GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2005-03 > 1111765651


From: "Ken Nordtvedt" <>
Subject: Re: [DNA] Re: [Beg to Adv -DNA] ASD Was: How significant is a 35/36 match...
Date: Fri, 25 Mar 2005 08:47:31 -0700
References: <64.51b56a44.2f758634@aol.com>


Anne, I must read up on the manipulations that can be done on an excel
database. It could be that an ASD calculation could be configured within
excel? I don't know.

But I don't think that people should be rushing to calculate ASDs for pairs
of people or for small groups. It won't mean much. It could start to take
on meaning as a cluster of related haplotypes within a surname project gets
bigger and bigger. But even then, since such clusters tend to sample some
lines with a go-getting member who lines up his cousins and therefore
accumulates multiple haplotypes, while other lines may be just represented
by single entries, there is a sampling problem. So until some protocols can
be developed for objective sampling of the descendant population of some
common ancestor or founder who is being sought, I really question the
immediate application of an ASD to genealogical clusters. More work has to
be done.

I think a few people on the list have tried to apply ASD type concepts to
their surname and greater dna family haplotype clusters. Perhaps they might
comment on how they have handled the issue of fair sampling of the
underlying descendant population?

And yes, you never have to calculate the squared differences for markers
with no differences. But you probably have to include the mutation rate of
the markers which had no differences. Otherwise you would be selectively
and in a biased way throwing away those markers which just happened to
generate no differences in your cluster. Usually those markers are the
one's with slow mutation rates, so that would not change your results too
much, but it is best to develop habits of not doing biased probability
measures in the first place.

I really hope someone develops the good protocols for applying an ASD
calculation to genealogical clusters; that would be best for everyone. And
when they do, I hope they write a clear and complete explanation of why they
did what they did, unlike FTDNA which just throws out "tools" for people to
use with the bare minimum of technical justification. People can always
delete the explanations if they choose, but some people want to understand
why a "tool" they are being given is valid.

Ken
>
> Thank you for this clear example. I suppose what would really be
wonderful
> is if SKS with the requisite skill wrote a calculator for the
math-stricken
> such that one could put in marker values for a group of participants and
the
> calculation would be made. I suppose the first thing would be to have to
> enter a value for the number of participants you have -- to build an
array large
> enough for each use of the calculator -- as that would vary. And there
> wouldn't be any need to put in markers which don't diverge is there?
>
> Anne
>
>
> ==============================
> Jumpstart your genealogy with OneWorldTree. Search not only for
> ancestors, but entire generations. Learn more:
> http://www.ancestry.com/s13972/rd.ashx
>
>



This thread: