Page 1 of 36Free preview · 1–5

About this document

This Nature research article introduces a method for detecting directional selection in ancient DNA while controlling for population structure. The authors analyse 15,836 ancient West Eurasians spanning 18,000 years, including 10,016 individuals with new data, alongside 6,438 contemporary individuals. They estimate selection coefficients for approximately 9.7 million variants and identify 479 independent loci with strong evidence of directional selection. Analyses also reveal coordinated changes in alleles associated with modern complex traits, including pigmentation, body fat and psychiatric and behavioural traits. The article emphasizes that modern genetic associations do not establish which phenotypes were adaptive in ancient societies. Methods, data and code availability statements, and extended data accompany the findings.

Free preview text

The first 5 pages, shown as text so search engines and readers can find this document.

Page 1

Nature | www.nature.com | 1 Article Ancient DNA reveals pervasive directional selection across West Eurasia Ali Akbari1,2,3 ✉, Annabel Perry2,3, Alison R. Barton2,3, Mohammadreza Kariminejad4, Steven Gazal5,6,7, Zheng Li8, Yating Zeng5,9, Alissa Mittnik2,10, Nick Patterson2,3, Matthew Mah1,11, Xiang Zhou12, Alkes L. Price3,13,14, Eric S. Lander3,15,16, Ron Pinhasi17,18, Nadin Rohland1,2,3, Swapan Mallick1,2,3,11 & David Reich1,2,3,11 ✉ Ancient DNA has transformed our understanding of population history1, but its potential to reveal as much about human evolutionary biology has not been realized because of limited sample sizes and the difficulty of distinguishing sustained rises in allele frequency increasing fitness—directional selection—from shifts due to migrations, population structure, or non-adaptive purifying or stabilizing selection2–7. Here we present a method for detecting directional selection in ancient DNA time-series data that tests for consistent trends in allele frequency change over time, and apply it to 15,836 West Eurasians (10,016 with new data). Previous work has shown that classic hard sweeps driving advantageous mutations to fixation have been rare over the broad span of human evolution8,9. By contrast, in the past ten millennia, we find that many hundreds of alleles have been affected by strong directional selection. We also document one-standard-deviation changes on the scale of modern variation in combinations of alleles that today predict complex traits. This includes decreases in predicted body fat and schizophrenia, and increases in measures of cognitive performance. These effects were measured in industrialized societies, and it remains unclear how these relate to phenotypes that were adaptive in the past. We estimate selection coefficients at 9.7 million variants, enabling study of how Darwinian forces couple to allelic effects and shape the genetic architecture of complex traits. Ancient DNA data hold extraordinary promise for revealing adapta- tion, making it possible to track effects across time and obtain direct measurements of selection coefficients10,11. Rather than being trapped in the present and studying the scars left by selection on the genomes of descendants12,13, ancient DNA makes it possible to test directly whether frequencies of variants shifted more than could be expected by chance2–5. Such data also make it easier to measure selection on variants not of recent mutational origin (standing variation), which is challenging to detect using retrospective methods14. Previous ancient DNA selection studies in West Eurasia2–5 (Europe and its neighbours in the Near East) have identified dozens of alleles influenced by selec- tion, but despite growth in reported ancient individuals to more than 10,000 today15, the number of genome-wide significant loci reported in a single study grew only from 12 in the first genome scan in 2015 (ref. 2) to 21 in 2024 (ref. 4), raising the concern that this approach might not deliver broad insights into human adaptation. Here we increase the yield of discoveries by 20-fold by adding more statistical power (due to a qualitatively new method and larger sample size) and reduc- ing artefactual signals (through intensive data cleaning). First, we increased power by testing for a consistent trend in allele frequency change over time. Several past studies have dealt with the challenge posed by population mixture by treating more recent popula- tions as linear combinations of more ancient ones, then searching for alleles whose frequencies were outliers compared with what would be expected from this history2,3. However, changes in frequency due to selection are often less than what can be expected from gene flow and random genetic drift6, and in this context, increasing sample size helps little. We used a qualitatively different approach, using the genetic similarity of each individual to every other, and testing whether the date when they lived provides additional predictive power for the allele frequencies of their population beyond what is expected from the empirical population structure. Our test is simple: at each variant https://doi.org/10.1038/s41586-026-10358-1 Received: 14 September 2024 Accepted: 4 March 2026 Published online: xx xx xxxx Check for updates 1Department of Genetics, Harvard Medical School, Boston, MA, USA. 2Department of Human Evolutionary Biology, Harvard University, Cambridge, MA, USA. 3Broad Institute of MIT and Harvard, Cambridge, MA, USA. 4Shamsipour Technical and Vocational College, Tehran, Iran. 5Department of Population and Public Health Sciences, Keck School of Medicine, University of Southern California, Los Angeles, CA, USA. 6Center for Genetic Epidemiology, Keck School of Medicine, University of Southern California, Los Angeles, CA, USA. 7Department of Quantitative and Computational Biology, University of Southern California, Los Angeles, CA, USA. 8Department of Biostatistics, School of Public Health, University of Michigan, Ann Arbor, MI, USA. 9Department of Biostatistics and Data Science, School of Public Health, The University of Texas Health Science Center at Houston, Houston, TX, USA. 10Department of Archaeogenetics, Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany. 11Howard Hughes Medical Institute, Harvard Medical School, Boston, MA, USA. 12Department of Statistics and Data Science, Yale University, New Haven, CT, USA. 13Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA, USA. 14Department of Biostatistics, Harvard T.H. Chan School of Public Health, Boston, MA, USA. 15Department of Systems Biology, Harvard Medical School, Boston, MA, USA. 16Department of Biology, Massachusetts Institute of Technology (MIT), Cambridge, MA, USA. 17Department of Evolutionary Anthropology, University of Vienna, Vienna, Austria. 18Human Evolution and Archaeological Sciences, University of Vienna, Vienna, Austria. ✉e-mail: Ali_Akbari@ fas.harvard.edu; reich@genetics.med.harvard.edu

Page 2

2 | Nature | www.nature.com Article we asked whether hypothesizing a non-zero selection coefficient s— causing allele frequency to trend in the same direction over all times and places—predicts frequency differences across populations signifi- cantly better than empirically measured population structure alone. Second, we increased power through a 14-fold increase in sample size, driven by 10,016 ancient individuals for whom we report new data, which combined with previously reported data yields a dataset of 15,836 people spanning 18,000 years (Supplementary Tables 1–3). We co-analysed with 6,438 modern people, sub-sampled so that their countries of origin were spread evenly over West Eurasia (Extended Data Fig. 1a,b). Third, we increased power by imputing diploid genotypes16, leverag- ing known patterns of allelic correlation in a modern reference panel to fill in missing data in all individuals. We also carried out intensive data cleaning (Supplementary Section 1), filtering out single-nucleotide polymorphisms (SNPs) susceptible to false positives. The final dataset included 8,074,573 SNPs and 1,665,051 insertions or deletions (indels) on chromosomes 1–22. Test of selection on single variants For each SNP in the genome, we estimated a selection coefficient, s, which we found has a standard error of typically 0.17% for common variants (Extended Data Fig. 1c,d). In theory, a valid test for selection should be Z, the number of standard errors this quantity is from zero, and we should be able to use a normal distribution to identify scores that pass the standard threshold of genome-wide significance (P%<%5%×%10−8). In practice, the median χ2 statistic (squared Z-score for the number of standard errors s is from zero) is inflated by 4.75-fold relative to a χ2 distribution with 1%d.f. In human genetics studies, such inflation can arise due to uncorrected population structure or non-normality of the null distribution due to case–control sample imbalance17. Inflation like this is often addressed by rescaling χ2 statistics by the median inflation across the genome18,19. But if the null is not normally distributed, or a substantial fraction of the genome carries real signal18, random sites will not provide an appropriate neutral baseline. These concerns apply here: the null distribution is not normal due to frequency-dependent biases from imputation and the non-linear transformation of allele frequencies (Supplementary Section 2), and most of the genome is in linkage disequilibrium with sites showing evidence of directional selection (Extended Data Fig.%2b and Supplementary Section 3). Instead, we calibrated our test by taking advantage of a finding about the connection between selection coefficients and associa- tions to phenotypes in living people. The proportion of SNPs showing significant association to a phenotype in genome-wide association studies (GWAS) increases with our selection statistic, plateauing at around 5.2 times the rate for random SNPs (Fig. 1a; this analysis relies on 1,317,022 SNPs with a genome-wide significant association to at least one phenotype for 452 traits in the UK Biobank, and conditions on minor allele frequency (MAF) to remove bias due to signals being easier to detect for higher MAFs). The plateau occurs at the same place when we control for ‘background selection’, the loss of neutral varia- tion linked to deleterious mutations removed by purifying selection, although the enrichment in GWAS variants is reduced (approximately 2.4 times; Extended Data Fig. 3a). This is the pattern expected for a true threshold for genome-wide significance: if SNPs beyond this threshold reflect a combination of true signal and false discoveries, we would expect enrichment to continue beyond it. This threshold occurs at Z%=%8.23, larger than the standard threshold (5.45) for genome-wide significance for a normal distribution, so we rescaled the naive score by this quantity to obtain an X-statistic (Z/1.51) whose significance threshold matches the standard threshold. To validate our procedure, we conducted realistic forward-in-time simulations (Fig. 1b, Extended Data Fig. 4 and Supplementary Sec- tion 2). The simulations specify a demographic history inspired by that of Europe and produce patterns of structure matching impor- tant features of real data. The simulations include coding, functional non-coding and neutral regions, recapitulating varying degrees of background selection. The simulations also permit exploration of the relationship between selected alleles and traits, in particular the phenomenon of ‘stabilizing’ selection to remove genetic variation when the average genetic predictor of a trait is at a fitness optimum. We verified that these simulations produce a deficiency of common polymorphisms influencing traits, matching patterns in real data20. Our test for directional selection is robust to every confounding fac- tor that we considered, including population structure, background selection due to purifying or stabilizing selection, sampling bias and selection in the ancestral population but not during the time transect itself (Supplementary Section 2). None of these produces even a hint of the enrichment in GWAS signals that we observe in data. By contrast, in simulations with directional selection, including where we shift the trait optimum, leading to deterministic changes in allele frequency to bring the population closer to that optimum21, we observed exactly this signal and could use it to provide a valid calibration of genome-wide sig- nificance (Fig. 1b, Extended Data Fig. 4 and Supplementary Section 2). To translate X to a posterior probability of selection π, we used a false discovery rate (FDR) approach (Fig. 1a,b and Supplementary Section 3). We fit a smooth function to the enrichment curve for GWAS signals and estimate that at X greater than our threshold of 5.45, π >%99%. A second line of evidence that these signals are real comes from com- puting the mean haplotype allele frequency (HAF) score, the sum of squared derived allele frequencies scaled by sample size in 200-kb win- dows surrounding each tested variant. Previous work22,23 has shown that directional positive selection on derived alleles is expected to increase HAF scores, whereas background selection is expected to decrease it (Extended Data Fig. 3b and Supplementary Section 4). After comput- ing the residual HAF score for each variant controlling for background selection24,25, we found that it increases with the X-statistic and begins to rise before 5.45, the standard threshold for genome-wide significance in GWAS (Fig. 1c and Extended Data Fig. 3c). Our forward-in-time simu- lations (Supplementary Section 2) confirm that the rise in HAF score occurs only for directional selection, with no hint of this pattern for other scenarios (Fig. 1d and Extended Data Fig. 4). We obtained a third line of evidence that we are detecting real sig- nals of adaptation by showing that |X|%>%5.45 variants are associated with some classes of traits much more than others. We found enrich- ment for SNPs contributing to blood–immune–inflammatory traits (95% confidence interval (CI) 3.12–8.24)3,5, compared with random SNPs with matched characteristics defining the baseline. For mental– psychiatric–nervous and behavioural traits, we did not detect enrich- ment (95% CIs of 0.90–1.08 and 0.99–1.46, respectively; Fig. 1e and Extended Data Fig. 5a). This cannot be explained by differences in allele frequencies or background selection as we controlled for these factors. The intensity of selection on both blood–immune–inflammatory and cardiometabolic traits increased significantly in the Bronze Age relative to the pre-farming period (Fig. 1e and Extended Data Fig. 5b), plausibly reflecting adaptation to new diets, larger population or living closer to domesticated animals26. Hundreds of cases of directional selection We found evidence of 479 independent loci (410 excluding the HLA region) with |X|%>%5.45, corresponding to a π%>%99% probability of selec- tion (Fig. 2a). To produce this list, we identified the strongest signal in the genome and considered all SNPs in linkage disequilibrium with it in unrelated Western European individuals (r2%>%0.05) to potentially reflect the same signal. We then found the second-strongest signal excluding these positions, and so on until no more variants passed this threshold (Extended Data Fig. 2b). Visualizations of the trajectories for these loci (Supplementary Section 5) and summary statistics for

Page 3

Nature | www.nature.com | 3 9.7 million variants (Supplementary Table 4) can be cross-referenced with GWAS and with their frequency trajectories at the AGES browser (https://reich-ages.rc.hms.harvard.edu). The actual number of loci under selection is likely to be much larger. Using a threshold of |X|%>%3.61 (FDR%=%50%), we identified 7,689 non-HLA loci, implying more than 3,800 independent episodes of selection. The exact number of distinct signals is difficult to count: residual linkage dis- equilibrium exists among candidate loci, which at some loci may cause some overestimation of signals (Supplementary Section 5), whereas truly selected alleles in linkage disequilibrium with nearby stronger alleles will be undercounted. Downsampling analyses showed that further increases in sample size are expected to increase the number of detected loci further, with people living more than 8,000 years ago providing the most added power (Extended Data Fig. 1d,e). To obtain insight into the phenotypes shaped by directional selec- tion, we took advantage of the fact that a high proportion (62%) of the variants with genome-wide evidence of selection are independently associated to a phenotype in at least one UK Biobank GWAS. However, biological interpretation is complicated as the allele that was the target of selection may differ from the tag SNP that we are using to represent the locus (and may even be in a neighbouring gene); because some alleles affect multiple phenotypes, because the relevant modern trait a b c d e Category Number of traits Heritability enrichment (95% CI) Z-score for selection stronger in Bronze Age than Neolithic or Mesolithic Historical period than Bronze Age Biomarkers 9 5.21 (3.17–7.26) 3.07 0.31 Blood–immune–inflammatory 20 5.68 (3.12–8.24) 3.96 1.92 Cardiometabolic 20 2.07 (1.27–2.87) 3.58 0.14 Life history–reproduction 7 1.89 (0.92–2.86) 2.54 0.00 Behavioural 13 1.23 (0.99–1.46) 1.96 0.95 Mental–psychiatric–nervous 13 0.99 (0.90–1.08) –0.61 0.33 Other 25 1.86 (1.23–2.49) 2.83 0.46 All 107 2.43 (1.86–3.00) 6.97 2.10 5.0 0 40 36 32 28 24 20 16 12 8 4 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 4.5 4.0 3.5 3.0 2.5 Enrichment Residual HAF/n Enrichment FDR 2.0 1.5 1.0 6 10 8 6 4 2 0 –2 5 4 3 2 1 0 0 1 2 3 4 |X| 5 6 7 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 FDR 0 1 2 3 4 |X| 5 6 7 0 1 2 3 4 |X| 5 6 7 0 1 2 3 4 |X| 5 6 7 Residual HAF/n FDR With directional selection No directional selection With directional selection No directional selection Fig. 1 | Multiple lines of evidence show that we are detecting genuine directional selection. a, Enrichment of SNPs significant in at least one of 452 UK Biobank GWAS, for SNPs with |X| above the value on the x axis, controlling for allele frequency. b, Simulations reveal enrichment of GWAS hits (blue; left axis) closely aligns with 1%−%FDR (orange; right axis) until a plateau as a function of |X|: 800 simulations with directional selection (200 each for selection coefficients of 0.01, 0.02 or 0.03 for model 2.1, soft sweep; and 0.05 for model 2.2, hard sweep) and 800 simulations without it. c, Residual mean HAF score ((HAF)/n) for SNPs with |X| above the value on the x axis, computed as observed minus expected, with n haploid sample size, from a linear regression correcting for background selection using as explanatory variables McVicker-B, Murphy-phastCons, Murphy-CADD, number of SNPs and heterozygosity in a 200-kb window. d, Residual mean HAF score in simulation for SNPs with |X| above the value on the x axis, computed as observed minus expected, with n haploid sample size, from a linear regression correcting for background selection using as explanatory variables McVicker-B, coding region annotation, functional non-coding region annotation, number of SNPs and heterozygosity in a 200-kb window. The shaded areas show 95% CIs (a–d). e, The heritability enrichment is a meta-analysis for annotations based on a binary selection annotation, with FDR either below 1% (1) or above 1% (0). The Z-score for change in selection intensity over time is based on a meta-analysis of heritability enrichment comparing key cultural transitions: Mesolithic–Neolithic to Bronze Age, and Bronze Age to historical era. We annotated each SNP according to whether it is among the top 5% with the highest probability of a stronger magnitude of selection coefficient in one time transect versus another. Cells shaded grey denote P < 0.05.

