Figure 2.—The distribution of fitness effects of deleterious
mutations represented as either (a) a continuous or (b) a discrete
function. The dashed lines in a and the solid lines in b
represent the 95% credibility intervals. (a) A transformation
of the gamma distribution to a log-scale. Note also the difference
in the minimum values for a and b.
These graphs are taken from a paper written by Adam Eyre-Walker, Megan Woolfit, and Ted Phelps. ‘Nes’ means the effective population size (assumed to be about 10,000 in humans) multiplied by the percentage decrease in fitness associated with a particular kind of mutation. So a Nes of 100 correspond to a 1% decrease in fitness. The point is that mildly deleterious mutations, ones that reduce fitness by something like 1%, are considerably more common than ones that drastically reduce fitness. This makes sense, because most non-synonymous mutations, ones that change an amino acid in a protein, don’t cause a big decrease in fitness. A few do, as when a mutation turns an amino acid into a stop codon, truncating the protein.
Note that this describes the spectrum of new mutations. The distribution of existing deleterious mutations in a population is quite different. Dominant lethal mutations are not passed on, hence do not build up with time. The dominant lethals you see are all new, freshly generated by mutation. On the other hand, a mutation that reduces fitness by 1% is only slowly eliminated by purifying selection, so its frequency builds up with time. Its equilibrium frequency is 100 times higher than that of a dominant lethal that occurs equally often.
So… most genetic load in humans is made up of many, many mutations that each have fairly small effects. A smaller fraction of the genetic load consists of mutations with big effects on fitness.
I think that this picture has some interesting implications. Consider paternal-age effects: several problems seem to be more common in children whose fathers were older than typical. Sometimes this effect is dramatic, as with Apert syndrome, but that is a special case: premeiotic cells with this mutation, the precursors of sperm cells, seem to have a growth advantage. Something similar probably happens in achondroplasia, classic dwarfism. More generally, one expects that the mutation rate rises with paternal age due to an increasing number of cell divisions in the male germ line. There is evidence, not utterly conclusive but fairly strong, of increased rates of autism and schizophrenia with paternal age.
We can then conclude that increased rates of disease with paternal age are driven by highly deleterious mutations, rather than an increase in the number of slightly deleterious mutations. If the mutation rate in men of a certain age is 1.5 times higher than average, dominant lethals would go up by 50% in their children, while the average number of of 1%-deleterious mutations would increase by only 0.5% – a totally insignificant change. This because those slightly-deleterious mutations have accumulated over many generations: one generation can’t make much difference. Still, the number of slightly deleterious mutations does vary between individuals: the distribution should be Poisson, although with N large enough to closely resemble a Gaussian distribution. And this distribution might be modified by selection: people on the high end may suffer materially reduced fitness. Theory suggests that they should.
One important point is that a single highly deleterious mutation has a good chance of pushing the whole organism in some odd direction in phenotype space. In other words, the same mutation that drops your IQ, or damages your heart, may also make you look funny. At lower IQs, more and more kids are considered to suffer from ‘organic’ retardation. On the other hand, a higher-than-average number of small-effect mutations should also interfere with really complex systems such as the brain (and reduce IQ), but because of the law of large numbers, wouldn’t tend to have any particular direction in phenotype space. As far as I can tell, an extra-large dose of small-effect mutations, which we will henceforth call genetic noise, would not make you funny-looking.
Individuals can vary in the amount of genetic noise they carry, and populations can as well, depending on the relative intensity of selection and on the mutation rate, which might also differ. For example, although having an unusually old father does not much affect the amount of genetic noise an individual carries, a culture in which fathers were typically 55 would undoubtedly accumulate an unusually high amount of genetic noise, over a couple of millennia.
If a kid’s parents have a higher-than-average amount of genetic noise, on average the kid will as well. This sure looks like what we usually call non-organic or familial retardation.
Most of the within-population variation in IQ looks to be familial rather than organic. If I’m right, this means that most IQ variation – what we might call the normal range – is caused by differences in the number of slightly deleterious mutations. None of them would show up in a QTL search, because all are rare. And that is where we stand thus far: no intelligence QTLs have been found – although you never know what you’ll see in the next population. On the other hand, shared chromosomal segments would mostly contain the same slightly deleterious mutations, and so IQ should correlate with genetic similarity, which is what Visscher has found.
Many great scientists and mathematicians have likely had relatively low levels of genetic noise combined with some fairly deleterious de novo mutations; with the net effect of a powerful mental engine strangely focused on some particular topic not directly related to fitness. Low noise, high weirdness. Math, not sheilas. One might look for advanced paternal age in such cases.
We know that IQ is associated with a number of good outcomes – greater longevity, for example. Some of that, maybe most, may be people practicing good health habits, but having less genetic noise could help in a more direct way.