key: cord-269124-oreg7rnj authors: Spyrou, Maria A.; Bos, Kirsten I.; Herbig, Alexander; Krause, Johannes title: Ancient pathogen genomics as an emerging tool for infectious disease research date: 2019-04-05 journal: Nat Rev Genet DOI: 10.1038/s41576-019-0119-1 sha: doc_id: 269124 cord_uid: oreg7rnj Over the past decade, a genomics revolution, made possible through the development of high-throughput sequencing, has triggered considerable progress in the study of ancient DNA, enabling complete genomes of past organisms to be reconstructed. A newly established branch of this field, ancient pathogen genomics, affords an in-depth view of microbial evolution by providing a molecular fossil record for a number of human-associated pathogens. Recent accomplishments include the confident identification of causative agents from past pandemics, the discovery of microbial lineages that are now extinct, the extrapolation of past emergence events on a chronological scale and the characterization of long-term evolutionary history of microorganisms that remain relevant to public health today. In this Review, we discuss methodological advancements, persistent challenges and novel revelations gained through the study of ancient pathogen genomes. The long shared history between humans and infectious disease places ancient pathogen genomics within the inter est of several fields such as microbiology, evolutionary biology, history and anthropology. Research on this topic aims to better understand the interactions between pathogens and their hosts on an evolutionary timescale, to uncover the origins of pathogens and to disentangle the genetic processes involved in their epidemic emer gence among human populations. Over the past 10,000 years, major transitions in human subsistence strategies, such as those that accompanied the Neolithic revolution 1 , likely exposed our species to a novel range of infectious agents 2 . Closer contact with domesticated animals would have increased the frequency of zoonotic transmission events, and higher human population densities would have enhanced the potential of pathogens to propagate within and between groups. Throughout human his tory, a number of epidemics and pandemics have been recorded or are hypothesized to have occurred (Fig. 1) . Although most of their causative agents still remain speculative, robust molecular methods coupled with archaeological and historical data can confidently demonstrate the involvement of certain pathogens in these episodes. The investigation of past infectious diseases has tra ditionally been conducted through palaeopathological assessment of ancient skeletal assemblages 3, 4 , although this approach is limited by the fact that most acute infec tions do not leave visible traces on bone. Since the 1990s, the field of ancient DNA (aDNA) has brought molecular techniques to this study, providing a diachronic genetic perspective to infectious disease research. Initial attempts relied on PCR technology [5] [6] [7] [8] [9] , which restricted the study of ancient microbial DNA to targeted, short genomic fragments that were amplified from ancient human remains. This method made infectious disease detection possible but gave limited information on the evolution ary history of the patho gen. In addition, complications associated with the study of aDNA, which is typically present at low quantities, is heavily fragmented and har bours chemical modifications [10] [11] [12] , hampered efforts to reproduce and authenticate early findings [13] [14] [15] . Over the past decade, major advancements in geno mics, in particular, the development of high throughput sequencing, also called next generation sequencing (NGS) 16 , radically increased the amount of data that can be retrieved from ancient remains. This techno logy has assisted the development of quantitative meth ods for aDNA authentication 11, 12, [17] [18] [19] and has enabled the retrieval of whole ancient pathogen genomes from archaeological specimens. The first such genome, pub lished in 2011 (reF. 20 ), was that of the notorious bacterial pathogen Yersinia pestis, the causative agent of plague. Since then, the field has expanded its directions to the in depth study of infectious disease evolution, providing a unique resource for understanding human history. Here, we review the latest methodological innovations that have aided the whole genome retrieval and evolu tionary analysis of various ancient pathogens (TAble 1) , most of which are still relevant to public health today. A scientific field focused on the study of whole pathogen genomes retrieved from ancient human, animal or plant remains. The cultural transition associated with the adoption of farming, animal husbandry and domestication as well as the practice of a sedentary lifestyle among human populations. The infectious disease transmission from animals to humans. In the second half of this Review, we highlight the util ity of this approach by discussing evolutionary events in the history of Y. pestis that have been uniquely revealed through the study of ancient genomes. Methods for isolating ancient microbial DNA The sweet spot for ancient pathogen DNA. The retrieval of DNA from ancient human, animal or plant remains carries with it a number of challenges, namely, its limited preservation and hence low abundance, its highly fragmented and damaged state and the perva sive modern DNA contamination that necessitates a confident evaluation of its authenticity 21, 22 . Efficient aDNA recovery is best accomplished via sampling of the anatomical element that contains the highest quantity of DNA from the target organism. For human aDNA analysis, bone and teeth have been the preferred study material, given their abundance in the archaeological record. Recent studies suggest that the inner ear por tion of the petrous bone 23 and the cementum layer of teeth 24 have the greatest potential for successful human DNA retrieval. However, petrous bone sampling and shotgun NGS sequencing of aDNA from five Bronze Age skeletons previously shown to be carrying Y. pestis failed to detect the bacterium in this source material, suggesting that its preservation potential for pathogen DNA is low 25 . Direct sampling from skeletal lesions, where present, has proved a rich source of aDNA for some chronic disease causing bacteria, such as Mycobacterium tuberculosis, which was isolated from vertebrae 26 ; Mycobacterium leprae, which could be isolated from portions of the maxilla and various long bones 27, 28 ; and Treponema pallidum subsp. pallidum and T. pallidum subsp. pertenue, which have been isolated from long bones 29 . Of note, the sampling methods for recovering pathogen DNA do not generally follow a standardized procedure, in part because of the great diversity in tissue tropism and resulting disease progression. In addition, acute blood borne infections do not typically produce diagnostic bone changes as opposed to those that affect their hosts chronically 3 . Therefore, if infections have caused mortal ity in the acute phase, as is the case for individuals from epidemic contexts who do not display skeletal evidence of infection, the preferred study material has been the inner cavities of teeth. Pathogen aDNA is thought to be preserved within the remnants of the pulp chamber, likely as part of desiccated blood 8, 17 . Consequently, tooth sampling has proved successful in the retrieval of whole genomes or genome wide data (that is, low coverage genomes that have provided limited analytical resolution) from ancient bacteria such as Y. pestis 20, [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] , Borrelia recurrentis 40 and Salmonella enterica 41 ; ancient eukaryotic pathogens such as Plasmodium falciparum 42 ; and ancient viruses such as hepatitis B virus (HBV) 43, 44 and human parvovirus B19 (B19V) 45 . Even M. leprae, which commonly manifests in the chronic form, has been retrieved from ancient teeth 27, 28 . Other types of specimen have also shown potential for aDNA retrieval. Examples are dental calculus as a source of oral pathogens, such as Tannerella forsythia 46 Pandemics refers to increased, often sudden, disease occurrence within populations across more than one region or continent, whereas epidemics refers to increased disease occurrences within a confined region or country. The evaluation of the health status of ancient individuals or populations, usually through the analysis of disease marker presence on skeletal assemblages. (aDNA). The DNA that has been retrieved from historical, archaeological or palaeontological remains. Tropism refers to the type of tissue or cell in which infection is established and supported. Segregating the metagenomic soup: methods for pathogen detection. Regardless of the source of genetic material, most ancient specimens yield complex metagenomic data sets. Poorly preserved aDNA usually makes up a miniscule fraction of the total genetic mate rial extracted from a sample (<1%), and the majority of DNA usually stems from organisms residing in the envi ronment 41 . Hence, specialized protocols are necessary for the detection and isolation of ancient pathogen DNA and its confident segregation from a rich environmental DNA background (Fig. 2) . In this context, laboratory based techniques are sep arated into those that target a specific microorganism and those that screen for several pathogenic micro organisms simultaneously (Fig. 2 ). Methods that screen for a single microorganism have used species specific assays of conventional or quantitative PCR (also known as real time PCR) 17,61-64 , as well as hybridization based enrichment techniques 17, 26, 28 (Fig. 2 ). These methods are particularly useful when the target microorganism is known, for example, in the presence of diagnostic skeletal lesions among the studied individuals 26, 28 , or when a hypothesis exists for the causative agent of an epidemic 17 . By contrast, broad laboratory based patho gen screening in aDNA research has used microarrays for both targeted enrichment 65 57 and HBV 54 that were sequenced using capillary sequencing (Sanger method). A term used to describe a specimen or data set that includes nucleic acid sequences from all organisms within the sampled proportion. The diagram provides an overview of techniques used for pathogen DNA detection in ancient remains by distinguishing between laboratory and computational methods. In both cases, processing begins with the extraction of DNA from ancient specimens 183 . As part of the laboratory pipeline, direct screening of extracts can be performed by PCR (quantitative (qPCR) or conventional) against species-specific genes, as done previously 17, 61, 63, 64 . PCR techniques alone, however, can suffer from frequent false-positive results and should therefore always be coupled with further verification methods such as downstream genome enrichment and/or next-generation sequencing (NGS) in order to ensure ancient DNA (aDNA) authentication of putatively positive samples. Alternatively , construction of NGS libraries 184, 185 has enabled pathogen screening via fluorescence-based detection on microarrays 66 and via DNA enrichment approaches 17 . The latter has been achieved, through single locus in-solution capture 26, 28 or through simultaneous screening for multiple pathogens using microarray-based enrichment of species-specific loci 65 and enables post-NGS aDNA authentication. In addition, data produced by direct (shotgun) sequencing of NGS libraries before enrichment can also be used for pathogen screening using computational tools. After pre-processing, reads can be directly mapped against a target reference genome (in cases for which contextual information is suggestive of a causative organism) or against a multigenome reference composed of closely related species to achieve increased mapping specificity of ancient reads. Alternatively , ancient pathogen DNA can also be detected using metagenomic profiling methods, as presented elsewhere 41, 71, 72 , through taxonomic assignment of shotgun NGS reads. Both approaches allow for subsequent assessment of aDNA authenticity and can be followed by whole pathogen genome retrieval through targeted enrichment or direct sequencing of positive sample libraries. detection 66 , whereby probes are designed to represent unique or conserved regions from a range of pathogenic bacteria, parasites or viruses. Although amplification based or fluorescence based approaches can be fast and cost effective for screening large sample collections 17, 38 , enrichment based techniques are usually coupled with NGS and therefore provide data that can be used to assess aDNA authenticity. When shotgun sequencing data are generated, com putational screening approaches can be used to detect the presence of pathogen DNA as well as for meta genomic profiling of ancient specimens (Fig. 2 ). In cases for which a causative agent is suspected, NGS reads can be directly mapped (for example, using the read align ment software Burrows-Wheeler aligner 67 ) against a specific reference genome or against a multigenome reference that includes several species of a certain genus with the purpose of achieving a higher mapping speci ficity to the target organism 34 (Fig. 2 ). In addition, broad approaches involve the use of metagenomic techniques for pathogen screening. Examples of tools that have shown their effectiveness with ancient metagenomic DNA include the widely used Basic Local Alignment Search Tool (BLAST) 68 ; the MEGAN Alignment Tool (MALT) 41 , which involves a taxonomic binning algorithm that can use whole genome databases (such as the National Center for Biotechnical Information (NCBI) Reference Sequence (RefSeq) database 69 ); Metagenomic Phylogenetic Analysis (MetaPhlAn) 70 , which is also integrated into the metagenomic pipeline MetaBIT 71 and uses thousands (or millions) of marker genes for the distinction of specific microbial clades; or Kraken 72 , an alignment free sequence classifier that is based on k-mer matching of a query to a constructed database. Taxonomic sequence assignments from the above methods, however, should be interpreted with caution, mainly because some pathogenic microorganisms have close environmental relatives that are often insuffi ciently represented in public databases. For example, a >97% sequence identity was shown between environ mental taxa and human associated pathogens such as M. tuberculosis and Y. pestis according to an analysis of 16S ribosomal RNA genes 73 . As such, given that envi ronmental DNA often dominates ancient remains that stem from burial contexts 74 , analyses should always ensure a qualitative assessment of assigned reads, that is, an evaluation of their mapping specificity and their genetic distance (also called edit distance) to the puta tively detected organism. In addition, one should con sider the known aDNA damage characteristics as criteria for data authenticity. Although several types of chemical damage can affect post mortem DNA survival, certain characteristics have been more extensively quantified. The first, termed depurination, is a hydrolytic mecha nism under which purine bases become excised from DNA strands. This process results in the formation of abasic sites and is a known contributor to the fragmen tation patterns observed in aDNA. As such, an increased base frequency of A and G compared with C and T immediately preceding the 5ʹ ends of aDNA fragments is often considered a criterion for authenticity 12 . A sec ond type of damage commonly identified among aDNA data sets is the hydrolytic deamination of C, whereby a C base is converted into U (and detected as its DNA analogue, T) 12, 75 . This base modification usually occurs at single stranded DNA overhangs that are most acces sible to environmental insults, resulting in an increased frequency of miscoding lesions at the terminal ends of aDNA fragments 11, 12 . Consequently, the evaluation of DNA damage profiles (for instance, by using map Damage2.0 (reF. 76 )) is a prerequisite for authenticating ancient pathogen DNA and is necessary for ensuring aDNA data integrity in general. More detailed overviews of authentication criteria in ancient pathogen research have been reviewed elsewhere 19, 73 . Targeted enrichment approaches to isolate whole ancient pathogen genomes. Evolutionary relationships between past and present infectious agents are best determined through the use of whole genome sequences of pathogens. However, the recovery of high quality data is often challenging owing to the aforementioned char acteristics of aDNA and therefore requires specialized sample processing. For example, in cases in which aDNA authenticity has already been achieved in the detection step, U residues resulting from post mortem C deami nation can be entirely 77 or partially 78 excised from aDNA molecules using the enzyme uracil DNA glycosylase (UDG) to avoid their interference with downstream read mapping and variant calling. In addition, given the low proportion of patho gen DNA in ancient remains, a common and cost effective approach for whole genome retrieval involves microarray based or in solutionbased hybridization capture. Both methods constitute a form of genomic selec tion of continuous or discontinuous genomic regions through the design and use of single stranded DNA or RNA probes that are complementary to the desired tar get. Microarray based capture utilizes densely packed probes that are immobilized on a glass slide 79 . It is cost effective in that it permits the parallel enrichment of molecules from several libraries that can be subsequently recovered through deep sequencing, although competi tion over the probes can impair enrichment efficiencies in specimens with comparatively lower target DNA con tents. Nevertheless, this type of capture has shown its effectiveness in the recovery of both ancient pathogen and human DNA 20, 26, 28, 41, 55, 80 . More recently, in solutionbased capture approaches have gained popularity owing to their capacity for greater sample throughput without compromising capture effi ciency [81] [82] [83] ; every sample library can be captured indi vidually, thus providing, in principle, an equal probe density per specimen. This technique has contributed to the increased number of specimens from which human genome wide single nucleotide polymorphism (SNP) data could be retrieved 84, 85 , even from climate zones that pose challenges to aDNA preservation (pre sented elsewhere [86] [87] [88] ). In addition, in solutionbased capture has recently become the preferred method for microbial pathogen genome recovery for both bacteria and DNA viruses (for examples, see reFs 34, 37, 41, 43, 45, 49, 50 ). Nevertheless, deep shotgun sequencing alone has also been used for human [89] [90] [91] and pathogen 28,33,48 high quality An algorithm that assigns metagenomic DNA reads to a species or a higher taxonomic rank (for example, genus or family) based on the sequence specificity. The matching, for each read, of multiple subsequences of length k without mismatches to a database. A hydrolytic reaction in which the β-N-glycosidic bond of a purine (adenine or guanine) is cleaved, causing its excision from a DNA strand. The hydrolytic removal of an amine group (NH 2 ) from a molecule. in ancient DNA studies, the term deamination most often refers to the deamination of cytosine residues into uracils. The identification of polymorphisms (nucleotide differences) in sequenced data by comparison to a reference. www.nature.com/nrg 328 | JuNe 2019 | volume 20 genome reconstruction, especially for specimens with fairly high endogenous DNA yields, although this frequently carries with it a greater production cost. In the absence of ancient pathogen genomes, the tim ings of infectious disease emergence and early spread are inferred mainly through comparative genomics of modern pathogen diversity 92, 93 , palaeopathological eval uation of ancient skeletal remains 94 or analysis of his torical records 95, 96 . Such approaches are highly valuable and, when combined, can be used to build an inter disciplinary picture of infectious disease history; however, limitations also exist. For example, the analysis of con temporary pathogen genetic diversity considers only a short time depth of available data and cannot predict evolutionary scenarios that derive from lineages that are now extinct. In addition, skeletal markers of specific infections in past populations only exist for a few con ditions and, when present, can rarely be considered as definitive, as numerous differential diagnoses can exist for a given skeletal pathology 97 . Similarly, historically recorded symptoms can often be misinterpreted given that past descriptions may be unspecific and do not always conform to modern medical terminology 98 . In the past decade, the reconstruction of ancient pathogen genomes has complemented such analyses with direct molecular evidence, often revealing aspects of past infections that were unexpected on the basis of existing data. The recent identification of HBV DNA in a mummified individual showing a vesicopustular rash 53 , which is usually considered characteristic of infection with VARV, highlights the importance of molecular methods in evaluating differential diagnoses. The oldest recovered genomic evidence of HBV to date was from a 7,000year old individual from present day Germany 44 , which shows that this pathogen has affected human populations since the Neolithic period. In addition, the virus was identified recently in human remains from the Bronze Age, Iron Age and up until the 16th century of the current era (ce) in Eurasia 43, 44, 53, 54 . Regarding bacterial pathogens, the identification of B. recurrentis in a 15th century individual from Norway 40 showed that -aside from Y. pestis -other vector borne pathogens were also circulating in medieval Europe. Furthermore, the causative agents of syphilis and yaws, T. pallidum subsp. pallidum and T. pallidum subsp. pertenue, respectively, were recently identified in different individuals from colonial Mexico 29 who exhib ited similar skeletal lesions. This study demonstrates the power of ancient pathogen genomics in distinguishing past infectious disease agents that are genetically and phenotypically similar but that differ greatly in their public health significance. Finally, the identification of G. vaginalis and S. saprophyticus in calcified nodules from a woman's remains (13th century Troy) 48 directly implicates these bacteria in pregnancy related com plications in the past. These findings, as well as other insights gained from analyses of ancient pathogen genomes (TAble 1) , demonstrate the ability of aDNA to contribute aspects of infectious disease history beyond those accessible by the palaeopathological, historical and modern genetic records. The reconstruction of whole pathogen genomes has not only been a tool for demonstrating infectious disease presence in the past but also aided in the robust infer ence of microbial phylogeography, which is important for understanding the processes that influence pathogen distribution and diversity over time. The evaluation of genetic relationships between ancient and modern pathogens is often conducted by direct whole genome or genome wide SNP compari sons of bacteria 20, 27, 29, 36, 48 , viruses 43, 44, 50, 53 or mito chondrial genomes and nuclear genome data from eukaryotic microorganisms 56, 59, 60 . Hence, accurate variant calling is critical for drawing reliable evolutionary inferences, although this process is often a challenge when handling data sets derived from samples with high rates of DNA fragmentation (resulting in ultrashort read data), low endogenous DNA content and high levels of DNA dam age. In these cases, increased accuracy is best achieved through stringent NGS read mapping parameters and through visual inspection of the sequences overlap ping the studied SNPs 35 . In addition, histograms of SNP allele frequencies -used to estimate the frequency of heterozygous calls in haploid organisms 26,52 -can often demonstrate the effects of environmental contamination on ancient microbial data sets 41 . Once variant calls are authenticated, one of the most common types of evolutionary inference in patho gen research is through phylogenetic analysis, which is a powerful means of resolving the genetic history of clonal microorganisms (Fig. 3) . Among the most commonly used tools in ancient microbial genomics are MEGA 99 , which comprises several phylogenetic methods; PhyML 100 , RAxML 101 and IQ TREE 102 , which implement maximum likelihood approaches; MrBayes 103 , which uses a Bayesian approach; and programs used for phylo genetic network inference, such as SplitsTree 104 . Two notable studies that examined phylogenetic relationships among ancient M. leprae genomes revealed a high strain diversity in Europe between the 5th and 14th centuries ce 27, 105 . Considered alongside the oldest palaeopatholog ical cases of leprosy dating to as early as the Copper and Bronze Age in Eurasia 106,107 and the high frequency of protective immune variants against the disease identi fied in modern day Europeans 108 , these results may sug gest a long history of M. leprae presence in this region. Moreover, the phylogenetic analysis of a 12th century S. enterica subsp. enterica genome from Europe showed its placement within the Paratyphi C lineage 109 . Further identification of the bacterium in 16th century colonial Mexico 41 revealed it as a previously unknown candidate pathogen that was likely introduced to the Americas through European contact. Given the low frequency of Paratyphi C today, these results may be indicative of a higher prevalence in past populations. Finally, an example from viral genomics is the recovery of HIV RNA from degraded serum specimens 57 , which high lighted the importance of archival collections in reconcil ing the expansion of recent pandemics. Specifically, these data were able to dispute a long standing hypothesis regarding the initiation of HIV spread in the USA. When the evolutionary histories of pathogens are influenced equally by mutation and recombination, additional tools have been used to identify recombining loci and to determine genetic relationships within and between microbial populations (Fig. 3) . For example, the programs ClonalFrameML 110 and Recombination Detection Program 4 (RDP4) 111 have been used to infer potential recombination regions within ancient 43, 44, 53 , respectively. In addition, principal component analysis (PCA) and ancient admix ture component estimation using the Bayesian modelling frameworks STRUCTURE 112 and fineSTRUCTURE 113 on both multilocus sequence typing (MLST) and whole genome data were recently used for population assignment of a 5,300year old H. pylori genome 49 . These analyses revealed key information on changes of the bacterial population structure that occurred in Europe over time. Furthermore, the recent study of ancient T. pallidum subsp. pallidum and T. pallidum subsp. pertenue 29 used the program TREE PUZZLE 114 , a maximum likelihoodbased phylogenetic algorithm, to gain a more robust phylogenetic resolution of ambiguous branching patterns among bacterial lineages. Such whole genome analyses of both clonal and recombining pathogens have helped to elucidate not only past infectious disease phylogeography but also possible zoonotic or anthroponotic transmission events that reveal disease interaction networks through time. Among others (TAble 1) , a notable example is that of 1,000year old pre Columbian M. tuberculosis genomes isolated from human remains, which showed a phylo genetic placement among animal adapted lineages, being most closely related to a strain circulating in modern day seals and sea lions 26 . Although the extent to which these strains were capable of human to human transmission is unclear, this study supports the existence of tuberculosis in pre Columbian South America and is helping to delineate the genomic and adaptive history of M. tuberculosis in the region before European contact 26 . Another example of intriguing evolutionary relationships revealed uniquely through the study of ancient pathogen genomes includes analy ses of Neolithic and Bronze Age HBV. These genomes grouped in extinct lineages that are most closely related to modern strains identified exclusively among African non human p ri ma te s 4 3 ,44 , a result that raises further questions regarding past transmission events in HBV history. Finally, the phylogenetic analysis of medieval M. leprae genomes suggested a European source for lep rosy in the Americas 28 , reinforcing the hypothesis that humans passed the disease to the nine banded arma dillo, the most common reservoir for this disease in the New World 115 . Importantly, the resolution of evolutionary analyses will depend on the quality, size and evenness of spatial sampling in the comparative data set. Therefore, the incomplete and often biased sampling of ancient and modern microbial strains can introduce challenges for discerning true biological relationships and past evo lutionary events. Nevertheless, in recent years, marked reductions in NGS costs 116 have aided the increased pro duction of large whole genome microbial data sets from present day strains. Current efforts for centralized data repositories that are continuously curated (such as the Pathosystems Resource Integration Center (PATRIC) database 117 and the recently introduced EnteroBase 118 ) and the development of robust phylogenetic frameworks that can accommodate genome wide data from >100,000 strains (for example, GrapeTree 119 ) are becoming valua ble for integrating large sample sizes into microbial evo lutionary analyses. In combination with the increasing number of ancient microbial data sets, these tools will aid in the evaluation of genetic relationships by offering higher resolution. Inferring divergence times through molecular dating. Apart from providing a molecular fossil record and revealing diachronic evolutionary relationships, a third analytical advantage gained from the retrieval of ancient pathogen genomes is that their ages can be directly used for calibration of a molecular clock. The ages of ancient specimens can be determined through contextual information, through archaeological artefacts or directly through radiocarbon dating, predominantly of bone or tooth collagen. Such temporal calibrations are required for high accuracy estimations of micro bial nucleotide substitution rates and in turn lineage divergence dates (Fig. 3) , particularly because both esti mations seem to be highly influenced by the time depth covered by the genomic data set 120 . For such analyses, the most widely used program is the Bayesian statistical framework BEAST 121, 122 . A characteristic example of how ancient calibration points can considerably affect divergence date estimates is that of M. tuberculosis. According to modern genetic data and human demographic events, the M. tuberculosis complex (MTBC) evolution was suggested to have fol lowed human migrations out of Africa, with its emer gence estimated at more than 70,000 years ago 93 . Recently, its emergence was re estimated to a maximum of 6,000 years ago on the basis of the 1,000year old myco bacterial genomes from Peru 26 , a result that was further The diagram is an overview of whole-genome analysis applied to date for ancient microbial data sets and distinguishes the methods used for clonal and recombining pathogens; of note, the depicted summary is not meant to represent an exhaustive pipeline of all possible analyses that could be undertaken. Ancient genome reconstruction is usually initiated through reference-based mapping or through de novo assembly of the data, although the latter has only been possible in exceptional cases of ancient DNA (aDNA) preservation 28, 44 . Subsequently , the genomes are assessed for their coverage depth and gene content for evaluation of their quality , which is also relevant for the comparative identification of virulence genes over their evolutionary time frames. Here, we show an example of virulence factor presence-or-absence analysis in the form of a heat map, as done previously 33, 34, 37, 41 . In addition, a comparison of the ancient genome or genomes with modern genomes can be carried out for single-nucleotide polymorphism (SNP) identification and for assessment of SNP effects (using SnpEff 186 ), which is particularly relevant for variants that seem to be unique to the ancient genome or genomes. Initial evolutionary inference can often be carried out through phylogenetic analysis and by testing for possible evidence of recombination in the analysed data set, for example, by comparing the support of different phylogenetic topologies 114 and by identifying potential recombination regions and homoplasies 110, 111 . If the data support clonal evolution, robust phylogenetic inference (for example, through a maximumlikelihood approach) is followed by assessment of the temporal signal in the data 124, 125 . If the data set shows a sufficient phylogenetic signal, molecular dating analysis and demographic modelling are considered possible, although the size of the data set will determine whether such analyses will be feasible and meaningful. Alternatively , if recombination is confirmed, genetic relationships between microbial clades or populations can be determined through phylogenetic network analysis 104 or through the use of population genetic methods such as principal component analysis (PCA) and identification of ancestral admixture components 112, 113 . In this case, the assessment of the temporal signal and proceeding with molecular dating analysis is cautioned and likely best performed after exclusion of recombination regions from all genomes in the data set. MRCA , most recent common ancestor. NGS, next-generation sequencing. A term used to describe that genome evolution occurs as a function of time and, therefore, the genetic distance between two living forms is proportional to the time of their divergence. A technique to estimate the age of a specimen on the basis of the amount of incorporated radiocarbon ( 14 C) that after the death of an organism gradually becomes lost over time. Denotes the frequency of substitution accumulation in an organism within a given time; usually represented as substitutions per site per year. The dates of separation between two phylogenetic lineages, for example, the split between two species. corroborated by the incorporation of 18th century European MTBC genomes in the dating analysis 26, 52, 123 . In molecular phylogenies, the length of each individ ual branch usually reflects the number of substitutions acquired by an organism within a given period of time and, as such, varying branch lengths should represent heterochronous sequences. Therefore, an important pre requisite for a robust dating analysis is that the nucleo tide substitution rate of the species whose phylogeny is to be dated behaves in a 'clock like' manner, meaning that phylogenetic branch lengths correlate with archaeological dates or sampling times. Such relationships can be assessed through date randomization and root-to-tip regression tests (Fig. 3) . The former is used to assess the effect of arbi trary exchange of phylogenetic tip dates on the nucleo tide substitution rate and divergence date estimates 124 , whereas the latter is used for estimation of a correlation coefficient (r) and coefficient of determination (R 2 ) by relating the tip date of each taxon to its SNP distance from the tree root (using, for example, the program Temp Est 125 ). The resulting values determine whether there is a temporal signal in the data and suggest whether branches within a phylogeny evolve at a constant rate, in which case a strict molecular clock 126 can be statistically tested, for example, using MEGA 99 or marginal likelihood estimations 127, 128 , and applied. If branches are affected by differences in their evolutionary rates, a relaxed clock 129 would be more appropriate. In general, a constant mole cular clock will rarely reliably describe the history of a microbial species, even more so for infectious pathogens whose replication rates vary between active and latent or between epidemic and dormant phases 120, 130 . In certain cases, neither of the two models may fit the data, such as when extensive rate variation weakens the temporal signal. This challenge was encountered in initial attempts to date the Y. pestis phylogeny using too few ancient cali bration points 36, 130 . Similar limitations can arise when the evolutionary history of a microorganism is vastly affected by recombination, as observed for HBV 44, 53 , although HBV molecular dating was recently attempted using a different genomic data set and suggested that the currently explored diversity of Old and New World pri mate lineages (including all human genotypes) may have emerged within the last 20,000 years 43 . Molecular dating analysis requires the use of an appro priate demographic model for the available data, which can be determined through model testing approaches (for example, through marginal likelihood estimations 127, 128 ). Currently, the most widely used models for estimating dates of divergence are the coalescent constant size 131 , which assumes a continuous population size history -and is unrealistic for epidemic pathogens -and the coalescent skyline 132 , which can estimate effective population size (N e ) changes over time. Moreover, the birth-death demographic model 133, 134 , which is cur rently unexplored within aDNA frameworks, may prove an insightful analysis tool in the future. This model has shown its applicability on comprehensive pathogen data sets from modern day epidemic contexts 133 . It has the ability to incorporate prior knowledge on incom plete sampling proportions and sampling biases within a data set, a frequent caveat of aDNA studies that is currently unaccounted for within molecular dating analy ses. Finally, recently developed fast dating algorithms should also be noted, for example, the least squared dating (LSD) program, which does not use constrained demographic models but can handle uncorrelated rate variation among phylogenetic branches and has shown potential for analysing large genomic data sets 135 . The pathogen best studied using aDNA analysis so far is Y. pestis, the causative agent of plague. To date, 38 ancient genomes of this bacterium have been published 20,30-39 (Fig. 4) , and their analyses have yielded valuable infor mation on past pandemic emergence as well as in depth microbial evolution. Integration of such knowledge into human population frameworks has provided key insights into the association of human migrations and infectious disease transmission in the past 31, 34 . This sec tion describes the evolutionary history of Y. pestis with the aim of demonstrating aspects of its emergence and spread as revealed through aDNA research. Plague is a well defined infectious disease caused by the Gram negative bacterium Y. pestis, which belongs to the fam ily Enterobacteriaceae. It evolved from a close relative, Yersinia pseudotuberculosis, which is an environmental enteric diseasecausing bacterium 136 . Although the two species are clearly distinguishable in terms of their vir ulence potential and transmission mechanisms, their nucleotide genomic identity reaches 97% among chromo somal protein coding genes 137 . In addition, they share the virulence plasmid pCD1, which encodes a type III secretion system common to three known pathogenic Yersinia: Y. pestis, Y. pseudotuberculosis and Yersinia enterocolitica. The distinct transmission mechanism and pathogenicity of Y. pestis are conferred by the unique acquisition of two plasmids, pPCP1, which contributes to the invasive potential of the bacterium 138 , and pMT1, which is involved in flea colonization 139, 140 , as well as by chromosomal gene pseudogenization or loss throughout its evolutionary history 141 . Y. pestis is not human adapted. Its primary hosts are sylvatic rodents such as marmots, mice, great gerbils, voles and prairie dogs, among others, in which it is continuously or intermittently maintained in so called reservoirs or foci [142] [143] [144] . Its global distribution includes numerous rodent species 144, 145 and encompasses regions in eastern Europe, Asia, Africa and the Americas (Fig. 4) , where the bacterium persists in active foci, some of which have existed for centuries or even millennia 31, 33, 34, 37, 130 . Y. pestis transmission among hosts is facilitated by a flea vector (Fig. 5) . The best yet characterized is the oriental rat flea, Xenopsylla cheopis, although others are also known to play important roles in Y. pestis transmission 142, 144, 146 . Notably, recent modelling infer ences suggest important roles for ectoparasites such as body lice and human fleas in its propagation during human epidemics 147 . Landmark studies investigating the classical model of transmission have shown that Y. pestis has the unique ability to colonize and form a biofilm within the flea, which blocks a portion of its A test that involves random shuffling of calibration points (tip dates) across a molecular phylogeny to evaluate the effect of randomizations compared to true data on the nucleotide substitution rate estimates. A test that uses a linear correlation to determine the relationship between branch lengths and sampling times within a time-dependent phylogeny. A mathematical model that aims to explain the size and density of a population over time. www.nature.com/nrg 332 | JuNe 2019 | volume 20 foregut, the proventriculus (Fig. 5 ). This phenotype is determined by the unique acquisition and activity of certain genomic loci in Y. pestis, namely, the Yersinia murine toxin (ymt) gene, which is present on the pMT1 plasmid 140, 141 and facilitates colonization of the arthropod midgut 141 . In addition, it is dependent on the pseudogenization of certain genes, namely, the biofilm downregulators rcsA, PDE2 (also known as rtn), PDE3 (also known as y3389) 141 and the ure ase gene ureD 148, 149 , which are, by contrast, active in Y. pseudotuberculosis. The biofilm prevents a blood meal from entering the flea's digestive tract, leaving it starving; as a result, the insect intensifies its feeding behaviour and promotes bacterial transmission to un infected hosts [150] [151] [152] . This continuous transmission cycle among fleas and rodents, also called the enzootic phase of maintenance (Fig. 5) , is thought to drive the preser vation of plague foci around the world and is depen dent on environmental and climatic factors as well as on host population densities 142, [153] [154] [155] . Disruption of this equilibrium for reasons that are not well understood can cause disease eruption among susceptible rodent species, leading to so called plague epizootics 142 (Fig. 5) . During that time, marked reductions in the rodent pop ulations force fleas to seek alternative hosts, which can lead to infections in humans and, as a result, trigger the initiation of epidemics or pandemics. Plague manifestation in humans has three disease forms, namely, bubonic, pneumonic and septicaemic 156 . Bubonic plague is the most common form of the disease and can cause up to 60% mortality when left untreated 157 . Subsequent to the bite of an infected flea, bacteria travel to the closest lymph node, where excessive replica tion occurs, giving rise to large swellings, the so called buboes. In addition, following primary bubonic plague, bacteria can disseminate into the bloodstream to cause septicaemia (secondary septicaemic plague) and to the lungs, causing secondary pneumonic disease. Both forms are highly lethal disease presentations and cause nearly 100% mortality when left untreated. Only the pneumonic form can result in direct human tohuman transmission. Early evolution: plague in prehistory. The time of divergence between Y. pestis and Y. pseudotuberculosis has been difficult to determine given the wide temporal interval produced by recent molecular dating attempts 92, 130, [169] [170] [171] [172] [173] [187] [188] [189] [190] [191] [192] [193] [194] [195] [196] [197] [198] [199] 200 , are shown as grey circles within their geographical country or region of isolation, and the size of each circle is proportional to the number of strains sequenced from each location (number indicated when more than one genome is shown). The areas highlighted in brown are regions that contain active plague foci as determined by contemporary or historical data. ybp, years before present. Adapted with permission from the 'Global distribution of natural plague foci as of March 2016' from https://www.who.int/csr/disease/plague/Plague-map-2016.pdf. based on aDNA data (13,000-79,000 years before present (ybp)) 33, 34 . Nevertheless, Y. pestis identification in human remains from Neolithic and Bronze Age Eurasia suggests that it caused human infections during these periods and originated more than 5,000 years ago 31, 33, 34 . These data have revealed important details about the early evolution of the bacterium. Genomic and phylogenetic analyses have shown that strains from the Late Neolithic and Bronze Age (LNBA) occupy a basal lineage in the Y. pestis phylogeny, and a recent study suggests the presence of even more basal variants in Neolithic Europe 31 (Fig. 6 ). Such analyses have demonstrated that, during its early evolution, the bacterium had not yet acquired impor tant virulence factors consistent with the complex trans mission cycle common to historical and extant strains. One of these genes is ymt, whose absence has been asso ciated with an inability for flea midgut colonization in Y. pestis 141 . In addition, these strains possess the active forms of the rcsA, PDE3, PDE2 and ureD genes, which suggests an impaired ability towards biofilm formation and blockage of the flea's proventriculus 141, 149 . Finally, they possess an active flagellin gene (flhD), which is pres ent as a pseudogene in all other Y. pestis, as it is a potent inducer of the innate immune response of the host 158 . As a result, during its initial evolutionary stages, Y. pestis may have been unable to efficiently transmit via a flea vector. Flea borne transmission of Y. pestis is a known prerequisite for bubonic plague development 141 ; hence, it has been suggested that this disease phenotype was not present during prehistoric times 33, 159 . In addition, these results have raised uncertainty regarding the pos sible vector and host mammalian species of the bacte rium. The Bronze Age in Eurasia was a period of intense human migrations, which shaped the genomic landscape of modern day Europe 85, 160 . Remarkably, the Y. pestis LNBA lineage was shown to mirror human movements during that time 34 and was found in regions that do not host wild reservoir populations today (Fig. 4) . The wide geographical distribution of these strains, their supposed limited bubonic disease potential and their relationship with human migration routes might together be indica tive of a different reservoir host species compared to wild rodents that have a central role in plague transmission in areas such as Central and East Asia, where the disease is endemic today. Nevertheless, an alternative mode of flea transmis sion, termed the early phase transmission, which occurs during the initial phases of infection and was suggested to be biofilm independent 161 , should also be considered as a possible way of Y. pestis propagation during its early evolution 34 . Although this transmission mechanism is currently not well understood, its comparative mode and efficiency in different rodent species have recently started to be assessed 162 evidence showing the full capacity for flea colonization similar to modern and historic strains was identified in two 3,800year old skeletons from the Samara region of modern day Russia 37 . Although this strain was shown to occupy a phylogenetic position among modern Y. pestis lineages (Fig. 6) , molecular dating analysis indicated that it originated ~4,000 years ago, suggesting that it over lapped temporally with the other Bronze Age strains that lacked the genetic prerequisites for arthropod transmis sion. Similar characteristics were previously identified in a low coverage 3,000year old isolate from modern day Armenia 33 , which suggests that multiple forms of the bacterium were circulating in Eurasia between 5,000 and 3,000 years ago that may have had different transmission cycles and produced different disease phenotypes. As the propagation mechanisms of those strains are still uncer tain, and the exact timing of flea adaptation in Y. pestis is unknown, additional metagenomic screening from human and animal remains may provide relevant infor mation on disease reservoirs and hosts across Neolithic and Bronze Age Eurasia. It is becoming increasingly apparent that, aside from plague, other infectious diseases, such as those caused by HBV 43,44 and B19V 45 (TAble 1) , were circulating dur ing the same time periods. Further pathogen screening coupled with a temporal assessment of human immune associated genomic variants 84 may reveal key aspects of disease prevalence and susceptibility during this pivotal period of human history. After the Bronze Age, bubonic plague has been associated with three historically recorded pandem ics. The earliest accounts of the so called first plague pandemic, which began with the Plague of Justinian (541 ce), suggest that it erupted in northern Africa in the mid6th century ce 163, 164 and subsequently spread through Europe and the vicinity until ~750 ce. The sec ond historically recorded plague pandemic began with the infamous Black Death (1346-1353 ce) 96 and con tinued with outbreaks in Europe until the 18th century ce. The most recent third plague pandemic began in the mid19th century in the Yunnan province of China, and it was during that time that Alexandre E. J. Yersin first described the bacterium in Hong Kong, in 1894 (Fig. 1) and has persisted until today in active foci in Africa, Asia and the Americas. Although the majority of mod ern plague cases derive from strains disseminated in this global dispersal, the pandemic is considered to have largely subsided since the 1950s 165 . The association of Y. pestis with the two earlier pan demics has, until recent years, been contentious. On the basis of their serological characterization, modern strains were traditionally grouped into three distinct biovars, namely, 'antiqua' , 'medievalis' and 'orientalis' , according to their ability to ferment glycerol and reduce nitrate 165, 166 . In addition, historical accounts of the dis ease seemed to correlate with the supposed distinct geographical distributions of these biovars 166 , and their phylogenetic relationships, as inferred from MLST data, reinforced the hypothesis that each was responsible for a single pandemic 136 . By contrast, later studies identified additional, atypical biovars 167 , and more robust phylo genetic analysis suggested that phylogeography does not correlate clearly with the phenotypic distinctions described between these bacterial populations 92, 130, 168 . Recent genomic analyses have revealed high genetic diversity of the bacterium in East Asia, which invaria bly led to the assumption that Y. pestis emerged there 130 . However, a strong research focus on the diversity of the bacterium in these endemic regions, mainly China, has contributed to a profound sampling bias in the available modern data (Fig. 4) . More recent investigations have revealed previously uncharacterized genetic diversity in the Caucasus region and in the central Asian steppe that ought to be further explored [169] [170] [171] [172] (Fig. 4) . Currently, the evolutionary tree of the bacterium is characterized by five main phylogenetic branches (Fig. 6) . The most ances tral, branch 0, includes strains distributed across China, Mongolia and the areas encompassing the former Soviet Union. The more phylogenetically derived branches 1-4 were formed through a rapid population expan sion event and are today found in Asia, Africa and the Americas 130 . Their wide distribution mainly reflects the geographical breadth of branch 1, which is associated with the third plague pandemic that spread worldwide during the 19th and 20th centuries 92 and is still respon sible for more confined epidemics such as those reported in Madagascar 173 . The analysis of aDNA from historical epidemic contexts has generated important information regard ing the evolutionary history of plague. The recovery of Y. pestis DNA via PCR from remnants of human den tal pulp suggested the involvement of the bacterium in both the first and second pandemics; however, these results were difficult to authenticate 8, 174, 175 . Subsequent PCR based SNP typing of ancient specimens offered some phylogenetic resolution and revealed an expected ancestral placement of medieval strains in the Y. pestis phylogeny [62] [63] [64] . More recently, full characterization and authentication of the bacterium were achieved using plasmid and whole genome enrichment coupled with NGS 17, 20, 35, 36 . Historical accounts of the first plague pandemic (6th to 8th centuries ce) suggest that the disease expanded mainly across the Mediterranean basin; however, its exact breadth and impact have been difficult to assess given the limited availability of historical and archaeo logical data, with the latter being currently under revision 176 . Two recent studies have reconstructed 6th century Y. pestis genomes from southern Germany 35,36 (Fig. 4) , a region that lacked historical documentation of the pandemic. Phylogenetic analysis showed that both genomes belong to a lineage that is today extinct and is closely related to strains from modern day China 35, 36 , which suggests the possibility of an East Asian origin of the first pandemic. This hypothesis was recently reinforced by the publication of a 2nd century to 3rd century Y. pestis genome from the Tian Shan mountains of modern day Kyrgyzstan 39 , which shares a common ancestor with the Justinianic plague lineage (Figs 4,6) . However, given the >300year age difference between these strains 35, 36, 39 , as well as the aforementioned East Asian sampling bias of modern Y. pestis data 130 , the geographical origin of the pandemic remains hypothet ical. Retrieval of additional Y. pestis strain diversity from that time period, particularly from areas known to have played an important role in the entry of this bacterium into Europe, that is, the eastern Mediterranean region, may hold clues about its putative source. The beginning of the second plague pandemic, 600 years later, was marked by the notorious Black Death of Europe (1346-1353 ce), estimated to have caused an up to 60% reduction of the continental pop ulation in only 5 years 96 . Historical records suggest that the first outbreaks occurred in the Lower Volga region of Russia, and the disease then spread into southern Europe through the Crimean peninsula 96 . Initial analy sis of Y. pestis via PCR from victims of the Black Death revealed a distinct phylogenetic positioning of two mid tolate14th century strains and led to the proposal that the disease entered the continent through independent pulses 64 . By contrast, whole genome analysis of ancient strains from western, northern and southern Europe demonstrated a lack of Y. pestis diversity during the Black Death, which suggests its fast spread through the continent and favours a single wave entry model of the bacterium into Europe 20,30,38 , although the possible presence of additional strain diversity during that time has recently been explored 30 . Intriguingly, the phylo genetic positioning of the Black Death Y. pestis genomes places them on branch 1, only two nucleotide substitu tions away from the 'star like' diversification of branches 1-4 (Fig. 6) , which gave rise to most of the strain diversity identified around the world today 38, 130 . After the Black Death, plague epidemics continued to affect Europe until the 18th century 177, 178 . Inferred climatic data from tree ring records in central Asia and Europe have recently suggested that such epidem ics were likely caused by multiple introductions of the bacterium into Europe as a result of climate driven disruptions of pre existing Asian reservoirs 179 . By con trast, ancient genetic and genomic evidence supports the persistence of the disease in Europe for 400 years after the Black Death 32, 38, 62 . Analysis of Y. pestis strains spanning from the late 14th to the 18th century ce has revealed the formation of at least two European lin eages that were responsible for the ensuing medieval epidemics (Fig. 6 ). Both lineages derive from the Black Death Y. pestis strain identified in 14th century west ern, northern and southern Europe 30, 32, 38 , suggesting that they likely arose locally. The first lineage survives today and gave rise to modern branch 1 strains 30, 38 (which are associated with the third plague pandemic), suggesting the European Black Death as a source for modern day epidemics 38 . The second lineage has not been identified among present day diversity and currently encompasses strains from 16th century Germany 38 and 18th century France (Great Plague of Marseille, 1720-1722 ce) (Fig. 6) . These phylogenetic patterns are consistent with a con tinuous persistence of the bacterium in Europe dur ing the second plague pandemic. In addition, they are supported by analyses of historical records that suggest the existence of plague reservoirs in the continent until the 18th century ce 180 . Y. pestis is absent from most of Europe today; specif ically, no active foci exist west of the Black Sea. Plague is thought to have disappeared from most of Europe at the end of the second pandemic (18th century ce). This finding is striking given the thousands of outbreaks that were recorded in the continent until that time 177, 178 . The reasons for its disappearance are unknown, although numerous hypotheses have been put forward 181 , includ ing a change in domestic rodent populations in Europe, namely, the replacement of the black rat, Rattus rattus, by the brown rat, Rattus norvegicus 181 ; an acquired plague immunity among humans and/or rodents 181 (although this hypothesis requires an update to accommodate the recent identification of Y. pestis in Europe 5,000 years ago 31, 33, 34 and the involvement of the bacterium in the first plague pandemic 35, 36 ); the increased living standards such as the better nutrition and hygienic conditions at the beginning of the Early Modern Era, which may have contributed to improved overall health conditions in Europe and likely decreased the number of rats and ecto parasites in human environments 181, 182 ; and the poten tial disruption of the European wild rodent ecological niche owing to habitat loss and industrialization start ing in 1700 ce 180 . Given the contribution that molecular data can offer in these discussions, future research on ancient sources of Y. pestis DNA will be instrumental in further revealing the history of one of humankind's most devastating pathogens. The analysis of ancient pathogen genomes has afforded promising views into past infectious disease history. For Y. pestis, aDNA exploration of its evolutionary past has revealed how a predominantly environmental bac terium and opportunistic gastroenteric pathogen deve loped into an extremely virulent form by acquisition of only a few virulence factors. We eagerly await revelations on a similar scale for other important pathogens that are expected to arise from deep temporal sampling and genomic reconstruction, as made possible through the recent advancements discussed here. Integration of ancient pathogen genomes into disease modelling and human population genetic frameworks, as well as their analysis alongside the information offered by the archaeological, historical and palaeopathological records, will help build a more interdisciplinary and com plete picture of host-pathogen interactions and human evolutionary history over time. Published online 5 April 2019 The origins of agriculture: population growth during a period of declining health Emerging and re-emerging infectious diseases: the third epidemiologic transition Identification of Pathological Conditions in Human Skeletal Remains 2nd edn The Global History of Paleopathology: Pioneers and Prospects Pre-Columbian tuberculosis in Northern Chile: molecular and skeletal evidence Identification of Mycobacterium tuberculosis DNA in a pre-Columbian Peruvian mummy Molecular analysis of skeletal tuberculosis in an ancient Egyptian population Detection of 400-year-old Yersinia pestis DNA in human dental pulp: an approach to the diagnosis of ancient septicemia The use of the polymerase chain reaction (PCR) to detect Mycobacterium tuberculosis in ancient skeletons Ancient DNA: extraction, characterization, molecular cloning, and enzymatic amplification Temporal patterns of nucleotide misincorporations and DNA fragmentation in ancient DNA The study provides a quantitative description of aDNA-associated patterns of nucleotide misincorporation and fragmentation that are currently used as primary authentication criteria Ancient DNA: do it right or not at all Absence of Yersinia pestisspecific DNA in human teeth from five European excavations of putative plague victims No proof that typhoid caused the Plague of Athens (a reply to Papagrigorakis et al.) Genome sequencing in microfabricated high-density picolitre reactors Targeted enrichment of ancient pathogens yielding the pPCP1 plasmid of Yersinia pestis from victims of the Black Death The Neandertal genome and ancient DNA authenticity Mining metagenomic data sets for ancient DNA: recommended protocols for authentication The study describes the first whole-genome sequence of an ancient bacterial pathogen through the use of high-throughput sequencing Genetic analyses from ancient DNA Optimal ancient DNA yields from the inner ear part of the human petrous bone Comparing ancient DNA preservation in petrous bone and tooth cementum Ancient pathogen DNA in human teeth and petrous bones Pre-Columbian mycobacterial genomes reveal seals as a source of New World human tuberculosis Ancient genomes reveal a high diversity of Mycobacterium leprae in medieval Europe The study presents the first de novo assembled ancient pathogen genome and an analysis of M Historic Treponema pallidum genomes from Colonial Mexico retrieved from archaeological remains Integrative approach using Yersinia pestis genomes to revisit the historical landscape of plague during the Medieval Period Emergence and spread of basal lineages of Yersinia pestis during the Neolithic Decline Eighteenth century Yersinia pestis genomes reveal the long-term persistence of an historical plague focus Early divergent strains of Yersinia pestis in Eurasia 5,000 years ago The study describes Y. pestis genomes from Bronze Age human remains and provides a chronological timing of virulence determinant acquisition during the early evolution of the bacterium The Stone Age plague and its persistence in Eurasia A high-coverage Yersinia pestis genome from a sixth-century justinianic plague victim Yersinia pestis and the Plague of Justinian 541-543 AD: a genomic analysis Analysis of 3800-year-old Yersinia pestis genomes suggests Bronze Age origin for bubonic plague Historical Y. pestis genomes reveal the European Black Death as the source of ancient and modern plague pandemics 137 ancient human genomes from across the Eurasian steppes Genomic blueprint of a relapsing fever pathogen in 15th century Scandinavia This paper presents the metagenomic tool MALT and is the first case study to demonstrate metagenomic detection of ancient pathogens in the absence of prior knowledge on the causative agent of an epidemic Plasmodium falciparum malaria in 1st−2nd century CE southern Italy Ancient hepatitis B viruses from the Bronze Age to the Medieval period The studies by Mühlemann (Nature, 2018) and Krause-Kyora (eLife, 2018) present a time transect of HBV genomes, spanning from the Neolithic period to the medieval period, and provide an overview of the HBV population history across millennia Ancient human parvovirus B19 in Eurasia reveals its long-term association with humans This study provides an analysis of the composition of human dental calculus from ancient individuals, showing the presence of oral microbiome bacterial DNA, periodontal pathogen DNA and proteins associated with host immunity Recovery of a medieval Brucella melitensis genome using shotgun metagenomics A molecular portrait of maternal sepsis from Byzantine Troy The study provides insights into the genomic history of H. pylori over several millennia through a population genomic analysis of a Copper Age strain against a worldwide data set 17th century variola virus reveals the recent history of smallpox Variola virus in a 300-year-old Siberian mummy Eighteenth-century genomes show that mixed infections were common at time of peak tuberculosis in Europe The paradox of HBV evolution as revealed from a 16th century mummy Tracing hepatitis B virus to the 16th century in a Korean mummy Second-pandemic strain of Vibrio cholerae from the Philadelphia cholera outbreak of 1849 Mitochondrial DNA from the eradicated European Plasmodium vivax and P. falciparum from 70-year-old slides from the Ebro Delta in Spain and 'patient 0'HIV-1 genomes illuminate early HIV/AIDS history in North America Characterization of the 1918 influenza virus polymerase genes The rise and fall of the Phytophthora infestans lineage that triggered the Irish potato famine Reconstructing genome evolution in historic samples of the Irish potato famine pathogen Screening ancient tuberculosis with qPCR: challenges and opportunities Genotyping Yersinia pestis in historical plague: evidence for long-term persistence of Y. pestis in Europe from the 14th to the 17th century Yersinia pestis DNA from skeletal remains from the 6th century AD reveals insights into Justinianic Plague Distinct clones of Yersinia pestis caused the black death Parallel detection of ancient pathogens via array-based DNA capture Ancient pathogen DNA in archaeological samples detected with a microbial detection array Fast and accurate long-read alignment with Burrows-Wheeler transform Basic local alignment search tool Reference sequence (RefSeq) database at NCBI: current status, taxonomic expansion, and functional annotation Metagenomic microbial community profiling using unique clade-specific marker genes metaBIT, an integrative and automated metagenomic pipeline for analysing microbial profiles from high-throughput sequencing shotgun data Kraken: ultrafast metagenomic sequence classification using exact alignments A robust framework for microbial archaeology Complications in the study of ancient tuberculosis: Presence of environmental bacteria in human archaeological remains DNA sequences from multiple amplifications reveal artifacts induced by cytosine deamination in ancient DNA 0: fast approximate Bayesian estimates of ancient DNA damage parameters Removal of deaminated cytosines and detection of in vivo methylation in ancient DNA Partial uracil-DNA-glycosylase treatment for screening of ancient DNA Hybrid selection of discrete genomic intervals on custom-designed microarrays for massively parallel sequencing Targeted investigation of the Neandertal genome by array-based sequence capture DNA analysis of an early modern human from Tianyuan Cave Application and comparison of large-scale solution-based DNA capture-enrichment methods on ancient Experimental conditions improving in-solution target enrichment for ancient DNA Genome-wide patterns of selection in 230 ancient Eurasians This study delineates large-scale population migrations into Europe during the Bronze Age by analysis of human genome-wide data of Genomic insights into the origin of farming in the ancient Near East Genetic origins of the Minoans and Mycenaeans Language continuity despite population replacement in Remote Oceania Ancient human genome sequence of an extinct Palaeo-Eskimo The complete genome sequence of a Neanderthal from the Altai Mountains A high-coverage genome sequence from an archaic Denisovan individual Yersinia pestis genome sequencing identifies patterns of global phylogenetic diversity The study shows a possible co-expansion of M. tuberculosis among human populations during out-of-Africa migrations The Bioarchaeology of Tuberculosis: a Global Perspective on a Re-Emerging Disease The Black Death Transformed: Disease and Culture in Early Renaissance Europe The Black Death Advances in Human Palaeopathology Past Human Infections MEGA7: molecular evolutionary genetics analysis version 7.0 for bigger datasets A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood RAxML version 8: a tool for phylogenetic analysis and post-analysis of large phylogenies IQ-TREE: a fast and effective stochastic algorithm for estimating maximum-likelihood phylogenies MrBayes 3: Bayesian phylogenetic inference under mixed models Application of phylogenetic networks in evolutionary studies Ancient DNA study reveals HLA susceptibility locus for leprosy in medieval Europeans Ancient skeletal evidence for leprosy in India Possible cases of leprosy from the Late Copper Age (3780-3650 cal BC) in Hungary Leprosy and the adaptation of human toll-like receptor 1 The study describes a 12th century Salmonella enterica subsp. enterica serovar Paratyphi C genome and its analysis alongside a comprehensive data set of thousands of S ClonalFrameML: efficient inference of recombination in whole bacterial genomes RDP4: detection and analysis of recombination patterns in virus genomes Inference of population structure using multilocus genotype data Inference of population structure using dense haplotype data TREE-PUZZLE: maximum likelihood phylogenetic analysis using quartets and parallel computing On the origin of leprosy DNA sequencing costs: data from the NHGRI Genome Sequencing Program (GSP). genome PATRIC, the bacterial bioinformatics database and analysis resource The paper introduces a web-based platform that performs genome assembly and multilocus sequence typing analysis and can be used for the retrieval of large data sets on enteric bacteria GrapeTree: visualization of core genomic relationships among 100,000 bacterial pathogens Genome-scale rates of evolutionary change in bacteria BEAST: Bayesian evolutionary analysis by sampling trees BEAST 2: a software platform for Bayesian evolutionary analysis The study describes the first sequenced ancient M. tuberculosis genome and shows the presence of mixed infections in 18th century Europe The performance of the date-randomization test in phylogenetic analyses of time-structured virus data Exploring the temporal structure of heterochronous sequences using TempEst (formerly Path-O-Gen) Bayesian analysis of elapsed times in continuous-time Markov chains Model selection and parameter inference in phylogenetics using nested sampling Improving the accuracy of demographic and molecular clock model comparison while accommodating phylogenetic uncertainty Relaxed phylogenetics and dating with confidence The study presents a comprehensive Y. pestis modern genomic data set from East Asia and demonstrates extensive clock-rate variations across the Y Bayesian coalescent inference of past population dynamics from molecular sequences Birth-death skyline plot reveals temporal changes of epidemic spread in HIV and hepatitis C virus (HCV) Estimating the basic reproductive number from viral sequence data Fast dating using least-squares criteria and algorithms Yersinia pestis, the cause of plague, is a recently emerged clone of Yersinia pseudotuberculosis Insights into the evolution of Yersinia pestis through whole-genome comparison with Yersinia pseudotuberculosis Early emergence of Yersinia pestis as a severe respiratory pathogen Murine toxin of Yersinia pestis shows phospholipase D activity but is not required for virulence in mice Role of Yersinia murine toxin in survival of Yersinia pestis in the midgut of the flea vector Retracing the evolutionary path that led to flea-borne transmission of Yersinia pestis This study presents a functional description of genes associated with flea-dependent colonization and transmission in Y. pestis Recovery of the Black-Footed Ferret Progress and Continuing Challenges. United States Geological Survey Scientific Investigations Report Natural history of plague: perspectives from more than a century of research Intraspecific diversity of Yersinia pestis Yersinia pestis in small rodents Comparative ability of oropsylla montana and xenopsylla cheopis fleas to transmit Yersinia pestis by two different mechanisms Human ectoparasites and the spread of plague in Europe during the second pandemic Silencing and reactivation of urease inYersinia pestis is determined by one G residue at a specific position in the ureD gene Silencing urease: a key evolutionary step that facilitated the adaptation of Yersinia pestis to the flea-borne transmission route Advances in Yersinia Research Transmission of Yersinia pestis from an infectious biofilm in the flea vector Observations on the mechanism of the transmission of plague by fleas Metapopulation dynamics of bubonic plague Plague dynamics are driven by climate variation Predictive thresholds for plague in Kazakhstan Yersinia pestisetiologic agent of plague The Genus Yersinia: From Genomics to Function Ecological opportunity, evolution, and the emergence of fleaborne plague This study presents human genome-wide data from the Bronze Age period that was used to delineate large-scale migrations across Eurasia. The same data set was later used for pathogen screening and could show the presence of Y. pestis, HBV and B19V in those populations The role of early-phase transmission in the spread of Yersinia pestis Infectious blood source alters early foregut infection and regurgitative transmission of Yersinia pestis by rodent fleas The Fate of Rome: Climate, Disease, and the End of an Empire The Justinianic plague: origins and effects Plague (World Health Organization Monograph Series Varietes de l'espece Pasteurella pestis: nouvelle hypothese Genetics of metabolic variations between Yersinia pestis biovars and the proposal of a new biovar, microtus Microevolution and history of the plague bacillus, Yersinia pestis Phylogeny and classification of Yersinia pestis through the lens of strains from the plague foci of commonwealth of independent states Yersinia pestis strains of ancient phylogenetic branch 0.ANT are widely spread in the high-mountain plague foci of Kyrgyzstan Nineteen whole-genome assemblies of Yersinia pestis subsp. microtus, including representatives of Biovars caucasica, talassica, hissarica, altaica, xilingolensis, and ulegeica Yersinia pestis strains isolated in the caucasus region Temporal phylogeography of Yersinia pestis in Madagascar: insights into the longterm maintenance of plague Yersinia pestis Orientalis in remains of ancient plague patients Orientalis-like Yersinia pestis, and plague pandemics Tracking mass death during the fall of rome's empire (I) Digitizing historical plague Les Hommes et la Peste en France et dans les Pays Européens et Méditerranéens. Tome II. -Les Hommes Face à la Peste Climate-driven introduction of the Black Death and successive plague reintroductions into Pandemic Disease in the Medieval World: Rethinking the Black Death The disappearance of plague: a continuing puzzle Mortality risk and survival in the aftermath of the medieval Black Death Complete mitochondrial genome sequence of a Middle Pleistocene cave bear reconstructed from ultrashort DNA fragments Illumina sequencing library preparation for highly multiplexed target capture and sequencing Single-stranded DNA library preparation from highly degraded DNA using T4 DNA ligase A program for annotating and predicting the effects of single nucleotide polymorphisms, SnpEff: SNPs in the genome of Drosophila melanogaster strain w1118 Six whole-genome assemblies of Yersinia pestis subsp. microtus bv. ulegeica (Phylogroup 0. PE5) strains isolated from mongolian natural plague foci Complete genome sequence of Yersinia pestis strain 91001, an isolate avirulent to humans Whole-genome sequencing and comparative analysis of Yersinia pestis, the causative agent of a plague outbreak in northern Peru Human plague associated with Tibetan sheep originates in marmots Thirty-two complete genome assemblies of nine Yersinia species, including Y. pestis, Y. pseudotuberculosis, and Y. enterocolitica Nine whole-genome assemblies of Yersinia pestis subsp. microtus bv. Altaica strains isolated from the altai mountain natural plague focus (No. 36) in Russia The Genus Yersinia: From Genomics to Function Complete genome sequence of Yersinia pestis strains Antiqua and Nepal516: evidence of gene reduction in an emerging pathogen Genome sequence of Yersinia pestis, the causative agent of plague Draft genome sequences of Yersinia pestis strains from the 1994 plague epidemic of Surat and 2002 Shimla outbreak in India Comparative genomics of 2009 seasonal plague (Yersinia pestis) in New Mexico Complete genome sequences of Yersinia pestis from natural foci in China A North American Yersinia pestis draft genome sequence: SNPs and phylogenetic analysis Yersinia pestis halotolerance illuminates plague reservoirs Mycobacterium leprae genomes from a British medieval leprosy hospital: towards understanding an ancient epidemic Characterization of the reconstructed 1918 Spanish influenza pandemic virus The authors thank C. Warinner for her valuable comments to the manuscript and M. Keller for his contributions in assembling comprehensive meta-information for the Y. pestis modern genomic data set. In addition, the authors thank all members of the Molecular Paleopathology and Computational Pathogenomics groups at the Max Planck Institute for the Science of Human History for insightful discussions during meetings. Moreover, they are grateful to M. O'Reilly, H. Shell and R. Barquera for extensive assistance with the graphics. This work was supported by the Max Planck Society. M.A.S. researched the literature and wrote the article. All authors provided substantial contributions to discussions of the content and reviewed and/or edited the manuscript. The authors declare no competing interests. Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Nature Reviews Genetics thanks E. Willerslev and other anonymous reviewer(s) for their contribution to the peer review of this work.