Mutation Rate Variation is a Primary Determinant of the Distribution of Allele Frequencies in Humans

Characterizing selection on complex traits through conditional frequency spectra

Natural selection on complex traits is difficult to study in part due to the ascertainment inherent to genome-wide association studies (GWAS). The power to detect a trait-associated variant in GWAS is a function of its frequency and effect size - but for traits under selection, the effect size of a variant determines the strength of selection against it, constraining its frequency. Recognizing the biases inherent to GWAS ascertainment, we propose studying the joint distribution of allele frequencies across populations, conditional on the frequencies in the GWAS cohort. Before considering these conditional frequency spectra, we first characterized the impact of selection and non-equilibrium demography on allele frequency dynamics forwards and backwards in time. We then used these results to understand conditional frequency spectra under realistic human demography. Finally, we investigated empirical conditional frequency spectra for GWAS variants associated with 106 complex traits, finding compelling evidence for either stabilizing or purifying selection. Our results provide insights into polygenic score portability and other properties of variants ascertained with GWAS, highlighting the utility of conditional frequency spectra.

Diverse genetic conflicts mediated by molecular mimicry and computational approaches to detect them

In genetic conflicts between intergenomic and selfish elements, driver and killer elements achieve biased survival, replication, or transmission over sensitive and targeted elements through a wide range of molecular mechanisms, including mimicry. Driving mechanisms manifest at all organismal levels, from the biased propagation of individual genes, as demonstrated by transposable elements, to the biased transmission of genomes, as illustrated by viruses, to the biased transmission of cell lineages, as in cancer. Targeted genomes are vulnerable to molecular mimicry through the conserved motifs they use for their own signaling and regulation. Mimicking these motifs enables an intergenomic or selfish element to control core target processes, and can occur at the sequence, structure, or functional level. Molecular mimicry was first appreciated as an important phenomenon more than twenty years ago. Modern genomics technologies, databases, and machine learning approaches offer tremendous potential to study the distribution of molecular mimicry across genetic conflicts in nature. Here, we explore the theoretical expectations for molecular mimicry between conflicting genomes, the trends in molecular mimicry mechanisms across known genetic conflicts, and outline how new examples can be gleaned from population genomic datasets. We discuss how mimics involving short sequence-based motifs or gene duplications can evolve convergently from new mutations. Whereas, processes that involve divergent domains or fully-folded structures occur among genomes by horizontal gene transfer. These trends are largely based on a small number of organisms and should be reevaluated in a general, phylogenetically independent framework. Currently, publicly available databases can be mined for genotypes driving non-Mendelian inheritance patterns, epistatic interactions, and convergent protein structures. A subset of these conflicting elements may be molecular mimics. We propose approaches for detecting genetic conflict and molecular mimicry from these datasets.

Digital Image Processing to Detect Adaptive Evolution

In recent years, advances in image processing and machine learning have fueled a paradigm shift in detecting genomic regions under natural selection. Early machine learning techniques employed population-genetic summary statistics as features, which focus on specific genomic patterns expected by adaptive and neutral processes. Though such engineered features are important when training data are limited, the ease at which simulated data can now be generated has led to the recent development of approaches that take in image representations of haplotype alignments and automatically extract important features using convolutional neural networks. Digital image processing methods termed α-molecules are a class of techniques for multiscale representation of objects that can extract a diverse set of features from images. One such α-molecule method, termed wavelet decomposition, lends greater control over high-frequency components of images. Another α-molecule method, termed curvelet decomposition, is an extension of the wavelet concept that considers events occurring along curves within images. We show that application of these α-molecule techniques to extract features from image representations of haplotype alignments yield high true positive rate and accuracy to detect hard and soft selective sweep signatures from genomic data with both linear and nonlinear machine learning classifiers. Moreover, we find that such models are easy to visualize and interpret, with performance rivaling those of contemporary deep learning approaches for detecting sweeps.

Reconstruction of the human amylase locus reveals ancient duplications seeding modern-day variation

Previous studies suggested that the copy number of the human salivary amylase gene, AMY1, correlates with starch-rich diets. However, evolutionary analyses are hampered by the absence of accurate, sequence-resolved haplotype variation maps. We identified 30 structurally distinct haplotypes at nucleotide resolution among 98 present-day humans, revealing that the coding sequences of AMY1 copies are evolving under negative selection. Genomic analyses of these haplotypes in archaic hominins and ancient human genomes suggest that a common three-copy haplotype, dating as far back as 800,000 years ago, has seeded rapidly evolving rearrangements through recurrent nonallelic homologous recombination. Additionally, haplotypes with more than three AMY1 copies have significantly increased in frequency among European farmers over the past 4000 years, potentially as an adaptive response to increased starch digestion.

Estimation of demography and mutation rates from one million haploid genomes

As genetic sequencing costs have plummeted, datasets with sizes previously un-thinkable have begun to appear. Such datasets present new opportunities to learn about evolutionary history, particularly via rare alleles that record the very recent past. However, beyond the computational challenges inherent in the analysis of many large-scale datasets, large population-genetic datasets present theoretical problems. In particular, the majority of population-genetic tools require the assumption that each mutant allele in the sample is the result of a single mutation (the "infinite sites" assumption), which is violated in large samples. Here, we present DR EVIL, a method for estimating mutation rates and recent demographic history from very large samples. DR EVIL avoids the infinite-sites assumption by using a diffusion approximation to a branching-process model with recurrent mutation. The branching-process approach limits the method to rare alleles, but, along with recent results, renders tractable likelihoods with recurrent mutation. We show that DR EVIL performs well in simulations and apply it to rare-variant data from a million haploid samples, identifying a signal of mutation-rate heterogeneity within commonly analyzed classes and predicting that in modern sample sizes, most rare variants at sites with high mutation rates represent the descendants of multiple mutation events.

Context-adjusted proportion of singletons (CAPS): a novel metric for assessing negative selection in the human genome

Interpretation of genetic variants remains challenging, partly due to the lack of well-established ways of determining the potential pathogenicity of genetic variation, especially for understudied classes of variants. Addressing this, population genetics methods offer a practical solution by evaluating variant effects through human population distributions. Negative selection influences the ratio of singleton variants and can serve as a proxy for deleteriousness, as exemplified by the Mutability-Adjusted Proportion of Singletons (MAPS) metric. However, MAPS is sensitive to the calibration of the singletons-by-mutability linear model, which results in biased estimates for certain variant classes. Building up on the methodology used in MAPS, we introduce the Context-Adjusted Proportion of Singletons (CAPS) metric for assessing negative selection in the human genome. CAPS produces corrected estimates with more accurate confidence intervals by eliminating the mutability layer in the model. Retaining the advantageous features of MAPS, CAPS emerges as a robust and reliable tool. We believe that CAPS has the potential to enhance the identification of new disease-variant associations in clinical and research settings, offering improved accuracy in assessing negative selection for diverse SNV classes.

Latent mutations in the ancestries of alleles under selection

We consider a single genetic locus with two alleles A1 and A2 in a large haploid population. The locus is subject to selection and two-way, or recurrent, mutation. Assuming the allele frequencies follow a Wright-Fisher diffusion and have reached stationarity, we describe the asymptotic behaviors of the conditional gene genealogy and the latent mutations of a sample with known allele counts, when the count n1 of allele A1 is fixed, and when either or both the sample size n and the selection strength |α| tend to infinity. Our study extends previous work under neutrality to the case of non-neutral rare alleles, asserting that when selection is not too strong relative to the sample size, even if it is strongly positive or strongly negative in the usual sense (α→-∞ or α→+∞), the number of latent mutations of the n1 copies of allele A1 follows the same distribution as the number of alleles in the Ewens sampling formula. On the other hand, very strong positive selection relative to the sample size leads to neutral gene genealogies with a single ancient latent mutation. We also demonstrate robustness of our asymptotic results against changing population sizes, when one of |α| or n is large.

Conditional frequency spectra as a tool for studying selection on complex traits in biobanks

Natural selection on complex traits is difficult to study in part due to the ascertainment inherent to genome-wide association studies (GWAS). The power to detect a trait-associated variant in GWAS is a function of frequency and effect size - but for traits under selection, the effect size of a variant determines the strength of selection against it, constraining its frequency. To account for GWAS ascertainment, we propose studying the joint distribution of allele frequencies across populations, conditional on the frequencies in the GWAS cohort. Before considering these conditional frequency spectra, we first characterized the impact of selection and non-equilibrium demography on allele frequency dynamics forwards and backwards in time. We then used these results to understand conditional frequency spectra under realistic human demography. Finally, we investigated empirical conditional frequency spectra for GWAS variants associated with 106 complex traits, finding compelling evidence for either stabilizing or purifying selection. Our results provide insight into polygenic score portability and other properties of variants ascertained with GWAS, highlighting the utility of conditional frequency spectra.

Bayesian estimation of gene constraint from an evolutionary model with gene features

Measures of selective constraint on genes have been used for many applications including clinical interpretation of rare coding variants, disease gene discovery, and studies of genome evolution. However, widely-used metrics are severely underpowered at detecting constraint for the shortest ∼25% of genes, potentially causing important pathogenic mutations to be overlooked. We developed a framework combining a population genetics model with machine learning on gene features to enable accurate inference of an interpretable constraint metric, shet. Our estimates outperform existing metrics for prioritizing genes important for cell essentiality, human disease, and other phenotypes, especially for short genes. Our new estimates of selective constraint should have wide utility for characterizing genes relevant to human disease. Finally, our inference framework, GeneBayes, provides a flexible platform that can improve estimation of many gene-level properties, such as rare variant burden or gene expression differences.

The Promise of Inferring the Past Using the Ancestral Recombination Graph

The ancestral recombination graph (ARG) is a structure that represents the history of coalescent and recombination events connecting a set of sequences (Hudson RR. In: Futuyma D, Antonovics J, editors. Gene genealogies and the coalescent process. In: Oxford Surveys in Evolutionary Biology; 1991. p. 1 to 44.). The full ARG can be represented as a set of genealogical trees at every locus in the genome, annotated with recombination events that change the topology of the trees between adjacent loci and the mutations that occurred along the branches of those trees (Griffiths RC, Marjoram P. An ancestral recombination graph. In: Donnelly P, Tavare S, editors. Progress in population genetics and human evolution. Springer; 1997. p. 257 to 270.). Valuable insights can be gained into past evolutionary processes, such as demographic events or the influence of natural selection, by studying the ARG. It is regarded as the "holy grail" of population genetics (Hubisz M, Siepel A. Inference of ancestral recombination graphs using ARGweaver. In: Dutheil JY, editors. Statistical population genomics. New York, NY: Springer US; 2020. p. 231-266.) since it encodes the processes that generate all patterns of allelic and haplotypic variation from which all commonly used summary statistics in population genetic research (e.g. heterozygosity and linkage disequilibrium) can be derived. Many previous evolutionary inferences relied on summary statistics extracted from the genotype matrix. Evolutionary inferences using the ARG represent a significant advancement as the ARG is a representation of the evolutionary history of a sample that shows the past history of recombination, coalescence, and mutation events across a particular sequence. This representation in theory contains as much information, if not more, than the combination of all independent summary statistics that could be derived from the genotype matrix. Consistent with this idea, some of the first ARG-based analyses have proven to be more powerful than summary statistic-based analyses (Speidel L, Forest M, Shi S, Myers SR. A method for genome-wide genealogy estimation for thousands of samples. Nat Genet. 2019:51(9):1321 to 1329.; Stern AJ, Wilton PR, Nielsen R. An approximate full-likelihood method for inferring selection and allele frequency trajectories from DNA sequence data. PLoS Genet. 2019:15(9):e1008384.; Hubisz MJ, Williams AL, Siepel A. Mapping gene flow between ancient hominins through demography-aware inference of the ancestral recombination graph. PLoS Genet. 2020:16(8):e1008895.; Fan C, Mancuso N, Chiang CWK. A genealogical estimate of genetic relationships. Am J Hum Genet. 2022:109(5):812-824.; Fan C, Cahoon JL, Dinh BL, Ortega-Del Vecchyo D, Huber C, Edge MD, Mancuso N, Chiang CWK. A likelihood-based framework for demographic inference from genealogical trees. bioRxiv. 2023.10.10.561787. 2023.; Hejase HA, Mo Z, Campagna L, Siepel A. A deep-learning approach for inference of selective sweeps from the ancestral recombination graph. Mol Biol Evol. 2022:39(1):msab332.; Link V, Schraiber JG, Fan C, Dinh B, Mancuso N, Chiang CWK, Edge MD. Tree-based QTL mapping with expected local genetic relatedness matrices. bioRxiv. 2023.04.07.536093. 2023.; Zhang BC, Biddanda A, Gunnarsson ÁF, Cooper F, Palamara PF. Biobank-scale inference of ancestral recombination graphs enables genealogical analysis of complex traits. Nat Genet. 2023:55(5):768-776.). As such, there has been significant interest in the field to investigate 2 main problems related to the ARG: (i) How can we estimate the ARG based on genomic data, and (ii) how can we extract information of past evolutionary processes from the ARG? In this perspective, we highlight 3 topics that pertain to these main issues: The development of computational innovations that enable the estimation of the ARG; remaining challenges in estimating the ARG; and methodological advances for deducing evolutionary forces and mechanisms using the ARG. This perspective serves to introduce the readers to the types of questions that can be explored using the ARG and to highlight some of the most pressing issues that must be addressed in order to make ARG-based inference an indispensable tool for evolutionary research.

Quantifying negative selection in human 3' UTRs uncovers constrained targets of RNA-binding proteins

Many non-coding variants associated with phenotypes occur in 3' untranslated regions (3' UTRs), and may affect interactions with RNA-binding proteins (RBPs) to regulate gene expression post-transcriptionally. However, identifying functional 3' UTR variants has proven difficult. We use allele frequencies from the Genome Aggregation Database (gnomAD) to identify classes of 3' UTR variants under strong negative selection in humans. We develop intergenic mutability-adjusted proportion singleton (iMAPS), a generalized measure related to MAPS, to quantify negative selection in non-coding regions. This approach, in conjunction with in vitro and in vivo binding data, identifies precise RBP binding sites, miRNA target sites, and polyadenylation signals (PASs) under strong selection. For each class of sites, we identify thousands of gnomAD variants under selection comparable to missense coding variants, and find that sites in core 3' UTR regions upstream of the most-used PAS are under strongest selection. Together, this work improves our understanding of selection on human genes and validates approaches for interpreting genetic variants in human 3' UTRs.

A mutation rate model at the basepair resolution identifies the mutagenic effect of polymerase III transcription

De novo mutations occur at substantially different rates depending on genomic location, sequence context and DNA strand. The success of methods to estimate selection intensity, infer demographic history and map rare disease genes, depends strongly on assumptions about the local mutation rate. Here we present Roulette, a genome-wide mutation rate model at basepair resolution that incorporates known determinants of local mutation rate. Roulette is shown to be more accurate than existing models. We use Roulette to refine the estimates of population growth within Europe by incorporating the full range of human mutation rates. The analysis of significant deviations from the model predictions revealed a tenfold increase in mutation rate in nearly all genes transcribed by polymerase III (Pol III), suggesting a new mutagenic mechanism. We also detected an elevated mutation rate within transcription factor binding sites restricted to sites actively used in testis and residing in promoters.

Scaling the discrete-time Wright-Fisher model to biobank-scale datasets

The discrete-time Wright-Fisher (DTWF) model and its diffusion limit are central to population genetics. These models can describe the forward-in-time evolution of allele frequencies in a population resulting from genetic drift, mutation, and selection. Computing likelihoods under the diffusion process is feasible, but the diffusion approximation breaks down for large samples or in the presence of strong selection. Existing methods for computing likelihoods under the DTWF model do not scale to current exome sequencing sample sizes in the hundreds of thousands. Here, we present a scalable algorithm that approximates the DTWF model with provably bounded error. Our approach relies on two key observations about the DTWF model. The first is that transition probabilities under the model are approximately sparse. The second is that transition distributions for similar starting allele frequencies are extremely close as distributions. Together, these observations enable approximate matrix-vector multiplication in linear (as opposed to the usual quadratic) time. We prove similar properties for Hypergeometric distributions, enabling fast computation of likelihoods for subsamples of the population. We show theoretically and in practice that this approximation is highly accurate and can scale to population sizes in the tens of millions, paving the way for rigorous biobank-scale inference. Finally, we use our results to estimate the impact of larger samples on estimating selection coefficients for loss-of-function variants. We find that increasing sample sizes beyond existing large exome sequencing cohorts will provide essentially no additional information except for genes with the most extreme fitness effects.

Tensor Decomposition-based Feature Extraction and Classification to Detect Natural Selection from Genomic Data

Inferences of adaptive events are important for learning about traits, such as human digestion of lactose after infancy and the rapid spread of viral variants. Early efforts toward identifying footprints of natural selection from genomic data involved development of summary statistic and likelihood methods. However, such techniques are grounded in simple patterns or theoretical models that limit the complexity of settings they can explore. Due to the renaissance in artificial intelligence, machine learning methods have taken center stage in recent efforts to detect natural selection, with strategies such as convolutional neural networks applied to images of haplotypes. Yet, limitations of such techniques include estimation of large numbers of model parameters under nonconvex settings and feature identification without regard to location within an image. An alternative approach is to use tensor decomposition to extract features from multidimensional data although preserving the latent structure of the data, and to feed these features to machine learning models. Here, we adopt this framework and present a novel approach termed T-REx, which extracts features from images of haplotypes across sampled individuals using tensor decomposition, and then makes predictions from these features using classical machine learning methods. As a proof of concept, we explore the performance of T-REx on simulated neutral and selective sweep scenarios and find that it has high power and accuracy to discriminate sweeps from neutrality, robustness to common technical hurdles, and easy visualization of feature importance. Therefore, T-REx is a powerful addition to the toolkit for detecting adaptive processes from genomic data.

Parsing variance by marker type: Testing biogeographic hypotheses and differential contribution of historical processes to population structure in a desert lizard

A fundamental goal of population genetic studies is to identify historical biogeographic patterns and understand the processes that generate them. However, localized demographic events can skew population genetic inference. Assessing populations with multiple types of genetic markers, each with unique mutation rates and responses to changes in population size, can help to identify potentially confounding population-specific demographic processes. Here, we compared population structure and connectivity inferred from microsatellites and restriction site-associated DNA loci among 17 populations of an arid-specialist lizard, the desert night lizard, Xantusia vigilis, in central California to test among historical processes structuring population genetic diversity. We found that both marker types yielded generally concordant insights into population genetic structure including a major phylogenetic break maintained between two populations separated by less than 10 km, suggesting that either marker type could be used to understand generalized demographic patterns across the region for management purposes. However, we also found that the effects of demography on marker discordance could be used to elucidate population histories and distinguish among competing biogeographic hypotheses. Our results suggest that comparisons of within-population diversity across marker types provide powerful opportunities for leveraging marker discordance, particularly for understanding the creation and maintenance of contact zones among clades.