Page 4

4 | Nature | www.nature.com Article may not be measured in one of the GWAS that we are analysing, or because the phenotype in modern societies may not have existed in the ancient societies where selection acted. The median selection mag- nitude |s| at the tag SNPs for non-HLA loci is 0.86% (range of 0.43–4.3%), and the median MAF is 13%. Standard errors in our estimates of |s| for common alleles are 0.17% (Fig. 2b and Extended Data Fig. 1c,d). We compared our results to those of five previous scans for selection in Holocene West Eurasia (four based on ancient DNA)2–5,13 (Extended Data Table 1 and Supplementary Section 6). Of 38 unique non-HLA loci that met the formal threshold for genome-wide significance in at least one previous study, 18 passed our π%>%99% threshold. The other 20 did not replicate, in most cases due to what appears to be incompletely controlled population structure driven by mixtures of populations with different allele frequencies before they came together and in a few cases due to failing quality control or fluctuating selection in space or time. We present a gallery of 36 single-allele trajectories of particular interest (Fig. 3) as well as estimates of how their selection coefficients changed over time (Extended Data Fig. 6). These loci are not necessar- ily those with the largest X scores, but are highlighted as they address long-standing debates. They include 27 passing the π%>%99% threshold, 5 with probable evidence of selection (57%%<%π%<%93%) and 4 with sur- prising negative findings. HLA-DQB1: selection in favour of the major risk factor for coeliac disease At the HLA region of chromosome 6, densely packed genes have key roles in microorganism recognition. rs3891176 (C>A, meaning that the ancestral allele is C and the newly arising mutation is A) is an excellent tag for HLA-DBQ1*02/DQ2 HLA-DQB1*02/DQ2, with individuals carrying two A alleles having a 19-fold higher susceptibility for coeliac disease or gluten sensitivity (Extended Data Fig. 7a,b). The A allele has a selec- tion coefficient of s%=%4.4% (π%>%99%), rising from approximately 0% to approximately 20% in the past 4,000 years (Fig. 3a). These findings suggest that the pathogenic exposures that drove its rise were not a phenomenon only or largely of the rise of agriculture27. ABO: positive selection for B at the expense of A The ABO gene modifies oligosaccharides in glycoproteins on the surface of red blood cells; its A, B and null (O) alleles encode distinct glycopro- tein alterations that modulate resistance and susceptibility to diverse pathogens28. The B allele rose from approximately 0% to approximately a b c 0 4.5 ABO SLC39A8 NSF KITLG DUOX2 MUC4 TCHH OAS1 SLC45A2 LCT TYR ATXN2 SLC22A4 FADS1 TLR1 CYP1A2 ADH1CDHCR7 FUT3 MUC5AC 106 90 70 50 30 10 –0.02 –0.01 0 0.01 0.02 105 104 104 105 Derived allele age from Relate (years before present) Derived allele selection coefficient (s) Simulation power (%) 106 4.0 Selection coefficient s (%) Favoured allele age given a selective sweep (years before present) 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0 0.05 0.10 0.15 0.20 MAF 0.25 0.30 0.35 0.40 0.45 0.50 2 4 6 8 10 PLA2G2A SLC44A5 FLG TCHH FCGR2A/3A/2B RBBP5, CNTN2, DSTYK, TMCC2 MARC1 SDC1 PCBP1-AS1 IL1 GCA, IFIH1 LCT, MCM6 CCR5 DNAH1, NISCH, TNNC1, SEMA3G RDUR MED12L MUC4 TLR1, TLR6 ADH1B/CABCG2 CENPE, SLC39A8 GYPA, GYPB TPPP ELOVL7 SLC45A2 SLC22A4/5 MHC CDHR3 CFTR, ASZ1 8p23.1 inv. locus SLC7A13, WWP1, CPNE3 NSMCE2 BNC2 9p21.1 9q31.2, RP11-240E2.2 ABO FADS1/2 DHCR7, NADSYN1 TYR, GRMS POU2F3 ESPL1 KITLG SH2B3, ATXN2 CRYL1 PCDH17 TNFSF13B PLEKHG3 IGHG2 SORD, DUOX1/2/A1 CSK, CYP1A1/2 ITGAM DEF8, MC1R, ZNF276, CDH15, ANKRD11 BLMH, KSR1 17q21.31 inv. locus GRB2, MRPS7, NUP85, GGA3 FUT3, FUT6 FUT2 ASXL1 CEACAM19 B3GALT5 SREBF2, CENPM, TNFRSF13C (BAFF-R) MUC2, MUC5AC RSPH4A, ZUFSP, KPNA5, RWDD1 |X| 12 14 16 18 20 1 2 3 4 5 6 7 Chromosome 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 Fig. 2 | Genome scan for directional selection. a, The x axis is the chromosomal position, and the y axis is the selection signal for each variant. The dotted line indicates our genome-wide significance threshold of |X|%=%5.45. For clarity, only a subset of loci are annotated. b, Selection coefficient (s) estimated from our scan plotted against MAF of tagging SNPs at independent loci with ( > 99%. Overlaid grids are simulation-based power estimates (90%, 70%, 50%, 30% and 10% probability of detection). The error bars indicate standard errors from 20,374 unrelated individuals. c, The estimated age of the favoured allele in a selective sweep versus the date of origin of the mutation inferred from Relate58, for tagging SNPs at independent loci with ( > 99%. The age of the sweep is defined as the time before present when the frequency of the favoured allele is expected to have been 0.0001 given the present-day frequency in 1000 Genomes Project European populations and assuming that the selection coefficient has been constant over time.

Page 5

Nature | www.nature.com | 5 10% over the past approximately 6,000 years (s%=%2.7%, π%>%99%), and was matched by a concomitant decrease in A frequency (Fig. 3b). The two alleles are associated with opposite effects on many phenotypes, suggesting that with changing pathogenic exposures, the optimal balance changed (Extended Data Fig. 7c,d). TCHH: selection for an allele that reduced male-pattern baldness An allele at missense SNP rs11803731 (A>T) in TCHH is a strong predictor of straight hair and male-pattern baldness in European individuals. The derived allele T is rare in African and East Asian individuals, and has been 0.25 a 0.16 0.08 0.06 0.12 0.12 0.05 0.04 0.03 0.02 0.01 0 0.10 0.08 0.06 0.04 0.02 0 0.08 0.04 0 0.05 0.04 0.03 0.02 0.01 0 0.06 0.04 0.02 0 0.15 0.6 1.0 0.5 0.30 1.0 0.05 0.04 0.03 0.02 0.01 0 0.8 0.6 0.4 0.2 0 0.25 0.20 0.15 0.10 0.05 0 0.4 0.3 0.2 0.1 0 0.8 0.6 0.4 0.2 0 0.4 0.2 0.10 0.8 0.6 0.4 0.2 0 0.8 0.9 1.0 0.5 0.4 0.3 0.2 0.1 0 0.9 0.8 0.7 0.6 0.5 0.8 0.7 0.6 0.5 0.6 0.4 0.2 0 0.08 0.06 0.04 0.02 0 0 0.08 0.04 0 0.12 0.55 0.12 0.15 0.06 0.05 0.04 0.03 0.02 0.01 0 0.10 0.05 0 0.10 0.08 0.06 0.04 0.02 0 0.45 0.35 0.25 0.15 0.10 0.08 0.06 0.04 0.02 0 HLA-DQB1 rs3891176 (C>A) Coeliac disease susceptibility CCR5-Δ32 rs333 (+>Δ32) HIV resistance BNC2 rs12350739 (G>A) Light hair or skin tone TSBP1 rs7775397 (T>G) Coeliac disease susceptibility *TLR1 rs5743618 (A>C) Leprosy resistance GCKR rs1260326 (C>T) Hyperlipidaemia or gout susceptibility SERPINA1 rs17580 (T>A) α1-Antitrypsin deficiency SERPINA1 rs28929474 (C>T) α1-Antitrypsin deficiency PTPN22 rs2476601 (G>A) Hypothyroidism susceptibility HFE rs1799945 (C>G) Haemochromatosis *IL23R rs11209026 (G>A) Crohn’s disease resistance *CYP1A2 rs2472304 (G>A) Blood pressure (decrease) *NADSYN1/DHCR7 rs2276360 (G>C) Vitamin D levels (increase) *ADH1B rs1229984 (C>T) Lower risk for alcoholism ABCG2 rs2231142 (G>T) Gout susceptibility APOE rs7412 (C>T) Hyperlipidaemia HLA-DQB1 rs1130398 (T>C) Coeliac disease susceptibility HLA-DRB1 rs660895 (A>G) Rheumatoid arthritis susceptibility GYPA rs7658293 (C>A) MNS blood group N (increase) DUOX2 rs57659670 (C>T) Ferritin level (increase) *SLC22A4 rs1050152 (C>T) Crohn’s disease susceptibility *SLC45A2 rs16891982 (C>G) Light hair or skin tone *TYR rs1042602 (C>A) Light hair or skin tone *TYR rs1126809 (G>A) Light hair or skin tone SLC24A5 rs1426654 (G>A) Light hair or skin tone CFTR-ΔF508 rs113993960 (+>Δ3) Cystic fibrosis MC1R rs1805008 (C>T) Light hair or skin tone (red hair) MC1R rs1805007 (C>T) Light hair or skin tone (red hair) DSTYK rs3851294 (G>A) Light hair or skin tone KITLG rs12821256 (T>C) Light hair or skin tone RALY/ASIP rs62209647 (G>C) Light hair or skin tone *ABO rs8176746 (G>T) Blood type B (increase) *TYK2 rs34536443 (G>C) Tuberculosis susceptibility *HLA-DRB1 rs3135388 (G>A) Multiple sclerosis susceptibility HFE rs1800562 (G>A) Haemochromatosis TCHH rs11803731 (A>T) Straight hair (increase) or baldness (increase) 0.20 0.15 0.10 0.05 0 8,000 4,000 0 Time before present (years) Derived allele frequencyDerived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 8,000 4,000 0 Time before present (years) 0.8 0.7 0.75 0.08 0.12 0.10 0.08 0.06 0.04 0.02 0 0.10 0.08 0.06 0.04 0.02 0 0.06 0.04 0.02 0 0.12 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0 0.30 0.25 0.20 0.15 0.10 0.05 0.08 0.04 0 0.65 0.55 0.45 0.35 0.6 0.5 0.4 0.3 0.2 0.1 0.6 0.4 0.2 0.6 0.06 0.05 0.04 0.03 0.02 0.01 0 0.5 0.4 0.3 0.2 0.1 0 0 Derived allele frequencyDerived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency Derived allele frequency 0.06 0.05 0.04 0.03 0.02 0.01 0 s = 4.4% |X| = 12.37  > 0.99 s = 2.7% |X| = 10.44  > 0.99 s = –0.9% |X| = 9.60  > 0.99 s = 2.3% |X| = 5.62  > 0.99 s = –1.9% |X| = 7.15  > 0.99 s = 4.2% |X| = 12.57  > 0.99 s = –2.1% |X| = 6.60  > 0.99 s = 2.7% |X| = 6.23  > 0.99 s = –1.3% |X| = 2.54  > 0.11 s = 2.3% |X| = 6.63  > 0.99 s = 1.6% |X| = 6.49  > 0.99 s = 1.3% |X| = 7.43  > 0.99 s = 1.9% |X| = 6.54  > 0.99 s = 1.9% |X| = 6.90  > 0.99 s = 1.2% |X| = 6.43  > 0.99 s = 0.6% |X| = 6.14  > 0.99 s = 5.6% |X| = 6.66  > 0.99 s = 2.4% |X| = 13.15  > 0.99 s = 1.7% |X| = 13.37  > 0.99 s = 1.0% |X| = 9.62  > 0.99 s = 1.4% |X| = 4.88  = 0.78 s = 0.0% |X| = 0.07  < 0.01 s = 1.0% |X| = 7.68  > 0.99 s = –0.9% |X| = 6.48  > 0.99 s = –0.9% |X| = 10.54  > 0.99 s = 1.3% |X| = 10.74  > 0.99 s = 1.8% |X| = 15.99  > 0.99 s = 0.9% |X| = 5.08  = 0.87 s = 0.9% |X| = 5.87  > 0.99 s = 2.3% |X| = 8.84  > 0.99 s = 0.9% |X| = 9.81  > 0.99 s = 1.0% |X| = 10.41  > 0.99 s = 2.0% |X| = 11.64  > 0.99 s = 0.5% |X| = 5.03  = 0.84 s = 1.6% |X| = 4.40  = 0.57 s = 2.5% |X| = 5.16  = 0.90 s = –2.3% |X| = 5.25  = 0.93 s = –0.0% |X| = 0.31  = 0.01 s = 0.4% |X| = 1.27  = 0.03 s = –0.5% |X| = 3.52  = 0.27 s = 2.1% |X| = 2.25  = 0.08 s = –0.6% |X| = 3.72  = 0.33 b c d e f g h i j k l m n o p q r s t u v w x y z aa ab ac ad ae af ag ah ai aj Fig. 3 | Gallery of single-variant allele frequency trajectories. a–aj, Each panel displays the derived allele frequency trajectory (solid line) over time for a variant (uncorrected for structure), along with selection coefficient (s), selection statistic (X) and posterior probability of selection (π). The shaded regions represent 95% CIs around the estimated trajectory. The circles represent frequencies in Western hunter–gatherers (green; n%=%131), early European farmers (orange; n%=%452) and steppe pastoralists (pink; n%=%293); the error bars indicate 95% CIs. The highlighted variants are not necessarily those with the strongest signals and may include negative results or variants in partial linkage disequilibrium. We highlight them here because of their biological interest and because they speak to long-standing debates. For panels d–f,ag,ai,aj, separate analyses are shown for transects before (blue) and after (red) a manually selected peak (marked by a black dashed line). In cases where π >%99%, the CI is shaded blue (or blue before and red after the split); otherwise, the shading is grey. Variants reported in other ancient DNA studies are marked with an asterisk.

