« Miss Universe 2004 | Gene Expression Front Page | The eternal recurrence of elite betrayal »
June 28, 2004

PTC, part II

For those who read my previous post on PTC taste/non-taste, and my ruminations on its implications for general gustatory preferences, I have read the full text (thanks to a gracious reader) of the recent paper which suggests that the PTC locus is being buffeted by "balancing selection" (you can download the paper here in PDF format). I think the authors have made a good case that internationally the PTC locus can be characterized as being under the influence of balancing selection, where heterozygotes have a some fitness advantage which results in the maintenance of high frequencies of both major alleles. I also agree with the authors that there is a strong likelihood that particular local populations might be more shaped by founder effects or positive selection for one of the alleles.

So why balancing selection? Basically, in balancing selection the frequencies of both alleles reach an equilibrium where the percentage of heterozygotes is maximized in a random mating population. The most powerful illustration of their argument is a graph which shows the "number of populations" as a function of the frequency of the "non-taster allele." The graph is shaped like an inverted "V," with a mode and mean in the region of the 50th percentile. As one shifts toward 0 or 100 percent, the number of populations displaying those frequencies drops.

One might note that some (many in fact) populations show frequencies shifted far from the central region of the graph. That might be suggestive of local positive selection for one of the alleles or random genetic drift. But note the most common trend is toward a cluster near the center of the graph. This is the pattern one would expect if one imagined that the frequencies have a tendency toward equilibriating where both are maintained within the population.

If you had positive selection for one of the alleles, it seems rather obvious that the frequencies would be heavily skewed in one direction (that is, most populations would show high frequencies of one of the alleles). On the other hand, what about a situation where there is no selection pressure. Well, in this circumstance, one would expect that the populations would cluster at either end of the graph, around 0 or 100 percent for the non-taster allele in this case.

In the event of random genetic drift being the exclusive driver of genetic change, the allele frequency moves in a "random walk" fashion up and down. If you had 100 populations where the allele frequency was initially at 50% (for simplicity), all these subpopulations would drift in a variety of directions as time progressed. But, if the frequency in a subpopulation hits 0 or 100 percent, you have reached "fixation," and the subpopulation would no longer change in frequency (we are assuming no mutation). As time progressed, naturally more and more subpopulations would reach fixation, and the 0 and 100 percent cluster would keep increasing, until all intermediate frequency populations were eliminated.

In other words, in a situation where random genetic drift is dominant, the graph should have been somewhat "U" shaped. In a case where positive selection was dominant, there should have been a strong skew toward one end of the graph. Instead, what you see is a mixed strategy in evidence in most populations.

The authors point out there do seem to be exceptions. Native Americans for example. But, they have good answers to the questions (founder effect and so forth), and already concede that they offer one (likely the major) answer, not the only one. I attempted to use PUBMED to see what the rates of Asian Indians of PTC non-taste were. There are several hints that this population has high rates of non-taste, the higher frequency of this phenotype among the Gypsies (in comparison to Eastern Europeans) is for example used to diagnose an exoganous South Asian origin for them. But, I found several high caste, likely vegetarian, populations which had high frequencies of tasters (far higher than the 40% quoted in the Tepper paper I commented on last year). Since the South Asian populations are generally assumed to have a lot of substructure, random genetic drift might have shifted several populations far off the "optimal" frequency. But, the 40% number no doubt comes from a summation of all the various populations. I do note that the authors of the current paper had 10 Pakistanis in their sample for the South Asian portion. This warrants further investigation, and I might actually email the head author, since he probably knows a lot more about the evidence for positive selection than I, as he must have combed the literature.

Addendum: Also, please note that the clarification of three possible phenotypes, with a middling taster group between the non-tasters and super-tasters, should be cautionary about the oversimplifications that go into making population models. And one must remember that even the authors of the above paper note that there are other alleles that exist at extremely low, but non-neglible, frequencies.

Posted by razib at 03:54 PM