Nick Wade in The New York Times has a piece out titled Still Evolving, Human Genes Tell New Story, based on a paper published today in PLOS, A Map of Recent Positive Selection in the Human Genome. This paper is an extension of the research project that emerges out of the International HapMap Project. In short the HapMap is an assay of ~300 individuals from 3 populations, European Americans from Utah, ethnic Yorubas from Nigeria, and a collection of Japanese and Chinese. What can a few hundred individuals tell you? A lot. Wade's piece is a soft landing survey of the major points. His forthcoming book, Before the Dawn, which would have more aptly been titled Still Evolving, elaborates on many of the points he brings up in in the piece. I won't scoop my own book review, but this article is simply an appetizer for what you'll find in Before the Dawn...the title hints at another popularization of scientific genealogy, but within the text there's a lot more stuff about functional genomics than phylogenetics. But, on to the paper...where the real meat is. How: The authors are basically using the three test populations as guinea pigs. Prior to the molecular era geneticists had to assay visible traits, the kink of a wing, a particular coat coloration, etc. In regards to humans pedigree analysis was about all there was. That all changed with the molecular revolution, as researchers began to have access to the informational guts of our genetics. Today, in the "postgenomic era" masses of direct sequence data can be massaged by computational algorithms to scry the information we desire. In this paper the authors are basically looking for signs of selection. Contrary to popular imagination selection, the motive engine behind evolution, is still happening, and it is happening with our own species. Selection occurs via differential fitness correlated with heritable variation. Variation is still around, as is differential fitness. The only question might be whether the variation is heritable...and I think reasonable people can agree that to some extent it is, variation is not totally uncorrelated with your genetic inheritance. While previous research tended to focus on zooming in on a candidate gene upon which we know selection occurred because of phenotypic variation we see around us (e.g., malarial resistence or lactose tolerance), the paper above assays the genome and attempts to detect tell-tale eddies in the sea of neutral genetic variation. Much of the genome is operationally neutral in its variation (pseudogenes, introns, etc.). Substitutions on a locus, a gene, from one allele to another are random and have no fitness implication. As Motoo Kimura showed this implies that the substitution rate is only proportional to the mutation rate (larger populations have weaker drift, but more mutations, smaller populations are the inverse). Additionally, recombination across the chromosomes in diploid organisms breaks apart synteny between alleles on the same sequence of the genome. With a few parameters like this, rates of recombination and neutral substitution, you can formulate a basic null hypothesis and computational models which allow you to test for deviations from expectation. And this is what the authors did in the paper. Basically, if you have a genome which is a flux of alleles moving from 0 to 100% in frequency you will have variation as different loci are in different states of polymorphism. Additionally, recombination, even if heterogeneous in its rate (as the authors assume, and is empirically attested), will break apart assocations between loci across a the genome so that you eventually attain linkage equilibrium. But the null hypothesis doesn't always hold, and this is evidence for selection. Sometimes genes are functionally constrained so powerful selection homogenizes a region of the genome so that no change is allowed. Sometimes you have a situation where two alleles exist at high frequency, and this could be some sort of balancing selection where heterozygotes are more fit than homozygotes. Voight et al., the PLOS paper, examine
instances of selection happening now
. In other words, the statistical test they use, which detects regions of "long haplotypes," that is, homogeneity that hasn't been disrupted on a region of the genome by recombination because selection has been too recent, is looking for current linkage disequilibrium regions. What is likely happening is that an allele is being selected for so strongly that it is dragging along other portions of the genome along with it. Over time recombination should break apart these associations, but until then you see evidence for selection via these long homogenized blocks. The statistical test used in the paper doesn't pinpoint those regions where fixation is nearly complete so that you can't discern the eddie of the long haplotype stranded in the middle of neutral variation. What: Wade's article covers most of the big points. But I'll go over it again. 1) Selection happens, and its happening. The whole idea that humans stopped evolving, that we are outside of God's evolutionary plan, is bunk. There is human variation to attest to selection, from skin color to variations in nutritional metabolism, to disease tolerances. The fact that not all of the regions of the genome detected by this group are at fixation suggests that selection continues (or did until recently, unless you believe in the ubiquity of balancing selection, in which case I have a bridge to sell you). Additionally, the test likely missed genes which are just now rising in frequency due to selection. 2) The time period in terms of "recent" is in the 10,000 to present range. This is a big window, and the paper is sketchy about giving too precise of a handle because they really used some back-of-the-envelope numbers. It seems they want to give you a taste for the character of selection that is reshaping the genome. Wade's article does point to one reality which is salient: the increase in population due to agriculture was really a rock that hit the human genome. 3) Some of the detected regions were already well known. For example, the lactose tolerance region in Europeans, ADH in Asians. Seeing as how the Neolithic Revolution increased population sizes and resulted in a radically alterted lifestyle it is no surprise that genes which control metabolization of our foods changed a lot. Additionally, pathogen resistence was a big issue as packing into unsanitary villages became the norm (many viral infections can't survive at the low densities of pre-agricultural peoples). Several genes related to skin color show evidence of recent selection in Europeans. There was the expected (to my mind) changes found in bone morphogenesis genes, agricultural populations exhibit a more gracile skull form and delictate dentition, the result of universal selective forces that arise out of transition to particular cereals as primary foostuffs. Finally, genes that effect brain and cognitive development are under recent selection, suggesting that behavior and personality might be traits which evolution has reshaped since settled agricultural life. It certainly seems plausible that denser life would result in different personality traits being selected for. 4) Variation is real, and selection is stochastic. Here the skin color data is illustrative. Northwest and Northeast Eurasians are both fair skinned, but their genes differ. Europeans seem to be selected for paleness on particular loci where East Asians are not, while on MC1R they exhibit extreme polymoprhism, in contrast to East Asians who seem subject to recent selection. My point is that phenotype is not that different, but it reaches its end state via an alternative path. Selection on a certain set of genes may allow one to attain the given phenoptype, and that might free other loci from selective constraint. The recent evolution of blondes paper hints at this insofar as MC1R was putatively left free to respond to sexual selection because Europeans were already fixed on other loci for pale skin. 5) Selective sweeps are powerful enough to be transregional. The lactose tolerance gene is a case in point, though it seems that the frequency increased first and is highest for adult milk digestion in Northern Europeans, the capacity is present in other Eurasian populations. This is not because Europeans expanded and filled western Eurasia, it is because the gene spread via selective pressures. Sweeps and linkage disequilibrium shared between populations is evidence of powerful genetic forces that undergird our genomes. The probability of fixation of a new positive mutation is 2s, where s is the selection coefficient, so deme-to-deme migration could result in wildfire spread of new traits (each introduction via intermarriage can be modeled as a mutation). 6) A few hundred individuals from 3 populations. Do you really think this is the end? Cheap genomics I hear your siren song! We live in the age of Kepler. The Principia has yet to be unveiled...and the sweet low hanging fruit beckons.