Tags

Doxxly editorial guide

Ancient DNA reveals pervasive directional selection across West Eurasia: summary, contents and background

This guide helps you understand the scope of Ancient DNA reveals pervasive directional selection across West Eurasia before or alongside reading the original document. It is based on the catalogue record, credited creator, source, classification and topics shown on this page. It does not replace the original work and does not invent claims about passages that Doxxly cannot verify.

Summary and overview

Ancient DNA reveals pervasive directional selection across West Eurasia is a 36-page documentcredited to Vesna F. Oakfield. This Nature research article introduces a method for detecting directional selection in ancient DNA while controlling for population structure. The authors analyse 15,836 ancient West Eurasians spanning 18,000 years, including 10,016 individuals with new data, alongside 6,438 contemporary individuals. They estimate selection coefficients for approximately 9.7 million variants and identify 479 independent loci with strong evidence of directional selection. Analyses also reveal coordinated changes in alleles associated with modern complex traits, including pigmentation, body fat and psychiatric and behavioural traits. The article emphasizes that modern genetic associations do not establish which phenotypes were adaptive in ancient societies. Methods, data and code availability statements, and extended data accompany the findings. It appears in the Science & Engineering collection, which groups it with other material that readers may use for comparison and further study.

The item can be approached in several ways. A reader looking for a quick orientation can begin with the free five-page preview, inspect the page structure and use the metadata to judge relevance. Someone undertaking deeper research can work through the complete file, search its indexed text, record useful page references and compare its treatment of Ancient DNA, Directional selection, Human evolution, Population genetics, West Eurasia, Polygenic adaptation, Genome-wide association studies, Archaeogenetics with the related resources linked below.

Rather than reducing the work to a single claim, this page treats it as a source that should be read in context. The title, creator, publication source, licence and format all help explain what kind of evidence it may offer. Readers should distinguish the original author's statements from Doxxly's catalogue description, and should cite the original edition when using the material in academic, professional or published work.

Key topics and reader takeaways

The catalogue connects this document with Ancient DNA, Directional selection, Human evolution, Population genetics, West Eurasia, Polygenic adaptation, Genome-wide association studies, Archaeogenetics. These terms offer practical entry points rather than a substitute for close reading. Start by identifying the definitions, examples, arguments, tables or visual explanations attached to each topic. Then note where the creator draws connections, makes distinctions or moves from background material to a conclusion, recommendation, calculation or narrative development.

A useful reading method is to separate description from interpretation. First record what the item actually presents: names, headings, figures, formulas, quotations or stated propositions. Next consider why those details matter within the whole work. Finally compare them with another source from the same subject area. This process helps turn a long digital file into evidence that can be evaluated rather than merely skimmed.

Table of contents and reading guide

The following is an editorial content guide based on the verified catalogue topics. It is not represented as the original author's formal table of contents. Use it to organise notes while moving through the 36 pages in the player.

  1. Part 1

    Ancient DNA

    Use this section of the reading guide to identify how Ancient DNA appears in Ancient DNA reveals pervasive directional selection across West Eurasia, note important terms, and connect the subject to the wider purpose of the work.

