Selection is often a mystery

If population genetics is “study of the distributions and changes of allele frequency in a population,” then the understanding of the maintenance of variation (or lack thereof) is one of the major topics of focus. In the first half of the 20th century when there was a lot more theory than data there were arguments about whether polymoprhism (in this era they’re talking about classical markers) was maintained through balancing selection or whether it was just a transient phenomena, and that at any given moment you’re just getting a snapshot of alleles sweeping up to fixation, or being purged out of the gene pool. In the second half of the 20th century it was all about neutral theory, and its discontents. Then the post-genomic era showed up, and geneticists had access to a lot of data and computational power to analyze it. Rather than relying on older molecular tests which were geared toward detecting inter-specific selection events population geneticists began scouring haplotype structure.

But even now there’s a lot of mystery. First, you might be able to adduce that selection is highly likely in a given region, but you may have no clue what that region does functionally (in some cases the region may not even be genic, in which case it has be a mysterious regulatory element). There are some good case studies where the mystery has cleared. Lactase persistence. The ways you can fight malaria. But over the past day I’ve been having to admit that it sure looks like the regions of the genome around pigmentation function are the targets of selection. But we don’t really know what selection is selecting for. And this is actually a set of selection events that I can imagine some day reaching a resolution into their probable cause. But we’re far from that.

A few years ago Eimear Kenney and company solved the mystery of why some Melanesian populations had very dark skins but blonde hair. I blogged about it, but didn’t read the paper too closely. Looking at the publication date, May 2012, I realize I was busy studying for some really big end of first year exams at that time, so that explains my lack of attention. In any case they found that a mutation, rs13289810 in TYRP1, results in blonde hair when it’s a homozygote. They didn’t find strong evidence for recent selection. That is there wasn’t a long haplotype block indicating a sweep in the past 10,000 years. The allele frequency difference across populations as well as long range linkage disequilibrium was suggestive of past selection.

This was in the Solomon Islands. Today I decided to see if there was any follow up on this work. Well, Heather Norton’s group published a paper, Distribution of an allele associated with blond hair color across Northern Island Melanesia. It’s on a different set of islands, but the same results pretty much hold. The allele has a recessive effect on hair color, not much on skin color (there was a small effect in the original paper, so it seems it’s not wholly tissue specific in expression). But I just kept staring at this map and the frequencies. Look at the derived proportions…they don’t get above 0.50. But in most of the populations they’re around in appreciable proportions. I had a hard time not thinking there wasn’t balancing selection going on here. That this was something old that was persisting, but not fixing.

I asked Carlos Bustamante, and he got back me on Twitter:

@razibkhan We estimated >10K yo. Gene is flanked by two recombination hot spots -> hard to estimate selection. @EimearEKenny @foodimprover

— Carlos Bustamante (@cdbustamante) August 17, 2015

I also had an exchange with the first author, and she pointed out in the supplements that the frequencies in the Solomons were quite curious too:

When they looked in the HGDP data set it’s ancestral everywhere else. The derived variant isn’t floating around at low frequencies. One might naively think that it’s overdominance, but I suspect we’re looking at some negative frequency dependent selection. In the 2014 paper by Norton et al. it’s pretty clear that this is distributed across rather disparate populations. It is unlikely in my opinion to be purely due to population structure, as diverse islands have been sampled. It looks to be an old variant that’s persisted, so it dates to the Pleistocene settlement of Near Oceania. It’s also found in Australia, though we don’t know the genetic basis.

Ten years ago I would have been super excited to know the genetic basis of an interesting trait like this. But now I’m left with why? Why? We’ll be grappling with a lot of why’s in the next few decades.