GENEALOGY-DNA-L Archives
Archiver > GENEALOGY-DNA > 2006-01 > 1136486940
From:
Subject: Re: [DNA] Y-STR Mutation rate
Date: Thu, 5 Jan 2006 13:49:00 EST
Ken ..
I think most of these efforts by population geneticists are useful,
informative & likely more accurate within the context of their particular analyses,
i.e. using their specific populations & data. There are many studies by Kimura
& followers predating the recent Ho, Howell, & Penny papers discussed in
the List in which reduction factors far less than 1/3 are estimated. Various
contexts: inter-species populations' autosomes, human mtDNA, human/chimp YSTR
& mtDNA, etc. I've seen 1/10 used as a rough estimate of a reduction ratio
for mother/offspring to population rates for control region mtDNA.
E.G., the recent Ho paper suggests we shorten the estimated time to the
Neanderthal/Human "split" & makes the point that though we don't know the
underlying causes, we can still attempt to measure the "decay" in mutation rates as
calibration age increases. Doesn't answer all our questions, but it's
helpful research. And perhaps this will help spark further research that will
yield more answers.
When you assume that growth rate of a haplotype is a function of the
haplotype, which seems very likely to me, your derivation shows dASD/dG always
negative, & we see an assumption that always retards the effective population
mutation rate from the f/s rate. AdditionaIly, I believe there are very likely
many specific ways in which your assumption is true, with computationally
different results. The specific population dynamics details: bottlenecks, founder
effect, differences in so-called “drift” due to mutational mechanisms
(underlying “mechanics” of which we’re ignorant or dimly aware & can’t measure) –
all these factors might possibly have a more specific influence on the
real-world retardation from f/s mutation rate of YSTRs.
So, in addition to knowing that dASD/dG is always negative, for specific
population comparisons with specific haplotypes, perhaps we can someday better
estimate the real-world retardation factor, and that factor might differ from
another “case” because the population dynamics differ & the mutational
characteristics of the specific haplotypes (in the time-frame of interest) might
differ.
Yes, dASD/dG is always negative, given your assumptions, and for a specific
situation, like that in the Zhivotovsky/Underhill paper comparing Bantu,
Gypsies, etc, *how* negative is it? What’s the actual derivative?
Finally, I see this view as likely realistic, v.s. pessimistic. :-) And
yes, any given population’s variance is a property of that population history,
perhaps similar to other populations, perhaps not. And continuing research
will hopefully yield ever more specific and realistic results, as we continue
to (hopefully) refine the hypotheses and theoretical consequences.
Best wishes …
Mike ...
===============================
Mike, If you are right in your pessimisim, then I'd say the "effective
mutation rate" is close to a useless parameter. It means that the age
estimates come from the other outside factors, for whatever they are worth.
Population variance then just tags along as a property of the unique history
of that population since its founder. What should we then conclude about
the community of population studies geneticists who to this day in their
publications keep churning out these assertions about absolute ages and
relative ages of populations based on the purported universal formulas?
Ken
In a message dated 1/5/2006 4:00:02 A.M. Pacific Standard Time,
writes:
Dear John C
I will try and answer your questions below, but by inference some
aspects of Mike and Ken's responses as well.
1) Yes I agree the mutation rate estimate for archaeology is an
effective rate not an absolute one. More correctly: it is an apparent
effective rate whose value depends on selection parameters including the
distribution of the magnitude of mutational effects on fitness,
population size and population variability parameters, and the mean and
variability of mutation rates amongst individuals. I have no problems
with that, these rates can still be used with appropriate caveats, and
as Ho says are probably rather stable after a million years or so, but
they do need to be underpinned to known dated events in the genealogical
record.
2) Purifying selection is not the same as random extinction it is still
stochastic but occurs more frequently than by chance alone. It can be
detected as distinct from that in many ways and leaves a very distinct
imprint on the genome. Similarly positive selection also leaves a
distinct and different imprint on the genome. One drives stability the
other drives change, both are a consequence of selection on the
individual of the species in a stable or changing environment. There is
another as well: counterbalancing selection, where diversity is selected
for and maintained. It may surprise many on the list to find that the
estimated age of the individual haplotypes in the autosomal human HLA
region are ~40 million years old and extend over considerable distances.
In fact these haplotypes are even older and predate the divergence of
humans from chimps. For instance see
http://www.genome.org/cgi/content/abstract/15/9/1250
3) Population size has a marked effect on whether a mutation is fixed or
lost especially at small population sizes. The usual consideration is
that: "The dynamics of mutations with selection coefficients much less
than the reciprocal of the effective population size [|s| <<1/2Ne)] are
expected to be entirely governed by random genetic drift. By increasing
the population size and allowing natural selection to occur, we expect
that mutations reaching fixation in our larger population-size lines
will have proportionately smaller average effects." What is often not
discussed is the actual proportion of mutations that have significant
effects, nor their distribution.
I don't think I made myself clear in previous posts, purifying selection
has a more marked impact at LARGER population sizes not smaller.
Appropriate adjustments above need to be made statement regards Y as
well as it is not autosomal.
4) Please note that the effective mutation rate estimates are all LOWER
than the absolute rate and decline with increasing time and then
stabilise according to the graphs of Ho et al. This is NOT what the
neutral model would predict, but it is consistent with purifying
selection.
See http://en.wikipedia.org/wiki/Neutral_theory_of_molecular_evolution
5) Regards my assertion that father son estimates of mutation rate could
only be used over a short time frame for direct lineage estimates, I
provided an extreme example of human, Neandertal and chimp to prove the
validity of the point. You stated:
"No. In the genealogical case, the individuals are all the same species,
and one can arguably assume the mutation rate is the same for all. In
the multi-species case, this assumption must be discarded."
I disagree on this point. There are various tests for this and with the
exception of a higher rate in the rodent/mouse lineage in mammals the
estimated effective rates are pretty constant across all mammalian
species. There is certainly no evidence I know of that they differ by
the amount required to explain the differences observed in this case,
which was nearly a 100 fold change.
6) Some time ago I posted a comment about individuals varying not in
their intrinsic mutation rate, but in their ability to repair mutations
and the consequences it may have. From memory one of these posts was
about whether SNPs and STRs would both vary in their "observed" mutation
rate as well. I recently came across a very elegant study in C Elegans
in which the gene I referred to MSH2 was knocked out. This elevated the
observed mutation rate by 100 fold for SNPs and 180 fold for STRs. They
then used this resource to study the effect of varying population sizes
on mutation accumulation over a number of generations and its impacts.
They clearly did not answer all questions but they did state:
"Our results indicate a large discrepancy between mutation rates
estimated from phenotypic and molecular analyses of the msh2 N = 1
lines. This is our most compelling evidence for a class of mutations
with small selective effects."
The link is
http://www.genetics.org/cgi/content/full/166/3/1269
My understanding is the senior author wrote a book with another
geneticist whose TMRCA equations are often mentioned on the list.
Anyway hope this adds to the debate about mutation rate, because the
process and its application underpin genetic genealogy as well as anthro
genealogy.
Cheers
John McEwan
-----Original Message-----
From: John Chandler [mailto:]
Sent: Thursday, 5 January 2006 10:02 a.m.
To:
Subject: Re: [DNA] Y-STR Mutation rate
John wrote:
> 1) You talk about "resetting" the molecular clock unpredictably due to
> selection and random extinction. Yes as I understand it this is what
> they are talking about, but it is not quite as bleak as you propose.
The
> first is most studies can calculate the "effective" mutation rates of
> various branches (they usually are dealing with a tree not a single
> lineage) and also for various genome locations. They can do what is
> called a relative rate test to check if the molecular clock ticks
faster
> in a branch or specific genome location...
Yes, they can do all this, but the result is just relative rates, not
absolute rates, and that's the whole problem in the first place.
> that. This can be due to positive or purifying selection
I have a question about that. From the description in Ho's paper and
elsewhere, I have concluded that the "purifying selection" is not
selection at all, but just a misnomer for random extinction. Is that
true? If not, how do you draw a distinction between "positive"
selection
on the one hand from "negative" selection on all other alleles on the
other?? This type of misnomer is very damaging because it invokes a
process that has very different consequences in the long run, precisely
where they hope to explain "puzzling" results. Random extinction causes
mutations to disappear within a population because each mutation starts
out as one individual, and static populations have a very high
probability that any given person's haplotype will die out. However,
in the long run, what happens is this: some mutations beat the odds and
take over instead of dying out -- these mutations then become the norm,
and the "purifying" then selects *for* them instead of against.
> 3) I don't agree with the conclusions of your point 4. As an example
> take the case of say neandertal's, chimps and humans: each has a sole
> direct lineage to each other, just the situation you outline for your
> genealogy comparisons but perhaps a LITTLE more extreme.
No. In the genealogical case, the individuals are all the same species,
and one can arguably assume the mutation rate is the same for all. In
the multi-species case, this assumption must be discarded. Sorry, but
that's life. The result is two new unknowns to be estimated and no
decent calibration for either one. We could eventually calibrate the
chimp rate, but the Neandertal rate is out of reach.
And it's worse than that. It is logically true that all chimps and all
humans must descend from one MRCA, but that ancestor could turn out to
be much further back than the estimated speciation event based on the
fossil record.
John Chandler
==============================
Jumpstart your genealogy with OneWorldTree. Search not only for
ancestors, but entire generations. Learn more:
http://www.ancestry.com/s13972/rd.ashx
==============================
Census images 1901, 1891, 1881 and 1871, plus so much more.
Ancestry.com's United Kingdom & Ireland Collection. Learn more:
http://www.ancestry.com/s13968/rd.ashx
=
Admin: FTDNA Humphrey project
Co-Admin: FTDNA Liles project
http://humphreygenealogy.com
multi-family project:Humphrey, Liles, Morton
This thread: