Does heterogeneity in opsin genes demand selection pressure

Mary K.Kuhner mkkuhner at kingman.genetics.washington.edu
Wed May 9 14:43:52 EST 2001


I think Hardy-Weinberg proportions are almost a complete red herring
in this discussion.

The basic question is:  can a population keep two alleles for a long
time even if there is no selection in their favor?

The answer to this depends on population size.  A tiny population
will rapidly lose one allele.  Imagine a group of four monkeys with
some R and some G alleles.  By chance all their children might get
R and not G, so that G is lost.  In a larger population the chance of
loss goes down, or equivalently the rate of loss goes down.  With
enough time one allele will always be lost unless the population is
infinitely big, or selection is happening, or mutation re-creates the
allele.

The average lifespan of an allele in a diploid population is proportional
to 2N, the number of copies of it, where N is the population size.
There are fewer copies of an X-linked gene, so its lifespan will be
proportional to 3/2 N.

If the apparent age of your allele is much greater than the predicted
lifespan of neutral alleles, it is probably not neutral.  One way to 
observe this would be to see the same alleles in several species
whose separation time is known.  Another would  be to look at
the accumulation of differences (besides the R/G difference) in
surrounding DNA.  Very old alleles will have lots of differences.

If the frequency of R and G is very equal this also suggests selection.
Neutral alleles usually have one rare and one common allele (this
is the most likely pattern by chance) whereas selection in favor
of RG heterozygotes would tend to keep freq(R)=freq(G).  This is
not proof, as neutral alleles can have equal frequencies by chance.

So, in order to decide whether or not the R/G polymorphism could
be neutral, you would want an estimate of N (long-term, not just
today) and an estimate of the age of your alleles.  You'd calculate
the expected lifespan of an allele, and compare it to the age of
your alleles.  A pop gen textbook should help here.

You'd also want to be fairly sure that R/G isn't being constantly
recreated by mutation, but since human R and G differ by something
like 7 critical mutations, if the apes are the same you could probably
feel confident of this.  If only 1 mutation differentiates the two
alleles you'd want to look deeper, into the actual sequences of
the two.

Mary Kuhner mkkuhner at genetics.washington.edu

---





More information about the Mol-evol mailing list