Discrete continuity in genetics

In the post below on skin color within a multiracial family I made the point that genetics is inherited in a discrete fashion. In the post-genomic era, or even the post-DNA era, this seems intuitively clear. Our genetic sequence, our genome, is a string of precisely four base pairs, A, G, T and C. The genome is digital, not analog. Case closed, right?

Not really. One of the main reasons I wrote the post below is the consistent misconception that genetics is blending, that children are a mix of the essences of their parents. This captures the expectation, but the variance. A natural inference of this model is that variation is diminished over the generations as it is homogenized through a process of mixture. Because Charles Darwin held to a system of blending inheritance he had to come up with ingenious ways to perpetuate and replenish variation. R.A. Fisher saw that Mendelianism was a way out, that discrete inheritance preserved information and genetic variation from generation to generation in full, mitigating the need for high mutational rates to battle homogenization, or to conceive of artificial barriers to breeding between demes.

Though it is easy to assert that Mendelianism "naturally" leads to the perpetuation of extant variation, it was harder to come to this consensus. One of the early posts on this weblog dealt with the early 20th century battle between the Mendelians and Biometricians. A large number of the former conceived of themselves as rebels overthrowing the outmoded Darwinian model, while the Biometricians fancied themselves the heirs of Charles Darwin. Under the leadership of Karl Pearson the Biometricians held that the Mendelian model could not account for continuous variation in phenotype, such as the famous bell curve which describes the nature of traits such as height or intelligence. This was nonsense, as some early Mendelians saw, and Will Provine in his book The Origins of Theoretical Population Genetics seems to suggest that one of the main blocks was simply a conflict of personality. Though Karl Pearson was no Fisher, he was a genius in his own right, and the basic reality that discrete processes can approximate continuous ones should have been clear to him.

Get it? If not, click below the fold for some graphs which I believe elucidate the issue pretty clearly.

Imagine a gene, locus "A," which comes in two flavors, A and a. The latter is a loss of function, while the former codes for 1 bristle on a flie. The alleles on the locus are additive and independent, so that the genotype Aa is between the values of AA and aa.

i-8ca9a514f3a680e8dc873c2d262d6d8e-image001.jpg

One could interpret A as dominant if you frame it as a dyad, "Bristled" vs. "unbristled." But in any case, this is a standard Mendelian single locus trait. But what if we add a second locus, "B," with the alleles B and b which are functionally cognate with A and a? That is, B contributes 1 bristle unit, while b contributes 0. This is now a two locus biallelic trait, and since we assume additivity and independence, the range of bristles can go from 4 to 0, all the way from AABB to aabb.

i-460e858eb53ccdeeaafc611657fc0940-image004.jpg

The graph illustrates the combinations. But it tells you nothing of expected frequencies. Going back to our first example, we can imagine it as a Hardy-Weinberg Equilibrium system, where the frequency of allele A is 0.5 and a is 0.5. From p2 + 2pq + q2 we get:

AA = 0.25
Aa = 0.5
aa = 0.25

Or, to be clearer in terms of realized combinations,

AA = 0.25
Aa = 0.25
aA = 0.25
aa = 0.25

In other words, not all combinations are created equal, there are multiple ways to get a unit of 1 bristle, but only one way to get 2 or 0 is in the first case. The same applies to the two locus biallelic case, assume that p and q are 0.5 again, you can generate a binomial probability mass distribution like so:

i-bfdf5582ee198773313c767ff810c411-image002.jpg

2 bristles is the modal unit, dropping off to 4 and 0 at the tails. There are many combinations that will give you 2 bristle units, only 1 will give you 0 or 4. This is clearly a discrete stepwise distribution, but it more than a simple "dominant/recessive" dyad. Now, let's crank this baby up, lets scale up to a 50 locus biallelic. Again, assume (extremely artificial) that all loci have frequences of 0.5 for a 1 bristle functional allele and a 0 bristle loss of function allele:

i-85e4ccfb0c60f79157aa2bf11c316f4e-image006.jpg

That's actually not a smoothed out graph in excel, the rather continuous form is simply an artifact of the human eye's perception. And so it is continuous variation on polygenic traits: as the number of loci increase, and their contribution to the sum totality of a trait decrease in average effect, they simulate the Central Limit Theorm, and approach a normal distribution. I like using the binomial distribution not only for its simplicity, but also because it too can approach the normal distribution as the n approaches ∞.

We perceive genetic variation as continuous, but its fundamental nature is discrete. Similarly, our everyday life is governed by the continuous motion of Newtonian Mechanics which is so well approximated by the calculus, but at its basic core the physical universe consists of discrete packets.

Tags

More like this

Science isn't perfect, it often misses obvious truths. Consider the 2005 Nobel in medicine, awarded for the work of Barry Marshall and J. Robin Warren in establishing the connection between Helicobacter pylori and ulcers. After the fact you hear many stories of doctors who had stumbled onto the…
One issue that has cropped up in the comments a few times here is a conflation between quantitative & population genetics. Though people seem to think they're interchangeable terms, they're distinct fields. That's why population genetics text books have chapters devoted specifically to…
I've talked about "the breeder's equation," R = h2S, before. R = response S = selection differential h2 = narrow sense heritability For example, if you have a population where the mean phenotypic value is 100, and you select a subpopulation with a mean value of 125 to breed the next generation,…
The first chapter of Evolutionary Genetics: Concepts & Case Studies gives a quick sketch of the arc of the field that the book covers via exposition of topical and current issues. Michael R. Dietrich focuses on the series of controversies which serve as "hinges of history." I have addressed…

I like using the binomial distribution not only for its simplicity, but also because it too can approach the normal distribution as the n approaches â.

In this case, â â 30. So, the next time someone asks you to count to â you can do it in two shakes of a lambs tail. In this case, 1 shake = 15.

I just want to make sure that I'm getting it right. Let's assume that a black man and white woman marry each other (non of them are bi-racial). Their kids' skin will be for sure brown skinned (not by any chance white or black), since
WW BB
%100 WB

Am I right?

Their kids' skin will be for sure brown skinned (not by any chance white or black), since
WW BB
%100 WB

Am I right?

assuming no gene-gene interaction effects, developmental or environmental variance, etc. yes.

Hi Razib,

Thanks for the prompt reply. The reason for asking that question is that My husband and I are expecting a baby girl. My husband is black and I'm Egyptian with a very fair skin and brown hair and eyes. Since my background is not in Genetics field, I would be thankful if you explain a little bit more about "gene-gene interaction effects, developmental or environmental variance, etc." that you mentioned in your answer.

Thaks and good luck,
Sarah

Thanks. I'm looking forward for that...
Sarah

A friend of mine was discussing someone's ethnic make-up. Said that the person had been genetically tested and was 35% Native American, 25% Caucasian, and 40% Black. Is that possible? I couldn't wrap my brain around the 35% (she may have said 33%).

"Is that possible? I couldn't wrap my brain around the 35% (she may have said 33%)."

Sure, Why wouldn't it be?

Imagine that one her maternal grandparents is of 2/3 Indigenous-American ancestry, and that one of the said person's paternal grand parents is of 3/4 Indigenous ancestry. Assuming that neither of the other two grandparents carry any genes exclusive to the indigenous American phylogeny, than each of the said person's parents will inherit, theoretically, half of their parent's DNA. One of the person's parents(the maternal)will be 1/3, .333(33.3%)indigenous-American, while the other parent (the paternal)will be 3/8, .375 (37.5%).

The middle index of 37.5%--what our hypothetical said person would inherit--and 33.3% = 35.4, averaging off at 35%.

By Aaronetic (not verified) on 14 Jan 2009 #permalink