GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2001-07 > 0994603462


From: Richard Gethmann <>
Subject: Re: [DNA] Projects and Generations
Date: Sun, 08 Jul 2001 10:44:22 -0400
References: <2449F3F867FBD211936600C04F68569EA3C3BD@smhrimsx.mhri.med.umich.edu>


Gregg,

I understand what you are saying below, a parsimonious tree is simply the tree
that requires the fewest mutational changes. In the tree as it is currently
constructed, there are four changes. By changing the tree slightly so that M25
and M09 have a common ancestor reduces the number of changes to three. That is
the "best fit" tree in that it has the fewest changes, but it does change the
genealogy lineages in some fashion for M25 and M09.

As a geneticist, I am bothered by the tree as currently constructed because it
requires two independent mutations to the same purple mutation.

There is another way of constructing the tree that allows for a single mutation
to purple, still keeps the number of mutations at four, doesn't change the
genealogy, and better illustrates what Gregg was talking about with respect to
reversions.

If we assume that the fourth son in generation 2 (the ancestor of the two
purple lines) was the first son with the purple mutation, then initially, all
three of his sons would have been purple. Somewhere, in generation 3, 4 or 5,
in the line leading to M15 and M14, a reversion occurred restoring the line to
yellow.

This changes neither the genealogy nor the total number of mutations, but it
does eliminates the problem of having two identical mutations.

Dick



"Bonner, Gregg" wrote:

> Sorry for the confusion - let me try it again.
> For the following I will use the 2nd diagram (the one with all the ovals
> colored in) from here:
> http://www.mumma.org/DNA.htm
> A parsimony tree is a tree which is constructed in such a way as to minimize
> the number of mutations required to explain all the data. People generally
> like parsimony trees based on the argument that mutations are rare, and
> hence you will want to minimize the number of them that you use to make the
> tree. In this figure the mutations are seen as a change in color. For the
> moment, I would like to forget all the Mumma genealogy which is known by
> virtue of census, wills, etc., and pretend that the diagram shown is based
> solely on DNA evidence. With the slight change of grouping together samples
> M09 and M25 with a common purple ancestor, we have a parsimony tree, which
> can be explained in only three mutations.
> 1) From the ancestral "yellow" state to purple to get to M09 and M25,
> 2) The change from yellow to blue to get to M14, and
> 3) The change from yellow to green for M06, M11, and M16.
> However, lets presume that M29 was really from that green line, but that
> M29's dad changed that part of the line back to yellow (i.e., he 'reverted'
> the line back to the original type). One way to test this is to test other
> descendants of the line further up the tree. Notice that M21 and M18 don't
> need to worry about this as much. If the father of M18 and M21 had reverted
> back to yellow from a long line of green, then you would expect to see M31
> as also being green. So M31 helps show that M18 and M21 were not the product
> of a hidden reversion. M29 does not have help like this. So if another
> volunteer were found that descended from this (presumed) green line in
> between the yellow parts (which would lead to M29), then he would become
> erroneously placed in the same group as M06/M11/M16. So what is best is to
> have a test subject from both sides of a branch point for every generation
> to rule out this possibility. What do I mean by branch point? In the chart,
> see Leonard Mumma at the top. Samples were found representing more than one
> branch (in this generation, from each of four sons) - that is branch point
> #1. Then from the son on the left, he had three sons represented (branch
> point #2). Et cetera to samples M18 and M21, totaling 5 branch points. Note
> sample M29 has no further branch points after the initial one. Note also
> that along the way to M18 and M21, the branch points are not in every
> generation. The branching origin takes place in generations 1, 2, 3, 6, and
> 8. So the greatest number of consecutive branch points is 3 (generations 1,
> 2, and 3). So this was my original question, "Which project shows the
> greatest number of consecutive branch points represented on more than one
> side of branch?" I wanted to know this so I could look and see how strong
> the case is against hidden reversion mutations. In the Mumma diagram, I see
> about 90 transmission events represented (90 father-to-son links), 3 of
> which (actually 4 if you don't group M09 and M25 together with a common
> purple ancestor) indicate a change, or mutation. If 3 or 4 of 90 are mutant,
> then it seems to me that 4 or 5 would not be impossible as well.
> Here I use family tree and phylogenetic tree equivalently, and I explained
> parsimony tree. By two samples per generation I mean representative of both
> sides of a branch point. Here M18 and M21 are the 2 samples of generation 9
> for that line. The line that is M18/M21 on one side, and M31 on the other,
> are the 2 sample sets of generation 7 for that line. Unfortunately, there is
> only one sample per generation per line in generation 8. I think it really
> unlikely for that line to mutate in generation 7, then revert in generation
> 8. However, if M29 really belongs with the green group (for example only),
> then one mutation in any of those 6 or so generations could revert it.
>
> The green line of M06/M11/M16 differs from the yellow line by a single
> repeat number (10 vs. 11 at DYS391). This is why I wanted to know what was
> the pressure to maintain any specific number. I could imagine how DYS391
> might not really care much whether it is 10 or 11, but does very much not
> want to be 9, for example. So it goes along merrily, changing from 10 to 11
> and back at a rate much greater than the rate of mutation in general. So
> then all those from the 11-repeat line would think they are a different line
> than the 10-repeat line. I could determine this in fact if someone had a
> line with a large number of consecutive branch points represented.
> I would call a person without kids just a dead end. And the main line I
> think I call just any specific line from a sampled person to the common
> ancestor. Or else the shared line between samples with a more recent common
> ancestor. By consecutive criteria, I just mean to say a tree with
> consecutive branch points with representatives of more than one side. If I
> were only interested in those samples that have consecutive branch point
> starting with the sample provider and going up, then I would myself be out
> of luck. My dad had no brothers. So my brother and I would make one branch
> point, but we could not have a second consecutive branch point because my
> dad would be the only one for that generation for that line. And I disagree
> with what was said about mtDNA. For me, it works in principle just like Y
> (for my genealogy purposes). The difference being only that the surname
> changes every generation, so it is harder to find people to recruit for my
> effort.
> Did I just make it worse?
>
> Cheers,
>
> Gregg
>
> ==============================
> Visit Ancestry.com for a FREE 14-Day Trial and enjoy access to the #1
> Source for Family History Online. Go to:
> http://www.ancestry.com/subscribe/subscribetrial1y.asp?sourcecode=F11HB

--
****************************************
RC Gethmann

****************************************
When you have eliminated the impossible,
whatever remains, however improbable,
must be the truth
A. Conan Doyle


This thread: