The major difficult arises from the fact that both positive selection and relaxed constraint produce a high ratio.

This may be true, but in the extreme case, where we are talking about a single allele (i.e. where nonsynonymous substitutions = zero), the diversity of synonymous substitutions in the population should be a direct measure of the allele's age (in the same population). Here we are NOT talking about ratios, but absolute numbers - this should help to distinguish between high positive selection (which would result in high ratios but low absolute numbers) and relaxed constraint (which would result in high ratios and high absolute numbers).

This is just an off-the-cuff observation, am I wrong?

Here we are NOT talking about ratios, but absolute numbers - this should help to distinguish between high positive selection (which would result in high ratios but low absolute numbers) and relaxed constraint (which would result in high ratios and high absolute numbers).

I'm not sure. why would selection give low absolute numbers, while relaxed constraint give high absolute numbers? also, note that absolute numbers are also influenced by the local mutation rate.

I'm not sure. why would selection give low absolute numbers, while relaxed constraint give high absolute numbers? also, note that absolute numbers are also influenced by the local mutation rate.

Since synonymous mutations are selection-neutral, given a large enough population, their number should be a very accurate measure of time - i.e. the probability of a synonymous mutation is constant, and will be preserved when it happens. It's like carbon-14 dating for DNA. If an allele is even moderately selected for, it will increase rapidly within the population. The result will be a small absolute number of synonymous substitutions (i.e. short time-depth, but wide distribution). In other words, high ratios are achieved by both the numerators and denominators being low numbers.

In the case of relaxed constraint, both synonymous and non-synonomous mutations are selection-neutral, so high ratios are achieved by both numerators and denominators being high numbers.

(The normal case, purifying selection results in low numerators and high denominators.)

He was definitely wrong about the field of economics being greener than biology. There have been only two or three advances in our understanding of how the economy works over the past century that have made a significant contribution to human welfare (taming the business cycle being the most important). You can't begin to count the number on the biology side of the equation, and the promise of future advances seems virtually unlimited.

Maybe he was imagining that a guy with his smarts could have risen to the very top of the economics profession by now, maybe even have won a "Nobel" prize, and not have to have slogged through the purgatory of being a post-doc. But in the long-run I suspect he would have felt far less fulfilled, maybe even sullied, by the experience, even if his academic salary would have been higher and he had penned a pop-science blockbuster like Freakanomics.

Having had some experience with Chinese people since I was a little tacker, I can say with some authority that they are just as capable of superstition as the rest of us ...

Personally, of course, I would bow down to Chinese women any day :-)

Another nice one, tks.

I wince a bit when Bruce says, "Our society, given its sordid history on race-related issues, is very confused about ..." You mean, as opposed to the brilliant and perfect way all other societies have dealt with race challenges? Seems to me that many tens of millions of people of many different races have come to this country and done awfully well for themselves. Whatever our gigantic and obvious imperfections, it seems to me that our batting average has to be above average.

But what the hell, still a good q&a.

Blowhard:

An excellent comment. I have always pointed this sort of thing out, not to deflect attention from the problems that still exist but to give PERSPECTIVE on the reality of the American situation. Because we make such a big deal out of any sort of problem, many of us grow up without realization that formerly and in other "societies" there existed no such freedom as has been created in America. This lends to his use of such words as "sordid" with regard to "race-related issues". By the same token (and ironically for those who dwell on imperfections and consistently bitch), this same criticism is a proof that this society is tops, if not in the top few, of all time. How do we react now that science more clearly and clearly demonstrates differences. It will middle the so-called right and left and show how similar the "beliefs" are, in a certain way, of both groups. I agree, a nice Q & A.

C

Off topic but I thought this may be of some interest to you:

Dr. J. P. Rushton is supposed to be on CNN's Paula Zahn Now at 8PM (EST) Tuesday according to a posting on the yahoo group evol-psych.

Since synonymous mutations are selection-neutral, given a large enough population, their number should be a very accurate measure of time

I think you're talking about sites that are polymorphic within a population. If we compare the chimp genome to the human genome, however, time is fixed (at about 5 million years) and the challenge to to sort through the sites that have become fixed.

synonymous mutations are selection-neutral

see this paper:
http://www.pnas.org/cgi/content/...ll/102/40/ 14338

I think you're talking about sites that are polymorphic within a population. If we compare the chimp genome to the human genome, however, time is fixed (at about 5 million years) and the challenge to to sort through the sites that have become fixed.

I'm talking about identifying whether an allele is under directional selection or relaxed constraint. If the allele is both widespread and has short time-depth, then obviously it is under directional selection. So, contrary to what I understand Lahn to be saying, it should be easily to distinguish between these two cases. Have I misunderstood him?

Rik,

From the link:

Surprisingly, from bacteria to mammals, the best indicator of a protein's relative evolutionary rate is the expression level of the encoding gene, measured in mRNA transcripts per cell (5, 6, 11-14). Highly expressed proteins evolve slowly, accounting for as much as 34% of rate variation in yeast (5).

The way they measure "evolutionary rate" is by looking at the synonymous variation. Isn't it the case that directional selection (i.e. non-synonymous variation) would decrease this? So maybe highly expressed proteins are evolving faster (in terms of functional evolution)? If expression is a proxy for importance, this might make sense.

Rik,

That was a very interesting link. The entire discussion was informative. I particularly liked the author’s analysis of “translational robustness” and fitness.

I wonder how much the paper’s results depend on the high selection pressure in yeast. With less selection pressure, have humans been losing “translational robustness” over hundreds of millions of years?


David: “The way they measure "evolutionary rate" is by looking at the synonymous variation.”

From what I read, they measured “evolutionary rate” by comparing amino acid substitutions in many proteins in a yeast clade. So the main focus was on “dN”, non-synchronous changes.

They did discuss “dS”, synchronous changes, to show evidence against the “Translational Efficiency” hypothesis.

If the allele is both widespread and has short time-depth, then obviously it is under directional selection.

this is essentially the test used by lahn to show selection in MCH1 and ASPM, except that he used haplotype length as a proxy for time.

the nonsynonymous/synonymous substitution ratio is not comparing alleles, it's comparing genes between species. that is, you're comparing two genome sequences-- the time of divergence between the two is fixed, and the sequences are representative of their respective species.

the nonsynonymous/synonymous substitution ratio is not comparing alleles

Why not?

Why not?

if you're comparing two genome sequences--say rat and mouse--you don't have polymorphism data, just a single sequence from each species. if there are different alleles of the gene segregating in mice, you can't tell.

if you're comparing two genome sequences--say rat and mouse--you don't have polymorphism data, just a single sequence from each species. if there are different alleles of the gene segregating in mice, you can't tell.

I mean, why not use this kind of analysis with allele data?

why not use this kind of analysis with allele data?

I think we're talking on different time scales here. if we want to know what makes humans different from mice, one would imagine that the differences are fixed in both populations, i.e. there is no polymorphism in either species at the relevant loci.

the signature of selection in polymorphism data only lasts for something like Ne (effective population size) generations, so it doesn't go back far enough to answer questions like the one above. for more recent selection, you use polymorphism data, but LD-based methods are more practical (and I think more powerful) than those based on the frequency spectrum).

for more recent selection, you use polymorphism data, but LD-based methods are more practical (and I think more powerful) than those based on the frequency spectrum).

Got it, thanks.