doi: 10.1126/science.aag0776. Epub 2016 Oct 13.
Detection of human adaptation during the past 2000 years
Affiliations
- PMID:27738015
- PMCID: PMC5182071
- DOI: 10.1126/science.aag0776
Item in Clipboard
Detection of human adaptation during the past 2000 years
Yair Field et al. Science..
Display options
Format
Display options
Format
Abstract
Detection of recent natural selection is a challenging problem in population genetics. Here we introduce the singleton density score (SDS), a method to infer very recent changes in allele frequencies from contemporary genome sequences. Applied to data from the UK10K Project, SDS reflects allele frequency changes in the ancestors of modern Britons during the past ~2000 to 3000 years. We see strong signals of selection at lactase and the major histocompatibility complex, and in favor of blond hair and blue eyes. For polygenic adaptation, we find that recent selection for increased height has driven allele frequency shifts across most of the genome. Moreover, we identify shifts associated with other complex traits, suggesting that polygenic adaptation has played a pervasive role in shaping genotypic and phenotypic variation in modern humans.
Copyright © 2016, American Association for the Advancement of Science.
Figures
(A) Simulated frequency trajectory for a derived allele that was selected from standing variation starting 100 generations ago. (B) Corresponding genealogy of 3000 present-day genomes. Lineages carrying the derived allele (D) are in blue; ancestral (A) are in black. Enlargement of the genealogy illustrates that tip branches carrying the favored allele (blue) are on average shorter than those carrying the disfavored allele (black). (C) Because favored alleles (blue) tend to have shorter tip branches, their haplotypes tend to have lower singleton density. (D) For each individual, we compute the distance between nearest singletons around the test SNP. (E) Distribution of singleton distances as a function of genotype at the simulated test SNP. (F) Mean tip length t is estimated for each allele from a likelihood model. Unstandardized SDS is a log-ratio of estimated tip lengths; this is standardized to mean 0, variance 1 within bins of derived allele frequency. In this simulated example, SDS is highly significant (P = 1 × 10−17 in favor of the derived allele; relative to neutral simulations). Compare with illustration of drift simulation (fig. S1).
(A) Mean tip length as a function of sample size, for a demographic model with strong recent growth (14) (additional models shown in fig. S10). (B) Power simulations for SDS (mean ± SD) under three models of selection with current derived allele frequency of 0.7: continuous hard sweep (orange); selection starting 100 generations ago (cyan); and hard sweep that stopped 100 generations ago (black). Right panel: corresponding simulations for iHS. (C) SDS and iHS, for sweeps that stopped in the past (s = 0.10), followed by neutral drift. (D) Power to detect simulated selection from present variation using half of the UK10K data. We biasedly sampled 1500 genomes out of 3195 without replacement so as to change the frequencies at randomly chosen SNPs. (E) Allele frequency differences between extant populations (1000-Genomes) versus SDS or iHS. SDS is most correlated with the difference between northern and southern Europe, whereas iHS reflects Europe versus Africa divergence. GBR, British; CEU, Utah residents (northwest European ancestry); IBS, Iberians (Spain); TSI, Tuscans (Italy); CHB, Han (China); JPT, Japanese; YRI, Yoruba (Nigeria); LWK, Luhya (Kenya).
(A) Manhattan plot of SDSP values indicates regions of genome-wide significance (P < 5 × 10−8;P values are two-sided tail probabilities of standard normal). (B) Distributions of singleton distances at the lactase locus, partitioned by genotypes at the causal site. Compare to simulated signals (Fig. 1E). (C) SDS signals for a curated set of segregating variants with known effects on pigmentation shows overall increase in derived allele frequencies (one-sidedP values). (D) Distribution of tSDS scores at 551 height-associated SNPs. tSDS is polarized so that tSDS > 0 implies increased frequency of the “tall” allele.
(A) Mean tSDS of SNPs, where tSDS > 0 implies increased frequency of the “tall” allele in a recent family-based study (20). Thex axis is ordered from least significant SNPs (P ~ 1) to most significant (P ~ 0), and SNPs are placed into bins of 1000 consecutive SNPs for easier visualization. (B) Covariance of height Z score and SDS, as a function of LD score, provides evidence that selection on height is truly polygenic (P = 2 × 10−11; LD score regression). (C) QQ-plot testing for a correlation between GWAS Z score and tSDS for 43 traits. tSDS > 0 implies increased frequency of the “trait-increasing” allele. Significant traits that are also nominally significant by LD score regression (P < 0.05, one-sided test) are labeled.
Comment in
- Geneticists are starting to unravel evolution's role in mental illness.Reardon S.Reardon S.Nature. 2017 Oct 30;551(7678):15-16. doi: 10.1038/nature.2017.22914.Nature. 2017.PMID:29094708No abstract available.
References
MeSH terms
Substances
Associated data
Grants and funding
LinkOut - more resources
Full Text Sources
Other Literature Sources
Miscellaneous