Quantifying genetic commonality
l.cook at man.ac.uk
Wed Feb 28 12:50:15 EST 2001
It is worth continuing this theme, because there seem to have been a number of
misleading conclusions in the press and elsewhere.
There is abundant evidence from morphological/metabolic/biochemical evidence (i.e.
the classical material of genetics) that human beings are pretty variable. This
evidence does not go away because sequences apparently look similar. In the light
of such evidence, how similar should the the sequences look?
>From enzyme surveys the mean heterozygosity, H, for humans appears to be about
0.125 (Nevo et al. 1984), which is high for mammals. If d is the probability of a
base difference between 2 individuals within one of the relevant genes, the
relation of H to d is roughly H = 1 - exp(-x.m.d),
where x is the number of nucleotides in the gene producing AA changes (about 850:
Li & Sadler, 1981) and m is the chance that an AA change will produce a scorable
difference (about 0.3 for electrophoresis).
Given this, d comes out at about 0.05. Some recorded figures are about this size,
some lower (but not by orders of magnitude), so the 90-odd percent base similarity
is by no means inconsistent with substantial classical genetic variability.
lcook at fs1.scg.man.ac.uk
The Manchester Muesum
University of Manchester
Manchester M13 9PL U.K.
Joe Felsenstein wrote:
> In article <96umk4$sfo$1 at mercury.hgmp.mrc.ac.uk>,
> Julian Assange <proff at iq.org> wrote:
> >Nancy Bennett <nancyb at ignet.com> writes:
> >> My question is -- can anybody even wildly estimate the degree to which any
> >> two randomly chosen humans (or let's make it easier and say any two randomly
> >> chosen Europeans) are, in fact, genetically similar? I recently read on the
> >> Human Genome Project web site that humans may have as few as "one gene in
> >> 500" that differs between individuals. This would suggest that, on average,
> >> the genes at 99.8% of all loci are identical. Is this an accurate
> >> estimate??
> >Yes, depending on what you mean by `gene' and `differ'. But it rarely
> >makes sense to talk about this data in absolute terms. You may have 19/20
> >positions correct for the state lottery, yet this does not mean you
> >have won 19/20ths of the lottery.
> I disagree.
> I don't think these figures are accurate. The usual figure for
> the degree of difference between the DNA sequence of different
> individuals (let's be more precise -- between haploid genomes such as
> the sequences of two gametes) is 1 in 1000 bases (sometimes this
> is given as 1 in 500, it varies a bit). But note that is bases, not
> genes. Genes are not 1 base long.
> Two (haploid) human genomes are about 99.8% similar, counting bases.
> They do not have 99.8% of all loci identical.
> Joe Felsenstein joe at genetics.washington.edu
> Dept. of Genetics, Univ. of Washington, Box 357360, Seattle, WA 98195-7360 USA
More information about the Mol-evol