Evolutionary pressures due to variation in climate play an important role in shaping phenotypic variation among and within species and have been shown to influence variation in phenotypes such as body shape and size among humans. Genes involved in energy metabolism are likely to be central to heat and cold tolerance. To test the hypothesis that climate shaped variation in metabolism genes in humans, we used a bioinformatics approach based on network theory to select 82 candidate genes for common metabolic disorders. We genotyped 873 tag SNPs in these genes in 54 worldwide populations (including the 52 in the Human Genome Diversity Project panel) and found correlations with climate variables using rank correlation analysis and a newly developed method termed Bayesian geographic analysis. In addition, we genotyped 210 carefully matched control SNPs to provide an empirical null distribution for spatial patterns of allele frequency due to population history alone. For nearly all climate variables, we found an excess of genic SNPs in the tail of the distributions of the test statistics compared to the control SNPs, implying that metabolic genes as a group show signals of spatially varying selection. Among our strongest signals were several SNPs (e.g., LEPR R109K, FABP2 A54T) that had previously been associated with phenotypes directly related to cold tolerance. Since variation in climate may be correlated with other aspects of environmental variation, it is possible that some of the signals that we detected reflect selective pressures other than climate. Nevertheless, our results are consistent with the idea that climate has been an important selective pressure acting on candidate genes for common metabolic disorders.
The human species inhabits a wide geographical range encompassing a diversity of climates, and adaptation to these climates likely played an important role in shaping genetic and phenotypic variation among populations. We hypothesized that spatially varying selective pressures related to climate shaped the frequencies of genetic variants in the energy metabolic pathway. To test this hypothesis, we examined patterns of genetic variation in 82 candidate genes for common metabolic disorders across the 52 globally dispersed populations of the Human Genome Diversity Project. We applied a combination of statistical approaches to test whether the geographic distribution of these variants could be accounted for by differing climates, consistent with a signal of spatially varying positive selection. For several climate variables, we observed signals in excess of that expected from human population history and chance alone. Significantly, many of these signals were from genes previously shown to affect cold tolerance and disease risk. Our results provide evidence that variation among human populations in susceptibility to common metabolic diseases may be due, in part, to different histories of selective pressures on genes in these disease pathways. Furthermore, our results point to additional genes and variants that are suitable targets for follow-up disease association studies.