Xinjun Zhang

Negative selection on deleterious mutations plays a key role in shaping human genetic variation, yet the dominance effects of these mutations remain poorly understood because existing statistical methods often cannot distinguish dominance effects from the overall selective effects. In this work, we take a fundamentally different approach by leveraging the distribution of Neanderthal ancestry across the human genome. Simulations show that recessive deleterious mutations can increase archaic introgressed ancestry through heterosis in the absence of positive selection, in contrast to the depletion expected under additive effects. We use this signal to develop DominL , a machine learning classifier trained on simulations of human demography with Neanderthal introgression to identify megabase-scale genomic windows enriched with recessive mutations. DominL demonstrates robust accuracy, with particularly high power in exon-dense regions. Applied to 7 non-African populations from the 1000 Genomes Project, DominL identifies approximately 3-9% of the genome as enriched for recessive mutations, with most high-confidence regions shared across populations. Predicted regions show patterns consistent with expected signatures from recessive mutations, including weakened background selection, depletion of runs of homozygosity, and enrichment of non-additive trait-associated variants. These regions also contain genes associated with metabolic and immune-related functions.

Hybridization is increasingly recognized as a widespread evolutionary process across animals, plants, and fungi, and plays a key role in driving speciation, shaping genetic diversity, and facilitating adaptation. Initial hybrid genomes are rarely observed empirically, making it challenging to observe in real time how competing forces like purifying selection and adaptive introgression interact to shape post-hybridization genomes. Here, we use replicated experimental hybrid populations between Drosophila simulans and D. sechellia to track directly genome-wide evolutionary dynamics across approximately 45 generations post-hybridization. We show that hybrid populations with primarily D. simulans ancestry rapidly recover wild-type levels of fertility and fecundity, whereas populations carrying predominantly D. sechellia ancestry experience persistent fitness reductions and frequent extinction. Genome-wide ancestry trajectories show pervasive purifying selection against minority-ancestry alleles, leading to rapid regression toward majority ancestry across the genome, with stronger purging on the X chromosome than autosomes. Following the initial purging, neutral minority ancestry stabilized at low frequencies across the genome. In striking contrast, we observe repeated increases in D. sechellia ancestry on chromosome 4 and multiple regions on chromosome 3R within D. simulans -majority populations, providing evidence for adaptive introgression. High-density sequencing revealed that these regions were repeatedly retained across independently evolving populations despite near-complete genome-wide purging. Simulations support a model in which recombination rapidly uncouples beneficial introgressed alleles from linked incompatibility-associated alleles during the initial generations, allowing positive selection to act. Our temporal allele-frequency analyses suggest that hybrid genome stabilization of the introgressed regions is highly repeatable but not identical, yielding parallel outcomes across replicate populations. We establish a tractable experimental framework to directly observe how hybrid genomes reorganize in real time, and demonstrate how purifying selection, positive selection, recombination, and stochastic processes interact to restructure hybrid genomes shortly after admixture.

Ghost introgression from unsampled hominin lineages has emerged as an increasingly important component of human evolutionary history. Recent studies suggest that deeply divergent hominin lineages may have contributed ancestry either directly to modern humans or indirectly through Denisovan introgression, while inference remains difficult due to few reference genomes, weak signal, and uncertainty in reconstructing deep genealogies. Here we show analytically and through simulations that Denisovan-mediated superarchaic introgression produces predictable shifts in local coalescent depth that can be approximated by scalable summary statistics, particularly pairwise sequence divergence, suggesting that substantial information regarding deeply divergent ancestry is preserved in sequence variations without explicit reconstruction of genealogies. Leveraging this insight, we develop DEEP ( D eep ancestry E stimation through E fficient P roxies), an ARG-free neural-network framework for identifying candidate regions of superarchaic ancestry. DEEP retains detectable power at low false positive rates across a broad range of demographic parameter space, remains scalable and recovers signals from small sample sizes. Applying DEEP to Oceanians, Tibetans, and Han Chinese, we identify approximately 0.4-0.6% of genomic windows with evidence of superarchaic ancestry. Candidate regions show both substantial overlap and notable heterogeneity across populations, with repeated enrichment near the HLA locus across all populations, suggesting immune-related regions recurrently retain deeply divergent ancestry.

Understanding the genetic architecture of complex traits in admixed populations remains challenging due to heterogeneous genetic backgrounds and demographic histories. Mischaracterizing admixture can bias genetic association estimates and limit the generalizability of biomedical findings. Here, we systematically evaluate how evolutionary forces—including admixture, natural selection, and demographic history—jointly shape complex trait architecture and influence genome-wide association study (GWAS) outcomes using a simulation-based framework complemented by empirical analyses. We model five human admixture scenarios and vary the correlation between causal variant effect sizes and selection coefficients to reflect different trait–fitness relationships. This framework enables simulation of complex trait phenotypes with environmental variance, allowing comprehensive assessment of GWAS power and fine-mapping precision across evolutionary contexts. We find that GWAS power is strongly modulated by both genetic architecture and demographic history. Traits with weak coupling between fitness and effect size, such as anthropometric traits, exhibit higher GWAS power than traits under stronger negative selection, including early-onset diseases. Because rare variants contribute substantially to heritability yet are poorly captured by GWAS, bottlenecked populations with fewer rare variants show enhanced power. Despite large differences in GWAS power, fine-mapping precision remains relatively consistent across traits and populations, improving primarily in regions of high recombination. Empirical analyses of diverse cohorts, the All of Us Research Program, support these patterns. Our findings highlight how evolutionary and demographic forces shape the genetic basis of complex traits in admixed populations and underscore the need for tailored study designs to improve GWAS accuracy and fine-mapping performance in diverse cohorts.

Scaling is a common practice in population genetic simulations to increase computational efficiency. However, few studies systematically examine the effects of scaling on diversity estimates and the comparability of scaled results to unscaled simulations and empirical data. We investigate the effects of scaling in two species, modern humans and Drosophila melanogaster. These species have stark differences in population size and generation time, necessitating moderate-to-no scaling for humans and dramatic scaling for Drosophila. We determine how coalescence, runtime, memory, estimates of diversity, the site frequency spectra, and linkage disequilibrium are influenced by scaling. We also examine the impact of simulated segment length and burn-in time on these metrics. Our results demonstrate that while computational efficiency improves with scaling, large scaling factors distort genetic diversity and dynamics between genetic variants, resulting in deviations from the intended model and empirical observations. Specifically, strongly scaled simulations may experience stronger negative selection on deleterious mutations, which amplifies background selection and purges linked mutations, leaving only rare strongly deleterious variants in the final population. We additionally show that a heuristic burn-in length of 10N generations is often insufficient for full coalescence in both models and alters expected linkage disequilibrium patterns. Finally, we provide considerations for conducting scaled simulations and offer potential strategies for the mitigation of scaling effects. For most nonmodel species simulations, we advocate for a bespoke scaling strategy drawn from these use cases.

Negative natural selection on deleterious mutations plays a key role in shaping human genetic variation. Understanding the dominance of deleterious mutations is critical as it can fundamentally impact the rate and efficiency of natural selection, the magnitude of inbreeding depression, and the prevalence and evolution of genetic diseases. Despite its inarguable importance, the dominance effects of mutations remain poorly understood in humans, primarily because existing statistical methods cannot distinguish them from the overall selective effects of mutations. In this work, we take a fundamentally different approach to infer dominance by leveraging the distribution of Neanderthal ancestry across the human genome. We show through simulations that recessive deleterious mutations lead to an increase in archaic introgressed ancestry in the absence of positive selection, contrary to what is expected when deleterious mutations are additive. Leveraging this unique pattern, we develop a machine learning classifier to infer dominance in genomic windows at a megabase resolution, trained on simulations of a human demographic model with Neanderthal introgression using fully recessive or additive mutations. Our method demonstrates robust accuracy at detecting genomic windows containing recessive deleterious mutations, with particularly high power in exon-dense regions. When applied to the non-African populations from the 1000 Genomes Project, we find that approximately 3-9% of the human genome is enriched for recessive mutations with most recessive regions shared across human populations. Furthermore, our method reveals that recessive deleterious mutations are not evenly distributed across the genome: regions enriched for recessive mutations are significantly depleted of haploinsufficient genes and runs of homozygosity, and are enriched with non-additive variants associated with complex traits. Overall, our Neanderthal ancestry-based approach reveals the presence of recessive deleterious mutations in the human genome and suggests that these mutations are found in regions containing genes associated with metabolism and immune-related traits.

Scaling is a common practice in population genetic simulations to increase computational efficiency. However, there exists a dearth of standardized guidelines for best practices. Few studies have examined the effects of scaling on diversity and whether the results are directly comparable to unscaled and empirical data. We examine the effects of scaling in two model populations, modern humans and Drosophila melanogaster . The reason is twofold: 1) due to the substantial difference in population sizes and generation times, human populations require moderate-to-no scaling, while more dramatic scaling is required for Drosophila ; and 2) model populations have empirical data for comparison. We determine whether coalescence, runtime, memory, estimates of diversity, the site frequency spectra, and deleterious variation are affected by scaling. We also explore the effect of varying the simulated segment length and burn-in times. We find that the typical 10N generation burn-in is often not sufficient for full coalescence to occur in human or Drosophila simulations. As expected, memory and runtime increase as the scaling coefficient decreases and the length of the simulated segment increases. We show that simulating larger segments in humans is preferable, as it produces a smaller variance in diversity estimates. Conversely, in Drosophila it is preferable to simulate smaller segments and concatenate them into full genome for achieving comparable levels of diversity to empirical data. We find that aggressive scaling leads to stronger negative selection and ultimately amplifies the strength of background selection on flanking variation. Author Summary: Scaling is a common approach to make population genetic simulations more computationally tractable. However, the implications of scaling and best practices for scaling are still unknown. This study highlights the importance of carefully considering scaling practices for forward-in-time population genetics simulations. We provide insights about the trade-offs between computational efficiency and accuracy of scaled simulations relative to empirical data, in human and Drosophila . We achieved this by varying the species demographic model; the method of coalescence; the simulated genomic element length; and the scaling factor. For each combination of parameters genetic diversity was quantified and computational was efficiency tracked. Our findings suggest that when simulating populations, such as humans, where moderate scaling is required, one should simulate larger genomic segments for more accurate measures of diversity. Scaling seems to cause an inflation of diversity in human simulations relative to empirical data. On the other hand, in populations where more aggressive scaling is required, such as Drosophila , simulating smaller segments is advantageous. The scaling factor increases substantially in Drosophila studies, and the simulated data experiences a drastic drop in diversity, relative to empirical data, and an increased effect of purifying selection.

