Graham Dellaire (popa0206 at PO-Box.McGill.CA) wrote:
: > ez005139 at dino.ucdavis.edu (Daniel Mcgoldrick) writes:
: > Is there any known relationship between the structure of a repeated
: > DNA element such as a VNTR or microsatellite and it's mutation rate?
: >> and anything else I can score off of a sequence.
: SNIP
: > I'm trying to construct a model that will give me an estimate
: > of the usefullness of a repeat structure (as seen on a sequencing gel)
: > for a certain range of divergence time that spans as many as five log
: > units. Ideally, it would be cool to just look at the sequence and calculate
: > "that one is informative for e.g. between 1 and 10 thousand years because
: > it has fifteen repeats of 2bp that are not interrupted...."
: >
: > Daniel J McGoldrick
: > UC Davis Bodega Marine Laboratory
: >
: Dan,
: First I think you should remember that even though you may think a sequence
: may have some predictive/informative value for mutation rates etc... you have to
: take into account position effects. Even if you have a sequence that is very repetitive
: whether it recombines/slips etc may depend on the chromosomal context it is in. Unfortunately
: the closer people look at DNA sequence they see that it is less and less predictive out of
: context. The whole debate on dating using molecular clocks is a little up in the air now.
: For example people are realizing hemoglobin in the mouse and man may be in different genomic contexts
: and therefore may not mutate at the same rate, on this basis you could not use accumulation or mutations
: in this locus from a common ancestral gene(s) to determine the time of divergence between the
: two species...
: G.
Interesting point. Sounds like I can only do this for organisms
where the map positions are known (maybe mice or humans?). Recent papers
seem to divide repeated structures into three classes including
microsatellites (1-2Bp), Simple Sequence repeats (3+bp), and larger VNTR's
and then see if they fit an infinite alleles model or stepwise model and
finally conclude that stepwise is not rejected. Multiple mutational
processes are suggested (e.g. slippage, unequal pairing followed by
recombination, and unequal sister chromatid exchange). One process might
make big jumps in allele sizes, another makes small jumps, and I still can't
figure out how slippage can make a larger allele than its ancestor.
Anyway, I don't see why a map parameter could not be thrown
into the model if it explains variance. The problem is, I think its
going to be very hard to compile data because the actual internal
structures (at the sequence level) are buried under the term
"microsatellite, STR etc.". Now I need to score "chromosomal context" too -
ouch!
As an optimist, though, I am operating under something like what
G.L. Stebbins once told me about clocks. He said, "We set our watches
forwards and back don't we - but still we have a clock".
Dan