  2. Part 2

    Directional selection

    Use this section of the reading guide to identify how Directional selection appears in Ancient DNA reveals pervasive directional selection across West Eurasia, note important terms, and connect the subject to the wider purpose of the work.

  3. Part 3

    Human evolution

    Use this section of the reading guide to identify how Human evolution appears in Ancient DNA reveals pervasive directional selection across West Eurasia, note important terms, and connect the subject to the wider purpose of the work.

  4. Part 4

    Population genetics

    Use this section of the reading guide to identify how Population genetics appears in Ancient DNA reveals pervasive directional selection across West Eurasia, note important terms, and connect the subject to the wider purpose of the work.

  5. Part 5

    West Eurasia

    Use this section of the reading guide to identify how West Eurasia appears in Ancient DNA reveals pervasive directional selection across West Eurasia, note important terms, and connect the subject to the wider purpose of the work.

  6. Part 6

    Polygenic adaptation

    Use this section of the reading guide to identify how Polygenic adaptation appears in Ancient DNA reveals pervasive directional selection across West Eurasia, note important terms, and connect the subject to the wider purpose of the work.

  7. Part 7

    Genome-wide association studies

    Use this section of the reading guide to identify how Genome-wide association studies appears in Ancient DNA reveals pervasive directional selection across West Eurasia, note important terms, and connect the subject to the wider purpose of the work.

  8. Part 8

    Archaeogenetics

    Use this section of the reading guide to identify how Archaeogenetics appears in Ancient DNA reveals pervasive directional selection across West Eurasia, note important terms, and connect the subject to the wider purpose of the work.

When a heading in the original differs from this guide, follow the original heading in your citation. Page numbers can also vary between editions. Doxxly's direct page field is useful for navigation, but formal references should include the title, creator, source and edition details available in the file itself.

About Vesna F. Oakfield

Doxxly credits Vesna F. Oakfield as the author or creator of Ancient DNA reveals pervasive directional selection across West Eurasia. This page does not add an unverified biography, job title or institutional affiliation. The attribution is preserved so readers can distinguish the creator from the hosting library and follow the original record when an external author link is available.

Author context matters because a work reflects a particular purpose, audience and point of view. Before relying on the material, consider what the creator set out to explain, when the source was produced and whether later research or revised guidance should also be consulted. For literature and philosophy, compare translations and editions; for technical material, check whether a newer version has superseded the one shown here.

About the source organisation

The recorded source for this item is reich.hms.harvard.edu. Doxxly provides access and discovery tools but does not claim to be the original publisher unless explicitly stated. Source links, creator attribution and the recorded Publicly available file collected for editorial review licence are displayed to help readers assess provenance and permitted use.

Provenance is especially important for reports, standards, presentations and working spreadsheets. Check the original source for amendments, supporting files and current policy. A trustworthy research workflow keeps the source URL with any notes or downloaded copy, verifies important figures against the original, and avoids treating an archived document as automatically current.

Visit reich.hms.harvard.edu

How to read and use this document

Begin with the title and opening pages, then use the guide above to decide whether to read sequentially or move to a particular topic. The Doxxly player provides previous and next navigation, a direct page field, zoom steps, fit-to-width and fit-to-page modes, rotation and fullscreen. These controls are useful when a page contains small type, a wide table, an image, a scanned layout or a dense sequence of notes.

The first five pages are open to every visitor. A plan unlocks the complete item and full-document search. Search is best used for distinctive names, technical terms or short phrases; broad words can produce too many possible contexts. When you find a useful passage, read the surrounding pages so that a sentence, chart or formula is not separated from its explanation.

If downloading is enabled, members can obtain the original PDF file. Keep the original format when formulas, slide sequencing, pagination or layout carries meaning. Respect the Publicly available file collected for editorial review terms displayed on this page, and use the rights-report link if you believe the attribution or availability needs review.

Research questions to consider

As you read Ancient DNA reveals pervasive directional selection across West Eurasia, ask which audience the creator addresses, what problem or idea organises the work, what evidence is presented, and which assumptions remain unstated. Note whether important terms are defined consistently and whether examples represent a broader pattern or a narrow case. For quantitative material, record units, time periods, formulas and data sources. For interpretive material, separate the creator's language from your own summary.

A second set of questions concerns relevance today. Which parts remain durable, which depend on their historical or institutional setting, and what would need confirmation from a newer source? Following the tags and the Science & Engineering collection can reveal useful comparisons. Agreement between sources can strengthen confidence, while disagreement can point to changing definitions, methods, evidence or priorities.

Finally, decide what you need from the item: background knowledge, a quotation, a worked example, a model, a visual explanation or a starting bibliography. That purpose should guide the depth of reading and the notes you keep. Doxxly helps locate and navigate the source; careful interpretation remains the reader's responsibility.

Explore more Science & Engineering resources

Frequently asked questions about Ancient DNA reveals pervasive directional selection across West Eurasia

Practical answers about reading, downloading and understanding this document on Doxxly.

More in Science & Engineering