Adaptive introgression (AI) facilitates local adaptation in a wide range of species. Many state-of-the-art methods detect AI with ad-hoc approaches that identify summary statistic outliers or intersect scans for positive selection with scans for introgressed genomic regions. Although widely used, approaches intersecting outliers are vulnerable to a high false-negative rate as the power of different methods varies, especially for complex introgression events. Moreover, population genetic processes unrelated to AI, such as background selection or heterosis, may create similar genomic signals to AI, compromising the reliability of methods that rely on neutral null distributions. In recent years, machine learning (ML) methods have been increasingly applied to population genetic questions. Here, we present a ML-based method called MaLAdapt for identifying AI loci from genome-wide sequencing data. Using an Extra-Trees Classifier algorithm, our method combines information from a large number of biologically meaningful summary statistics to capture a powerful composite signature of AI across the genome. In contrast to existing methods, MaLAdapt is especially well-powered to detect AI with mild beneficial effects, including selection on standing archaic variation, and is robust to non-AI selective sweeps, heterosis from deleterious mutations, and demographic misspecification. Further, MaLAdapt outperforms existing methods for detecting AI based on the analysis of simulated data and on a validation of empirical signals through visual inspection of haplotype patterns. We apply MaLAdapt to the 1000 Genomes Project human genomic data, and discover novel AI candidate regions in non-African populations, including genes that are enriched in functionally important biological pathways regulating metabolism and immune responses.

Adaptive introgression (AI) facilitates local adaptation in a wide range of species. Many state-of-the-art methods detect AI with ad-hoc approaches that identify summary statistic outliers or intersect scans for positive selection with scans for introgressed genomic regions. Although widely used, these outlier-based approaches are vulnerable to a high false-negative rate as the power of different methods vary, especially for complex introgression events. Moreover, population genetic processes unrelated to AI, such as background selection or heterosis, may create similar genomic signals as AI, compromising the reliability of methods that rely on neutral null distributions. In recent years, machine learning (ML) methods have been increasingly applied to population genetic questions. Here, we present an ML-based method called MaLAdapt for identifying AI loci from genome-wide sequencing data. Using an Extra-Trees Classifier algorithm, our method combines information from a large number of biologically meaningful summary statistics to capture a powerful composite signature of AI across the genome. In contrast to existing methods, MaLAdapt is especially well-powered to detect AI with mild beneficial effects, including selection on standing archaic variation, and is robust to non-AI selection sweeps, heterosis, and demographic misspecifications. Further, MaLAdapt outperforms existing methods for detecting AI based on the analysis of simulated data and on a validation of empirical signals through visual impaction of haplotype patterns. We apply MaLAdapt to the 1000 Genomes Project human genomic data, and discover novel AI candidate regions in non-African populations, including genes that are enriched in functionally important biological pathways regulating metabolism and immune responses.

The apportionment of human genetic diversity within and between populations has been measured to understand human relatedness and demographic history. Likewise, the distribution of archaic ancestry in modern populations can be leveraged to better understand the interaction between our species and its archaic relatives. Resolving the interactions between modern and archaic human populations can be difficult, as archaic variants in modern populations have been shaped by genetic drift, bottlenecks and gene flow. Here, we investigate the distribution of archaic variation in Eurasian populations. We find that archaic ancestry coverage at the individual- and population-level present distinct patterns in modern human populations: South Asians have nearly twice the number of population-unique archaic alleles compared with Europeans or East Asians, indicating that these populations experienced differing demographic and archaic admixture events. We confirm previous observations that East Asian individuals have more Neanderthal ancestry than European individuals, but surprisingly, when we compare the number of single nucleotide polymorphisms with archaic alleles found across a population, Europeans have more Neanderthal ancestry than East Asians. We compare these results to simulated models and conclude that these patterns are consistent with multiple admixture events between modern humans and Neanderthals. This article is part of the theme issue ‘Celebrating 50 years since Lewontin's apportionment of human diversity’.