Re: Hamilton's rule


Perplexed in Peoria wrote:
> "aj" <aj@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message news:dikvcs$1h79$1@xxxxxxxxxxxxxxxxxxxxxx
> > I have a question about the 99% figure you use (see your post below).
> > Correct me if I'm wrong (I suppose I don't have to suggest this in
> > this, or any, newsgroup, but what the heck), but I understood this
> > figure to refer to sequence similarity, and not the percentage of
> > identical genes. Thus two individuals of a hypothetical haploid species
> > with 100 genes, each 100 base pairs long, could show 99% sequence
> > similarity (9900 identical base pairs of 10,000 total base pairs), and
> > yet have every gene be different (1 base pair difference in each of 100
> > genes). We know that point mutations can alter gene function, so it
> > seems reasonable to look at the probability of identity of genes rather
> > than of base pairs.
> >
> > Perhaps I've misapprehended the meaning of 99% similarity in this
> > genetic context. I'd like to know.
>
> I believe that you are correct, aj, in your guess that figures like 99%
> typically refer to percent identity at the base pair level. Or else
> percent synonymity at the codon level.
>
> But, IIRC, a recent press release on the comparison between human and
> chimp genome gave a figure of 95% similarity but then said that this
> meant that 5% of human genes (i.e. reading frames) have no recognizable
> homologs in the chimp genome. And further, that only about 25% of genes
> are identical at the protein sequence level - never mind the base pair
> sequence.
>
> Of course, none of this has anything to do with Hamilton's rule.

Agreed. The sequence similarity / proportion of identical genes
distinction is a red herring. Hamilton was fundamentally after an
explanation for why certain unselfish *phenotype* abound in the natural
world -- e.g. altruism. He started by constructing the usual one locus,
two allele models that are standard as a first grasp at population
genetical problems, and derived the results presented in 1963 and 1964.
But there was a more general principle at work, which he elucidated in
his 1970 paper. Say we have a three allele model, where two alleles are
different in sequence and yet code for exactly the same altruistic
phenotype, and the third allele codes for selfishness. In terms of
molecular evolution it may be interesting to follow the dynamics of
these three alleles. But in terms of the evolution of altruism -- which
is what Hamilton was, and I am, interested in -- we might as well count
the first two alleles as a single altruistic variant.

Hopefully this will help to illustrate why sequence similarity and
proportion of identical genes are besides the point. What is crucial is
that relatedness measures a statistical association between social
partners. If the only cause of this statistical association is
coancestry, and in the majority of cases assuming this is so will lead
us to a very good approximation, then we may phrase relatedness in
terms of probabilities of identity *by descent*. However, other causes
of statistical associations are possible, for example due to
environmental sorting of individuals with similar (or dissimilar)
phenotypes. Say there is a genetic basis to prefering red wine rather
than white wine. Then statistical associations at the wine-preference
loci will tend to emerge between individuals standing close together at
wine stores, because their environment has sorted them according to
their genetics. Hamilton was quite aware of this, but he realised that
this genetic association (relatedness) would not select for altruism in
the wine-store model, because it is the relatedness at the altruism
loci and not at the wine-preference loci that is of interest. Hamilton
did suggest the greenbeard example -- essentially, where the
wine-preference and the altruism are pleiotropic effects of the same
gene -- to illustrate selection for altruism not based on coancestry.


.

Quantcast