Recurrent mutation in the ancestry of a rare variant

Recurrent mutation produces multiple copies of the same allele which may be co-segregating in a population. Yet, most analyses of allele-frequency or site-frequency spectra assume that all observed copies of an allele trace back to a single mutation. We develop a sampling theory for the number of latent mutations in the ancestry of a rare variant, specifically a variant observed in relatively small count in a large sample. Our results follow from the statistical independence of low-count mutations, which we show to hold for the standard neutral coalescent or diffusion model of population genetics as well as for more general coalescent trees. For populations of constant size, these counts are distributed like the number of alleles in the Ewens sampling formula. We develop a Poisson sampling model for populations of varying size and illustrate it using new results for site-frequency spectra in an exponentially growing population. We apply our model to a large data set of human SNPs and use it to explain dramatic differences in site-frequency spectra across the range of mutation rates in the human genome.

Bayesian estimation of gene constraint from an evolutionary model with gene features

Measures of selective constraint on genes have been used for many applications including clinical interpretation of rare coding variants, disease gene discovery, and studies of genome evolution. However, widely-used metrics are severely underpowered at detecting constraint for the shortest ~25% of genes, potentially causing important pathogenic mutations to be overlooked. We developed a framework combining a population genetics model with machine learning on gene features to enable accurate inference of an interpretable constraint metric, s het . Our estimates outperform existing metrics for prioritizing genes important for cell essentiality, human disease, and other phenotypes, especially for short genes. Our new estimates of selective constraint should have wide utility for characterizing genes relevant to human disease. Finally, our inference framework, GeneBayes, provides a flexible platform that can improve estimation of many gene-level properties, such as rare variant burden or gene expression differences.

Scaling the Discrete-time Wright Fisher model to biobank-scale datasets

The Discrete-Time Wright Fisher (DTWF) model and its large population diffusion limit are central to population genetics. These models describe the forward-in-time evolution of the frequency of an allele in a population and can include the fundamental forces of genetic drift, mutation, and selection. Computing like-lihoods under the diffusion process is feasible, but the diffusion approximation breaks down for large sample sizes or in the presence of strong selection. Unfortunately, existing methods for computing likelihoods under the DTWF model do not scale to current exome sequencing sample sizes in the hundreds of thousands. Here we present an algorithm that approximates the DTWF model with provably bounded error and runs in time linear in the size of the population. Our approach relies on two key observations about Binomial distributions. The first is that Binomial distributions are approximately sparse. The second is that Binomial distributions with similar success probabilities are extremely close as distributions, allowing us to approximate the DTWF Markov transition matrix as a very low rank matrix. Together, these observations enable matrix-vector multiplication in linear (as opposed to the usual quadratic) time. We prove similar properties for Hypergeometric distributions, enabling fast computation of likelihoods for subsamples of the population. We show theoretically and in practice that this approximation is highly accurate and can scale to population sizes in the billions, paving the way for rigorous biobank-scale population genetic inference. Finally, we use our results to estimate how increasing sample sizes will improve the estimation of selection coefficients acting on loss-of-function variants. We find that increasing sample sizes beyond existing large exome sequencing cohorts will provide essentially no additional information except for genes with the most extreme fitness effects.

The effect of mutation subtypes on the allele frequency spectrum and population genetics inference

Population genetics has adapted as technological advances in next-generation sequencing have resulted in an exponential increase of genetic data. A common approach to efficiently analyze genetic variation present in large sequencing data is through the allele frequency spectrum, defined as the distribution of allele frequencies in a sample. While the frequency spectrum serves to summarize patterns of genetic variation, it implicitly assumes mutation types (A→C vs C→T) as interchangeable. However, mutations of different types arise and spread due to spatial and temporal variation in forces such as mutation rate and biased gene conversion that result in heterogeneity in the distribution of allele frequencies across sites. In this work, we explore the impact of this simplification on multiple aspects of population genetic modeling. As a site's mutation rate is strongly affected by flanking nucleotides, we defined a mutation subtype by the base pair change and adjacent nucleotides (e.g. AAA→ATA) and systematically assessed the heterogeneity in the frequency spectrum across 96 distinct 3-mer mutation subtypes using n = 3556 whole-genome sequenced individuals of European ancestry. We observed substantial variation across the subtype-specific frequency spectra, with some of the variation being influenced by molecular factors previously identified for single base mutation types. Estimates of model parameters from demographic inference performed for each mutation subtype's AFS individually varied drastically across the 96 subtypes. In local patterns of variation, a combination of regional subtype composition and local genomic factors shaped the regional frequency spectrum across genomic regions. Our results illustrate how treating variants in large sequencing samples as interchangeable may confound population genetic frameworks and encourages us to consider the unique evolutionary mechanisms of analyzed polymorphisms.

Limited role of generation time changes in driving the evolution of the mutation spectrum in humans

Recent studies have suggested that the human germline mutation rate and spectrum evolve rapidly. Variation in generation time has been linked to these changes, though its contribution remains unclear. We develop a framework to characterize temporal changes in polymorphisms within and between populations, while controlling for the effects of natural selection and biased gene conversion. Application to the 1000 Genomes Project dataset reveals multiple independent changes that arose after the split of continental groups, including a previously reported, transient elevation in TCC>TTC mutations in Europeans and novel signals of divergence in C>Gand T>A mutation rates among population samples. We also find a significant difference between groups sampled in and outside of Africa in old T>C polymorphisms that predate the out-of-Africa migration. This surprising signal is driven by TpG>CpG mutations and stems in part from mis-polarized CpG transitions, which are more likely to undergo recurrent mutations. Finally, by relating the mutation spectrum of polymorphisms to parental age effects on de novo mutations, we show that plausible changes in the generation time cannot explain the patterns observed for different mutation types jointly. Thus, other factors - genetic modifiers or environmental exposures - must have had a non-negligible impact on the human mutation landscape.

Adaptive sequence divergence forged new neurodevelopmental enhancers in humans

Searches for the genetic underpinnings of uniquely human traits have focused on human-specific divergence in conserved genomic regions, which reflects adaptive modifications of existing functional elements. However, the study of conserved regions excludes functional elements that descended from previously neutral regions. Here, we demonstrate that the fastest-evolved regions of the human genome, which we term "human ancestor quickly evolved regions" (HAQERs), rapidly diverged in an episodic burst of directional positive selection prior to the human-Neanderthal split, before transitioning to constraint within hominins. HAQERs are enriched for bivalent chromatin states, particularly in gastrointestinal and neurodevelopmental tissues, and genetic variants linked to neurodevelopmental disease. We developed a multiplex, single-cell in vivo enhancer assay to discover that rapid sequence divergence in HAQERs generated hominin-unique enhancers in the developing cerebral cortex. We propose that a lack of pleiotropic constraints and elevated mutation rates poised HAQERs for rapid adaptation and subsequent susceptibility to disease.

Identifying rare variants inconsistent with identity-by-descent in population-scale whole-genome sequencing data

Analyses of genetic variation typically assume that rare variants within a population are inherited from a single common ancestral event identity-by-descent (IBD). However, there are genetic and technical processes through which rare variants in population genetic data may deviate from this simple evolutionary model, including recurrent mutations, gene conversions and genotyping error. All these processes can decrease the expected length of shared background haplotype surrounding a rare variant if that variant was inherited from a single event descending from a common ancestor. No method exists to computationally infer rare variants inconsistent with this simple model-denoted here as 'IBD-inconsistent'-using unphased population sequencing data.We hypothesized that the difference in shared haplotype background length can distinguish variants consistent and inconsistent with this simple IBD transmission population sequencing data without pedigree information. We implemented a Bayesian hierarchical model and used Gibbs sampling to estimate the posterior probability of IBD state for rare variants, using simulated recurrent mutations to demonstrate that our approach accurately distinguishes rare variants consistent and inconsistent with a simple IBD inheritance model.Applying our method to whole-genome sequencing data from 3,621 human individuals in the UK10K consortium, we found that IBD-inconsistent variants correlated with higher local mutation rates and genomic features like replication timing. Using a heuristic to categorize IBD-inconsistent variants as gene conversions, we found that potential gene conversions had expected properties such as enriched local GC content.By identifying IBD-inconsistent variants, we can better understand the spectrum of recent mutations in human populations, a source of genetic variation driving evolution and a key factor in understanding recent demographic history.

On the prospect of achieving accurate joint estimation of selection with population history

As both natural selection and population history can affect genome-wide patterns of variation, disentangling the contributions of each has remained as a major challenge in population genetics. We here discuss historical and recent progress towards this goal-highlighting theoretical and computational challenges that remain to be addressed, as well as inherent difficulties in dealing with model complexity and model violations-and offer thoughts on potentially fruitful next steps.

The contribution of evolutionarily volatile promoters to molecular phenotypes and human trait variation

BACKGROUND

Promoters are sites of transcription initiation that harbour a high concentration of phenotype-associated genetic variation. The evolutionary gain and loss of promoters between species (collectively, termed turnover) is pervasive across mammalian genomes and may play a prominent role in driving human phenotypic diversity.

RESULTS

We classified human promoters by their evolutionary history during the divergence of mouse and human lineages from a common ancestor. This defined conserved, human-inserted and mouse-deleted promoters, and a class of functional-turnover promoters that align between species but are only active in humans. We show that promoters of all evolutionary categories are hotspots for substitution and often, insertion mutations. Loci with a history of insertion and deletion continue that mode of evolution within contemporary humans. The presence of an evolutionary volatile promoter within a gene is associated with increased expression variance between individuals, but only in the case of human-inserted and mouse-deleted promoters does that correspond to an enrichment of promoter-proximal genetic effects. Despite the enrichment of these molecular quantitative trait loci (QTL) at evolutionarily volatile promoters, this does not translate into a corresponding enrichment of phenotypic traits mapping to these loci.

CONCLUSIONS

Promoter turnover is pervasive in the human genome, and these promoters are rich in molecularly quantifiable but phenotypically inconsequential variation in gene expression. However, since evolutionarily volatile promoters show evidence of selection, coupled with high mutation rates and enrichment of QTLs, this implicates them as a source of evolutionary innovation and phenotypic variation, albeit with a high background of selectively neutral expression variation.

De novo mutation rates at the single-mutation resolution in a human HBB gene-region associated with adaptation and genetic disease

Although it is known that the mutation rate varies across the genome, previous estimates were based on averaging across various numbers of positions. Here, we describe a method to measure the origination rates of target mutations at target base positions and apply it to a 6-bp region in the human hemoglobin subunit beta (HBB) gene and to the identical, paralogous hemoglobin subunit delta (HBD) region in sperm cells from both African and European donors. The HBB region of interest (ROI) includes the site of the hemoglobin S (HbS) mutation, which protects against malaria, is common in Africa, and has served as a classic example of adaptation by random mutation and natural selection. We found a significant correspondence between de novo mutation rates and past observations of alleles in carriers, showing that mutation rates vary substantially in a mutation-specific manner that contributes to the site frequency spectrum. We also found that the overall point mutation rate is significantly higher in Africans than in Europeans in the HBB region studied. Finally, the rate of the 20A→T mutation, called the “HbS mutation” when it appears in HBB, is significantly higher than expected from the genome-wide average for this mutation type. Nine instances were observed in the African HBB ROI, where it is of adaptive significance, representing at least three independent originations; no instances were observed elsewhere. Further studies will be needed to examine mutation rates at the single-mutation resolution across these and other loci and organisms and to uncover the molecular mechanisms responsible.

Mutation saturation for fitness effects at human CpG sites

Whole exome sequences have now been collected for millions of humans, with the related goals of identifying pathogenic mutations in patients and establishing reference repositories of data from unaffected individuals. As a result, we are approaching an important limit, in which datasets are large enough that, in the absence of natural selection, every highly mutable site will have experienced at least one mutation in the genealogical history of the sample. Here, we focus on CpG sites that are methylated in the germline and experience mutations to T at an elevated rate of ~10-7 per site per generation; considering synonymous mutations in a sample of 390,000 individuals, ~ 99 % of such CpG sites harbor a C/T polymorphism. Methylated CpG sites provide a natural mutation saturation experiment for fitness effects: as we show, at nt sample sizes, not seeing a non-synonymous polymorphism is indicative of strong selection against that mutation. We rely on this idea in order to directly identify a subset of CpG transitions that are likely to be highly deleterious, including ~27 % of possible loss-of-function mutations, and up to 20 % of possible missense mutations, depending on the type of functional site in which they occur. Unlike methylated CpGs, most mutation types, with rates on the order of 10-8 or 10-9, remain very far from saturation. We discuss what these findings imply for interpreting the potential clinical relevance of mutations from their presence or absence in reference databases and for inferences about the fitness effects of new mutations.

Human Prehistoric Demography Revealed by the Polymorphic Pattern of CpG Transitions

The prehistoric demography of human populations is an essential piece of information for illustrating our evolution. Despite its importance and the advancement of ancient DNA studies, our knowledge of human evolution is still limited, which is also the case for relatively recent population dynamics during and around the Holocene. Here, we inferred detailed demographic histories from 1 to 40 ka for 24 population samples using an improved model-flexible method with 36 million genome-wide noncoding CpG sites. Our results showed many population growth events that were likely due to the Neolithic Revolution (i.e., the shift from hunting and gathering to agriculture and settlement). Our results help to provide a clearer picture of human prehistoric demography, confirming the significant impact of agriculture on population expansion, and provide new hypotheses and directions for future research.

Sequencing of 53,831 diverse genomes from the NHLBI TOPMed Program

The Trans-Omics for Precision Medicine (TOPMed) programme seeks to elucidate the genetic architecture and biology of heart, lung, blood and sleep disorders, with the ultimate goal of improving diagnosis, treatment and prevention of these diseases. The initial phases of the programme focused on whole-genome sequencing of individuals with rich phenotypic data and diverse backgrounds. Here we describe the TOPMed goals and design as well as the available resources and early insights obtained from the sequence data. The resources include a variant browser, a genotype imputation server, and genomic and phenotypic data that are available through dbGaP (Database of Genotypes and Phenotypes)1. In the first 53,831 TOPMed samples, we detected more than 400 million single-nucleotide and insertion or deletion variants after alignment with the reference genome. Additional previously undescribed variants were detected through assembly of unmapped reads and customized analysis in highly variable loci. Among the more than 400 million detected variants, 97% have frequencies of less than 1% and 46% are singletons that are present in only one individual (53% among unrelated individuals). These rare variants provide insights into mutational processes and recent human evolutionary history. The extensive catalogue of genetic variation in TOPMed studies provides unique opportunities for exploring the contributions of rare and noncoding sequence variants to phenotypic variation. Furthermore, combining TOPMed haplotypes with modern imputation methods improves the power and reach of genome-wide association studies to include variants down to a frequency of approximately 0.01%.

Novel Compound Heterozygous Mutations in CRTAP Cause Rare Autosomal Recessive Osteogenesis Imperfecta

Whole-exome sequencing (WES) has advantages over the traditional molecular test by screening 20,000 genes simultaneously and has become an invaluable tool for genetic diagnosis in clinical practice. Here, we reported a family with a child and a fetus presenting undiagnosed skeletal dysplasia phenotypes, while the parents were asymptomatic. WES was applied to the parents and affected fetus to identify the genetic cause of the phenotypes. We identified novel compound heterozygous mutations consisting of a single-nucleotide variant (SNV) and a large deletion in the CRTAP gene (NM_006371.4:c.1153-3C > G/hg19 chr3:g.32398837_34210906del). Genetic alterations of CRTAP are known to cause osteogenesis imperfecta (OI) in an autosomal recessive manner. Further examination of the proband’s elder sibling who was diagnosed as OI after birth found that she shares the inherited compound heterozygous mutations of CRTAP; thus, the findings support the disease-causing role of CRTAP mutations. Through the in vitro molecular test and in silico analysis, the deleterious effects of the splicing-altering SNV in CRTAP (c.1153-3C > G) on gene product were confirmed. Collectively, our WES-based pathogenic variant discovery pipeline identifies the SNVs and copy number variation to delineate the genetic cause on the proband affected with OI. The data not only extend the knowledge of mutation spectrum in patients with skeletal dysplasia but also demonstrate that WES holds great promise for genetic screening of rare diseases in clinical settings.

Natural Selection Shapes Codon Usage in the Human Genome

Synonymous codon usage has been identified as a determinant of translational efficiency and mRNA stability in model organisms and human cell lines. However, whether natural selection shapes human codon content to optimize translation efficiency is unclear. Furthermore, aside from those that affect splicing, synonymous mutations are typically ignored as potential contributors to disease. Using genetic sequencing data from nearly 200,000 individuals, we uncover clear evidence that natural selection optimizes codon content in the human genome. In deriving intolerance metrics to quantify gene-level constraint on synonymous variation, we discover that dosage-sensitive genes, DNA-damage-response genes, and cell-cycle-regulated genes are particularly intolerant to synonymous variation. Notably, we illustrate that reductions in codon optimality in BRCA1 can attenuate its function. Our results reveal that synonymous mutations most likely play an underappreciated role in human variation.

Mutational bias and the protein code shape the evolution of splicing enhancers

Exonic splicing enhancers (ESEs) are enriched in exons relative to introns and bind splicing activators. This study considers a fundamental question of co-evolution: How did ESE motifs become enriched in exons prior to the evolution of ESE recognition? We hypothesize that the high exon to intron motif ratios necessary for ESE function were created by mutational bias coupled with purifying selection on the protein code. These two forces retain certain coding motifs in exons while passively depleting them from introns. Through the use of simulations, genomic analyses, and high throughput splicing assays, we confirm the key predictions of this hypothesis, including an overlap between protein and splicing information in ESEs. We discuss the implications of mutational bias as an evolutionary driver in other cis-regulatory systems.

Splicing is regulated by cis-acting elements in pre-mRNAs such as exonic or intronic splicing enhancers and silencers. Here the authors show that exonic splicing enhancers are enriched in exons compared to introns due to mutational bias coupled with purifying selection on the protein code.

Extreme differences between human germline and tumor mutation densities are driven by ancestral human-specific deviations

Mutations do not accumulate uniformly across the genome. Human germline and tumor mutation density correlate poorly, and each is associated with different genomic features. Here, we use non-human great ape (NHGA) germlines to determine human germline- and tumor-specific deviations from an ancestral-like great ape genome-wide mutational landscape. Strikingly, we find that the distribution of mutation densities in tumors presents a stronger correlation with NHGA than with human germlines. This effect is driven by human-specific differences in the distribution of mutations at non-CpG sites. We propose that ancestral human demographic events, together with the human-specific mutation slowdown, disrupted the human genome-wide distribution of mutation densities. Tumors partially recover this distribution by accumulating preneoplastic-like somatic mutations. Our results highlight the potential utility of using NHGA population data, rather than human controls, to establish the expected mutational background of healthy somatic cells.

Nucleosome positioning stability is a modulator of germline mutation rate variation across the human genome

Nucleosome organization has been suggested to affect local mutation rates in the genome. However, the lack of de novo mutation and high-resolution nucleosome data has limited the investigation of this hypothesis. Additionally, analyses using indirect mutation rate measurements have yielded contradictory and potentially confounding results. Here, we combine data on >300,000 human de novo mutations with high-resolution nucleosome maps and find substantially elevated mutation rates around translationally stable (‘strong’) nucleosomes. We show that the mutational mechanisms affected by strong nucleosomes are low-fidelity replication, insufficient mismatch repair and increased double-strand breaks. Strong nucleosomes preferentially locate within young SINE/LINE transposons, suggesting that when subject to increased mutation rates, transposons are then more rapidly inactivated. Depletion of strong nucleosomes in older transposons suggests frequent positioning changes during evolution. The findings have important implications for human genetics and genome evolution.

Nucleosome organization has been suggested to affect local mutation rates in the genome. Here, the authors analyse data on >300,000 human de novo mutations and high-resolution nucleosome maps and provide evidence that nucleosome positioning stability modulates germline mutation rate variation across the human genome.

Applicability of the Mutation-Selection Balance Model to Population Genetics of Heterozygous Protein-Truncating Variants in Humans

The fate of alleles in the human population is believed to be highly affected by the stochastic force of genetic drift. Estimation of the strength of natural selection in humans generally necessitates a careful modeling of drift including complex effects of the population history and structure. Protein-truncating variants (PTVs) are expected to evolve under strong purifying selection and to have a relatively high per-gene mutation rate. Thus, it is appealing to model the population genetics of PTVs under a simple deterministic mutation-selection balance, as has been proposed earlier (Cassa et al. 2017). Here, we investigated the limits of this approximation using both computer simulations and data-driven approaches. Our simulations rely on a model of demographic history estimated from 33,370 individual exomes of the Non-Finnish European subset of the ExAC data set (Lek et al. 2016). Additionally, we compared the African and European subset of the ExAC study and analyzed de novo PTVs. We show that the mutation-selection balance model is applicable to the majority of human genes, but not to genes under the weakest selection.

Genetic intolerance analysis as a tool for protein science

Recent advances in whole genome and exome sequencing have dramatically increased the database of human gene variations. There are now enough sequenced human exomes and genomes to begin to identify gene variations that are notable because they are NOT observed in sequenced human genomes, apparently because they are subject to "purifying selection", exemplifying genetic intolerance. Such "dysprocreative" gene variations are embryonic lethal or prevent reproduction through any one of a number of possible mechanisms. Here we review an emerging quantitative approach, "Missense Tolerance Ratio" (MTR) analysis, that is used to assess protein-encoding gene (cDNA) sequence intolerance to missense mutations based on analysis of the >100 K and growing number of currently available human genome and exome sequences. This approach is already useful for analyzing intolerance to mutations in cDNA segments with a resolution on the order of 90 bases. Moreover, as the number of sequenced genomes/exomes increases by orders of magnitude it may eventually be possible to assess mutational tolerance in a statistically robust manner at or near single site resolution. Here we focus on how cDNA intolerance analysis complements other bioinformatic methods to illuminate structure-folding-function relationships for the encoded proteins. A set of disease-linked membrane proteins is employed to provide examples.

The evolution of Great Apes has shaped the functional enhancers' landscape in human embryonic stem cells

High-throughput functional assays of enhancer activity have recently enabled the genome-scale definition of molecular, structural, and biochemical features of these genomic regulatory regions. To infer the evolutionary origin of DNA sequences operating as functional enhancers in human embryonic stem cells (hESC), we examined the patterns of evolutionary conservation and divergence in the genome-wide functional enhancers' landscape of hESC. We show that a prominent majority (up to 94%) of DNA sequences identified in hESC as functional enhancers are conserved in humans and our closest evolutionary relatives, Chimpanzee and Bonobo. More than 91% of functional enhancers that are highly conserved in both Chimpanzee and Bonobo, are conserved among other Great Apes and >75% are conserved in the Rhesus genome. In striking contrast, <5% of DNA sequences operating in hESC as functional enhancers are conserved in rodents. Conserved in primates enhancers' sequences are complemented by 1619 sequences of enhancers that are specific to humans. Enhancers that harbor human-specific sequences appear enriched among the invariant enhancer module maintaining activity in different pluripotent states and these regions are associated with pluripotency- and embryonic-lineage-related genes. However, functional enhancers make up only a minority of all conserved in primates or human-specific transcription factor binding sites. Our analyses revealed that sequences that are conserved during ~8 million years of primate evolution dominate the genomic landscape of functional enhancers in both primed and naïve hESC. Collectively, these observations revealed thousands of evolutionarily conserved sequences that function as a core regulatory network in human embryonic stem cells which has recently undergone further extension after divergence of modern humans from our closest relatives, Chimpanzee and Bonobo.

Inductive determination of allele frequency spectrum probabilities in structured populations

We present a method for inductively determining exact allele frequency spectrum (AFS) probabilities for samples derived from a population comprising two demes under the infinite-allele model of mutation. This method builds on a labeled coalescent argument to extend the Ewens sampling formula (ESF) to structured populations. A key departure from the panmictic case is that the AFS conditioned on the number of alleles in the sample is no longer independent of the scaled mutation rate (θ). In particular, biallelic site frequency spectra, widely-used in explorations of genome-wide patterns of variation, depend on the mutation rate in structured populations. Variation in the rate of substitution across loci and through time may contribute to apparent distortions of site frequency spectra exhibited by samples derived from structured populations.

Adaptive Landscape of Protein Variation in Human Exomes

The human genome contains hundreds of thousands of missense mutations. However, only a handful of these variants are known to be adaptive, which implies that adaptation through protein sequence change is an extremely rare phenomenon in human evolution. Alternatively, existing methods may lack the power to pinpoint adaptive variation. We have developed and applied an Evolutionary Probability Approach (EPA) to discover candidate adaptive polymorphisms (CAPs) through the discordance between allelic evolutionary probabilities and their observed frequencies in human populations. EPA reveals thousands of missense CAPs, which suggest that a large number of previously optimal alleles experienced a reversal of fortune in the human lineage. We explored nonadaptive mechanisms to explain CAPs, including the effects of demography, mutation rate variability, and negative and positive selective pressures in modern humans. Many nonadaptive hypotheses were tested, but failed to explain the data, which suggests that a large proportion of CAP alleles have increased in frequency due to beneficial selection. This suggestion is supported by the fact that a vast majority of adaptive missense variants discovered previously in humans are CAPs, and hundreds of CAP alleles are protective in genotype-phenotype association data. Our integrated phylogenomic and population genetic EPA approach predicts the existence of thousands of nonneutral candidate variants in the human proteome. We expect this collection to be enriched in beneficial variation. The EPA approach can be applied to discover candidate adaptive variation in any protein, population, or species for which allele frequency data and reliable multispecies alignments are available.

Predicting the clinical impact of human mutation with deep neural networks

Millions of human genomes and exomes have been sequenced, but their clinical applications remain limited due to the difficulty of distinguishing disease-causing mutations from benign genetic variation. Here we demonstrate that common missense variants in other primate species are largely clinically benign in human, enabling pathogenic mutations to be systematically identified by the process of elimination. Using hundreds of thousands of common variants from population sequencing of six non-human primate species, we train a deep neural network that identifies pathogenic mutations in rare disease patients with 88% accuracy and enables the discovery of 14 new candidate genes in intellectual disability at genome-wide significance. Cataloging common variation from additional primate species would improve interpretation for millions of variants of uncertain significance, further advancing the clinical utility of human genome sequencing.

Massive variation of short tandem repeats with functional consequences across strains of Arabidopsis thaliana

Short tandem repeat (STR) mutations may comprise more than half of the mutations in eukaryotic coding DNA, yet STR variation is rarely examined as a contributor to complex traits. We assessed this contribution across a collection of 96 strains of Arabidopsis thaliana, genotyping 2046 STR loci each, using highly parallel STR sequencing with molecular inversion probes. We found that 95% of examined STRs are polymorphic, with a median of six alleles per STR across these strains. STR expansions (large copy number increases) are found in most strains, several of which have evident functional effects. These include three of six intronic STR expansions we found to be associated with intron retention. Coding STRs were depleted of variation relative to noncoding STRs, and we detected a total of 56 coding STRs (11%) showing low variation consistent with the action of purifying selection. In contrast, some STRs show hypervariable patterns consistent with diversifying selection. Finally, we detected 133 novel STR-phenotype associations under stringent criteria, most of which could not be detected with SNPs alone, and validated some with follow-up experiments. Our results support the conclusion that STRs constitute a large, unascertained reservoir of functionally relevant genomic variation.

An Upper Limit on the Functional Fraction of the Human Genome

For the human population to maintain a constant size from generation to generation, an increase in fertility must compensate for the reduction in the mean fitness of the population caused, among others, by deleterious mutations. The required increase in fertility due to this mutational load depends on the number of sites in the genome that are functional, the mutation rate, and the fraction of deleterious mutations among all mutations in functional regions. These dependencies and the fact that there exists a maximum tolerable replacement level fertility can be used to put an upper limit on the fraction of the human genome that can be functional. Mutational load considerations lead to the conclusion that the functional fraction within the human genome cannot exceed 15%.

Substitutions into amino acids that are pathogenic in human mitochondrial proteins are more frequent in lineages closely related to human than in distant lineages

Propensities for different amino acids within a protein site change in the course of evolution, so that an amino acid deleterious in a particular species may be acceptable at the same site in a different species. Here, we study the amino acid-changing variants in human mitochondrial genes, and analyze their occurrence in non-human species. We show that substitutions giving rise to such variants tend to occur in lineages closely related to human more frequently than in more distantly related lineages, indicating that a human variant is more likely to be deleterious in more distant species. Unexpectedly, substitutions giving rise to amino acids that correspond to alleles pathogenic in humans also more frequently occur in more closely related lineages. Therefore, a pathogenic variant still tends to be more acceptable in human mitochondria than a variant that may only be fit after a substantial perturbation of the protein structure.

The population genetics of human disease: The case of recessive, lethal mutations

Do the frequencies of disease mutations in human populations reflect a simple balance between mutation and purifying selection? What other factors shape the prevalence of disease mutations? To begin to answer these questions, we focused on one of the simplest cases: recessive mutations that alone cause lethal diseases or complete sterility. To this end, we generated a hand-curated set of 417 Mendelian mutations in 32 genes reported to cause a recessive, lethal Mendelian disease. We then considered analytic models of mutation-selection balance in infinite and finite populations of constant sizes and simulations of purifying selection in a more realistic demographic setting, and tested how well these models fit allele frequencies estimated from 33,370 individuals of European ancestry. In doing so, we distinguished between CpG transitions, which occur at a substantially elevated rate, and three other mutation types. Intriguingly, the observed frequency for CpG transitions is slightly higher than expectation but close, whereas the frequencies observed for the three other mutation types are an order of magnitude higher than expected, with a bigger deviation from expectation seen for less mutable types. This discrepancy is even larger when subtle fitness effects in heterozygotes or lethal compound heterozygotes are taken into account. In principle, higher than expected frequencies of disease mutations could be due to widespread errors in reporting causal variants, compensation by other mutations, or balancing selection. It is unclear why these factors would have a greater impact on disease mutations that occur at lower rates, however. We argue instead that the unexpectedly high frequency of disease mutations and the relationship to the mutation rate likely reflect an ascertainment bias: of all the mutations that cause recessive lethal diseases, those that by chance have reached higher frequencies are more likely to have been identified and thus to have been included in this study. Beyond the specific application, this study highlights the parameters likely to be important in shaping the frequencies of Mendelian disease alleles.

What determines the frequencies of disease mutations in human populations? To begin to answer this question, we focus on one of the simplest cases: mutations that cause completely recessive, lethal Mendelian diseases. We first review theory about what to expect from mutation and selection in a population of finite size and generate predictions based on simulations using a plausible demographic scenario of recent human evolution. For a highly mutable type of mutation, transitions at CpG sites, we find that the predictions are close to the observed frequencies of recessive lethal disease mutations. For less mutable types, however, predictions substantially under-estimate the observed frequency. We discuss possible explanations for the discrepancy and point to a complication that, to our knowledge, is not widely appreciated: that there exists ascertainment bias in disease mutation discovery. Specifically, we suggest that alleles that have been identified to date are likely the ones that by chance have reached higher frequencies and are thus more likely to have been mapped. More generally, our study highlights the factors that influence the frequencies of Mendelian disease alleles.