOGY-DNA > 2005-10 > 1128796870

From: "Ken Nordtvedt" <>
Subject: Re: [DNA] Another failure to get DYS463
Date: Sat, 8 Oct 2005 12:41:10 -0600
References: <> <>

I have been wondering lately about the conventional wisdom you express
below, because having found some SNPs which rather clearly sit within a
haplogroup of interest to me (I1c), I have been formulating how vigorously
to encourage a testing lab to invest in enough testing of the SNPs to find
if they are "private" or have a decent descendant population found today.

Let's say I have very good reason to believe the SNP lies within haplogroup
or clade X (this can be established by measuring the extended STR
haplotype). Haplogroup X has some population history P(g) in each
generation from g = 1 to today's g = G. So the SNP occured in one of those
G generations. What's the probability of that generation being an early,
middle, or recent generation? It seems to me it is about the same for any

If it occured in generation g it had an equal chance to occur in each of the
P(g) members of that generation. On average each member of that generation
has left P(G)/P(g) descendants in today's population. So the number of
folks today carrying a SNP which occured in generation g is the mutation
rate times P(g) times P(G)/P(g) with the reference to generation g and its
P(g) dropping out of the product in the end. So any particular SNP we see
today seems about equally likely to have been created in each of the
generations into the past as far back as we can limit things by its known
nesting within a clade with finite age as determined by other means such as
SNPs or STRs.

Or let's restate this conclusion in terms of the present-day search process.
If a new SNP is found by a process whose success was about proportional to
its frequency of occurence in today's population, then its age
distribution is uniformally distributed by whatever nesting property we can
establish for this SNP within the tree.

True, there are many more chances for a SNP mutation to first occur in more
recent generations, but more recent mutations have a smaller multiplication
factor for their descendant populations in today's population, and these two
factors cancel each other out on average.

The population which provides the dna material being searched for the SNPs
must be brought into the picture. For instance, a European population will
not harbor very downstream SNPs in yhaplogroup C3; their downstream SnPs
will be found within the haplogroups common for the source population, etc.

This was not an academic response to John's comment. I am genuinely
interested in discussion from others who might see this differently,
because we have a growing inventory of supposed or suspected "private" SNPs,
and there is a question of how much resources should be invested in
thoroughly checking out how private they are, and how to structure the
arguments to the testing companies about why they should invest in
developing capabilities to measure these SNPs for us commercially.


Also note that
> most SNPs are "private".
> John Chandler

This thread: