Maria C. Ávila Arcos

Admixture between populations is a common feature of human history. Admixture events introduce new genetic variation that can fuel evolution. Characterizing the significance of admixture events on the evolution of populations across various species is of great interest to evolutionary geneticists. Local ancestry inference (LAI) methods infer genetic ancestry of an individual at a particular chromosomal location. Certain methods specialize in detecting archaic introgression, which consists of interbreeding between modern and archaic humans like Neanderthals and Denisovans. Most current LAI methods allow the detection of a single archaic ancestry, and post-processing may distinguish between multiple waves of introgression. These methods vary in how they choose archaic or modern reference genomes for the inference. Here, we present a new hidden Markov model-based method (DAIseg), which has the advantage of simultaneously distinguishing between multiple waves of ancient and recent admixture, using only modern human reference genomes. Simulations demonstrate that DAIseg achieves higher overall performance than state-of-the-art methods. We also apply DAIseg to Papuan populations to jointly detect Denisovan and Neanderthal introgressed segments, and identify a higher number of archaic segments than previous methods. Analysis of inferred introgressed segments shows that we can identify evidence for two Denisovan introgression events in Papuans. Overall, on top of being able to deal with both archaic and recent admixture, DAIseg provides a more principled approach for detecting and classifying Denisovan and Neanderthal segments which will improve downstream analysis of introgressed segments to infer the impact of archaic introgression in humans.

Studying the evolutionary history of archaic segments in recently admixed individuals requires inferring both continental and archaic ancestry in admixed genomes. Here, we present TRACTINATOR , the first deep-learning method for simultaneous inference of continental and archaic ancestry in admixed human genomes. The model combines SNP sequences, population allele-frequency information, and S* statistics to improve both inference tasks. By learning relationships between haplotypes and population allele frequencies, TRACTINATOR can generalize across genomic regions and even across different genomic datasets. We train our model using both real and synthetic data, and show that augmenting with synthetic data improves accuracy for both continental and archaic ancestry inference. Finally, we apply TRACTINATOR to admixed Latin American populations from the 1,000 Genomes Project, revealing how archaic ancestry is distributed within chromosomal segments of African, European and Indigenous American ancestry in Latin American individuals. For candidates of adaptive introgression, we also infer whether the archaic haplotype was introduced via European or Indigenous American ancestors.

When modern humans expanded from Africa into Eurasia, they interbred with archaic hominins such as Neanderthals and Denisovans. This introgression shaped human evolution, yet most insights have been gained from present-day genomes, leaving little known about how archaic variants evolved after interbreeding. Ancient genomes offer a direct view of this process, but low coverage and poor quality have limited their use. Recent advances in genotype imputation offer a way to overcome these challenges by reconstructing missing information from reference panels and recovering evolutionary signals from low-coverage data. Here, we show that imputation enables accurate detection and quantification of archaic introgression in ancient genomes, improves local archaic ancestry inference, and that regions of archaic ancestry are imputed with especially high accuracy. We further demonstrate that imputed genomes can reconstruct the trajectories of introgressed haplotypes, distinguish populations across time and geography, and identify both known and additional candidates for adaptive introgression.

Motivation: Local Ancestry Inference (LAI) allows us to study evolutionary processes in admixed populations[1], uncover ancestry-specific disease risk factors[2], and to better understand the demographic history of these populations[3]. Many methods for LAI exist, however, these methods usually focus on cases of intercontinental admixture. In this work, we evaluate both existing and novel methods in challenging scenarios, such as downsampled reference panels, intracontinental admixture, and distant admixture events. Results: We present four novel LAI implementations based on neural network architectures, including Bidirectional Long Short-Term Memory and Transformer networks which have not previously been used for LAI. We compare these novel implementations to existing methods for LAI across a variety of scenarios using the 1 Thousand Genomes dataset and other synthetic datasets. We find that while all networks achieve high performance for intercontinental admixture scenarios, inference power is comparatively low for scenarios of intracontinental or distant admixture. We further show how our implementations achieve the best performance of all methods through specialized preprocessing and inference smoothing steps. Availability: All implementations and benchmarking code available at https://github.com/Jazpy/LAINNs .

RNA viruses cause substantial global morbidity, yet their impact prior to the twentieth century remains obscured. While ancient DNA studies have transformed our understanding of past pathogens, ancient RNA (aRNA) isolation is largely restricted to exceptionally preserved samples. Here, we simultaneously recover aDNA and aRNA from non–formalin-fixed human lung specimens and reconstructed an 18th-century Human Rhinovirus (HRV) A genome—the oldest human RNA virus identified to date. The RNA is highly fragmented, with distinctive terminal misincorporations and coverage patterns consistent with double-stranded RNA. Phylogenetic analyses indicate that this historical HRV genome is an extinct lineage related to contemporary genotypes, providing a unique perspective on rhinovirus evolution. These findings demonstrate that centuries-old medical specimens can retain informative aRNA, expanding the temporal scope of paleovirology.

Pathogen emergence and adaptation are constant, but the mechanisms underlying pathogen success as well as host susceptibility and resistance are often only observable in time series data. Ancient DNA research of pathogens and their hosts provides unique insights into past occurrences, including the changes that led to pathogen jumps between animals and humans, pandemic outbreaks, the timing of such events and the genetic, cultural and ecological factors that affect pathogen success and human survival and recovery. Recent technological improvements and the increasing number of ancient samples analysed have enabled the unprecedented investigation of pathogen evolution. Such studies are poised to benefit from the increased diversity of sequenced ancient pathogens, adoption of a broader framework that considers the entire ecosystem of host–pathogen interactions and growing collaboration with related fields.

Salmonella enterica is a widespread pathogen of major global health relevance, with over 2,500 serovars classified into non-typhoidal and typhoidal groups. Within the typhoidal group, S. enterica Paratyphi C causes paratyphoid fever in humans. Ancient DNA from this bacterium has previously been recovered from epidemic-associated burials in Eurasia and Mexico, dating back 6,000 to 300 years. Here, we analyzed dental DNA from seven individuals (radiocarbon dated to 1800–1940 CE) buried at the Temple of the Immaculate Conception in Mexico City, Central Mexico, and identified ancient S. enterica Paratyphi C DNA in a young female with Native American ancestry. Using an in-house targeted enrichment strategy and deep shotgun sequencing, we reconstructed a S. enterica Paratyphi C genome (COYC5) with ∼11X coverage. Phylogenetic and comparative genomic analyses place COYC5 in close association with previously reported S. enterica Paratyphi C genomes associated with the 1545 cocoliztli epidemic in southern Mexico, as well as with European strains. Divergence estimates indicate that the Mexican and European lineages shared a common ancestor approximately 1,400 years before present (yBP), reflecting an ancient evolutionary split predating European colonization of Mexico. In contrast, the divergence between COYC5 and the southern Mexican genomes occurred around 516 yBP, coinciding with the onset of the colonial period. This pattern supports a European introduction of S. enterica Paratyphi C during colonization, followed by its local diversification within Mexico. Despite this regional differentiation, the conserved presence of key virulence loci—such as SPI-7 and an active shufflon system—across COYC5, southern Mexican, and European genomes underscores the enduring pathogenic potential of S. enterica Paratyphi C. The identification of this bacterium in 19th-century Mexico City provides the first genomic evidence of its persistence in urban contexts beyond major epidemic outbreaks, offering new insights into its evolutionary trajectory in Mexico. These findings raise new questions about how the pathogen spread and persisted across different ecological, social, and epidemiological contexts in ancient Mexico and the Americas.

Paleogenomic studies suggest that Mammuthus columbi derives from an ancient hybridization between Mammuthus primigenius and Mammuthus trogontherii . Although its habitat spanned from North to Central America, available genetic data are limited to temperate regions, leaving gaps in knowledge of the species’ demographic history on the continent. In this study, we generated 61 capture-enriched M. columbi mitogenomes from the Basin of Mexico, located in Central Mexico. Our analysis reveals that these mitogenomes belong to a mitochondrial lineage distinct from that of other North American mammoth mitogenomes. These divergent mitogenomes suggest a deep population structure in their ancestors and challenge prior assumptions based on geographically restricted samples. Our findings underscore the importance of wider spatial sampling to reconstruct mammoths’ evolutionary history and demonstrate the feasibility of studying megafauna from tropical latitudes.

We study the gene MUC19 , for which some modern humans carry a Denisovan-like haplotype. MUC19 is a mucin, a glycoprotein that forms gels with various biological functions. We find diagnostic variants for the Denisovan-like MUC19 haplotype at high frequencies in admixed American individuals and at highest frequency in 23 ancient Indigenous American individuals, all pre-dating population admixture with Europeans and Africans. We find that the Denisovan-like MUC19 haplotype is under positive selection and carries a higher copy number of a 30–base-pair variable number tandem repeat, and that copy numbers of this repeat are exceedingly high in admixed American populations. Finally, we find that some Neanderthals carry the Denisovan-like MUC19 haplotype, and that it was likely introgressed into modern human populations through Neanderthal introgression rather than Denisovan introgression.

When modern humans expanded from Africa into Eurasia, they interbred with archaic hominins such as Neanderthals and Denisovans. This groundbreaking discovery, made in part possible through the genomic analysis of archaic remains, reshaped our understanding of human origins, and opened new research avenues to study the effects and consequences of archaic introgression. While significant progress has been made in identifying and quantifying archaic introgression, most studies have focused on contemporary individual genomic data. As a result, relatively little is known about the evolution of archaic variants within modern human populations shortly after interbreeding occurred. While analysis of ancient DNA from modern humans offers the potential to address this scientific gap, its poor quality has hindered its exploration. However, recent studies have shown that imputation using contemporary reference panels can accurately infer missing genotypes in ancient genomes. Here, we investigate the feasibility of using imputation to improve both global and local archaic ancestry inference in ancient genomes, by downsampling to different low-coverage values and imputing 20 high-coverage (>10X) ancient genomes, representing individuals from diverse temporal and geographical contexts. We tested the reliability of detecting and quantifying archaic introgression using D-statistics and the f4-ratio . We identified consistent results from the imputed and the original ancient genomes, suggesting that genomic estimates to detect and quantify introgression work well in imputed genomes. Regarding Local Ancestry Inference we find that we can identify more introgressed segments in imputed genomes than in non-imputed genomes. Surprisingly, we find that imputation accuracy is even higher in regions of archaic ancestry compared to other regions of the genome, facilitating the detection of introgressed segments in imputed genomes. Imputation also allows detection of Denisovan segments in Siberian and Alaskan individuals. We show that segments identified even at 0.0625X coverage can be used to reconstruct the history of introgressed haplotypes. For example, comparisons with archaic segments in contemporary humans reveal that the oldest individual analyzed, Ust’Ishim, carried segments that are now found exclusively in either European or East Eurasian populations, indicating that these segments co-occurred in a shared ancestral population. By comparing only ancient individuals, we demonstrate that clustering based solely on the identified archaic segments effectively groups individuals into genetic clusters that correspond to populations defined by geography and time. This analysis shows a clear distinction between individuals with Denisovan introgression and those without, as well as a more detailed separation between Mesolithic and Neolithic Europeans, with the latter closely resembling Central Eurasian individuals. Furthermore, despite the small sample size in this study, we are able to reconstruct the origins of genes identified as candidates for adaptive introgression in contemporary populations, such as the BCN2 gene, which is an adaptively introgressed gene identified in contemporary West Eurasians. Additionally, we identify new candidates for adaptive introgression including LEMD2 and MLN in Europe. In this gene-region, imputation helps resolve the introgressed haplotype, which is closest to the Vindija Neanderthal haplotype.