19 Emerging Technologies in DNA Sequencing

Chromosome 21 is the smallest human chromosome and represents a model for physical mapping of the entire human genome. Three copies of this autosome cause Down syndrome, the most frequent genetic disorder associated with significant mental retardation. Five years ago, in the framework of the international human genome project, and as an integrated part of the chromosome 21 community, a consortium, of academic groups from Japan and Germany was formed to map and sequence this chromosome. Several whole-genome and chromosome-specific DNA libraries were constructed for mapping purposes. Using a combination of nested-deletion [1] and shotgun sequencing approaches, our center determined and analyzed over 17 million bases of high-quality data. In total, over 33.5 million basepairs of DNA, distributed over four contigs, were sequenced with very high accuracy [2]. The largest of these contigs is nearly 25.5Mb long. Only three small clone gaps remain, which together comprise approximately 100kb. Thus, we achieved a coverage of 99.7% of 21q. In addition, 211,116 bp from the short arm were also sequenced. Analysis of the chromosome revealed far fewer genes than previously estimated. Instead of 800-1000 expected genes, we found only 225 genes, of which 127 are known genes and 98 are predicted genes. In addition, 59 pseudogenes were identified. The completion of the 21q sequence provides a unique resource for understanding the molecular pathophysiology of Down syndrome, as well as all other monogenic and complex disorders that map to this chromosome, including Alzheimer’s disease, leukemia, autoimmune disease, epilepsy and manic-depressive psychosis. It also stands as a structural framework from which the complete molecular architecture of the chromosome can be determined. With the announcement of the first “working draft” of the human genome by the international human genome sequencing consortium this past June, there has been a great interest in the total number of genes in the human genome[3]. Some of the current estimates range from as low as 34,000 genes to as high as 140,000 genes. The low gene-density of chromosome 21 was quite unexpected. Most striking is a 7-Mb region near the region near the centromere that contains only one gene. This region is much larger than the whole genome of some species, such as Escherichia coli, yet evolutionary processes permitted the existence of such a gene-poor SNA segment. This finding leads us to propose that there are additional large gene-less regions in other human/mammalian chromosomes. By combining the gene numbers of chromosomes 21 and 22 [4] (770genes; 2-3% of human genome) and assuming that together they represent the average gene content of the human genome, we estimate that the total number of human genes may be close to 40,000. As part of the international human genome sequencing project, we are also sequencing parts of chromosomes 11 and 18. Our high-throughput sequence production line generates about 10 Mb per day. In order to handle such a high volume of output, we have developed a system for automated data assembly, annotation and release. All the data is released through our website and through the DNA Database of Japan immediately according to the policies established by the international consortium. In addition to chromosomes 11, 18 and 21, we are also interested in the structure of the entire human genome. In collaboration with the University of Tokyo, we have developed a database called HGREP [5] (Human Genome Reconstruction Project) that contains working draft and finished sequences which cover more than 85% of the human genomic sequences. Sequence entries are aligned along the chromosomes based on sequence similarities to STS markers, BAC-end and other entry sequences. Further, biological features, such as genes, gene functions, repeats and CpG islands, are fully annotated on these entries. To identify evolutionally conserved and biologically important information in the genome, we are taking the strategy of comparative genomic sequence analysis between human and other related genomes such as those of rodents and primates. We are currently sequencing several regions of the mouse, including some regions that are counterpart to human chromosome 21, and are preparing genomic resources for sequencing of the chimpanzee. Through comparative methods, we are also trying to understand telomeric regions, structures at the end of chromosomes that play an important role in several biological functions and have been associated with human diseases like an oncogenesis and a cellular aging. The structure and function of all genes and their regulatory regions can only be fully understood by looking at how they interact under different circumstances and in several different backgrounds. Determining which genes and elements are human-specific will lead to a deeper understanding of who we are and will advance medical sciences drastically.

ChemInform is a weekly Abstracting Service, delivering concise information at a glance that was extracted from about 100 leading journals. To access a ChemInform Abstract of an article which was published elsewhere, please select a “Full Text” option. The original article is trackable via the “References” option.

1. Edward R.B. McCabe, MD, PhD* 1. 2. *Physician-in-Chief, Mattel Children’s Hospital at UCLA; Professor and Executive Chair, Department of Pediatrics, UCLA School of Medicine, Los Angeles, CA. Birth defects are the leading cause of infant mortality in the United States, representing more than 20% of all infant deaths. This infant mortality rate from birth defects exceeds that from sudden infant death syndrome, low-birthweight/short gestation, respiratory distress syndrome, and maternal complications. In addition, birth defects and genetic diseases represent major sources of morbidity for those who survive. As our ability increases to care effectively for those who have infectious diseases and other acute illnesses, individuals who have chronic illnesses due to genetic etiologies represent an increasing proportion of patients seen in the general pediatrician”s office. The Human Genome Project was initiated on October 1, 1990, and has a projected funding period of 15 years. The goal is to sequence the entire human genome, representing three billion base pairs that contain the coding sequences for approximately 75,000 genes. During the latter half of this century, investigations into the genetics of disease gathered increasing momentum. In addition to fundamental investigations into human genetics, technologic tools were developed that permitted large-scale genomic sequencing. These tools included the polymerase chain reaction (PCR), which permits amplification of hundreds of thousands or even millions of copies of DNA and requires only limited sequence data for its success; automated DNA sequencing, which allows increased sequence processing and decreased cost compared with manual methods; and improved information systems, which permit sophisticated analysis and assembly of the three billion base pairs of DNA in the human genome. Thus, the Human Genome Project represents the current chapter in our understanding, but it is neither the first nor the final chapter in this story. Once we know the sequences of all of the human genes, we must learn their functional roles in human development and disease pathogenesis. The Human Genome Project has been referred to as the “moon shot …

With the development of new sequencing techniques, whole genomes of many species have become available. This huge amount of data gives rise to new opportunities and challenges. These new sequences provide valuable information on relationships among species, e.g. genome recombination and conservation. One of the principal ways to investigate such information is multiple sequence alignment (MSA). Currently, there is large amount of MSA data on the internet, such as the UCSC genome database, but how to effectively use this information to solve classical and new problems is still an area lacking of exploration. In this thesis, we explored how to use this information in four problems, i.e. sequence orthology search problem, multiple alignment improvement problem, short read mapping problem, and genome rearrangement inference problem. For the first problem, we developed a EM algorithm to iteratively align a query with a multiple alignment database with the information from a phylogeny relating the query species and the species in the multiple alignment. We also infer the query's location in the phylogeny. We showed that by doing alignment and phylogeny inference together, we can improve the accuracies for both problems. For the second problem, we developed an optimization algorithm to iteratively refine the multiple alignment quality. Experiment results showed our algorithm is very stable in term of resulting alignments. The results showed that our method is more accurate than existing methods, i.e. Mafft, Clustal-O, and Mavid, on test data from three sets of species from the UCSC genome database. For the third problem, we developed a model, PhyMap, to align a read to a multiple alignment allowing mismatches and indels. PhyMap computes local alignments of a query sequence against a fixed multiple-genome alignment of closely related species. PhyMap uses a known phylogenetic tree on the species in the multiple alignment to improve the quality of its computed alignments while also estimating the placement of the query on this tree. Both theoretical computation and experiment results show that our model can differentiate between orthologous and paralogous alignments better than other popular short read mapping tools (BWA, BOWTIE and BLAST). For the fourth problem, we gave a simple genome recombination model which can express insertions, deletions, inversions, translocations and inverted translocations on aligned genome segments. We also developed an MCMC algorithm to infer the order of the query segments. We proved that using any Euclidian metrics to measure distance between two sequence orders in the tree optimization goal function will lead to a degenerated solution where the inferred order will be the order of one of the leaf nodes. We also gave a graph-based formulation of the problem which can represent the probability distribution of the order of the query sequences.

The 1000 Genomes Project: paving the way for personalized genomic medicine.

Background and objective Development of the next generation sequencing (NGS) platform was driven by the completion of the Human Genome Project in 2003. With the availability of NGS, the time taken for sequencing of humongous genomic regions was greatly reduced and data generated per unit DNA was also significantly increased. Though the cost to use NGS in a clinically setting is far from ideal, economically speaking, there is a significant decrease in the average cost per sequenced base. To validate findings of NGS on mutation detected for FBN1, TGFBR2, RAF1, RTEL1, LMNA, MID2, KCNK9, DMD, SMARCA2 and IQSEC2 by using gold standard, Sanger Sequencing.

At the First (1996) and Second (1997) International Strategy Meetings on Human DNA Sequencing, the attendees discussed and strongly endorsed the concept that human genomic DNA sequence produced by large-scale DNA sequencing centers funded by the Human Genome Project should release data as rapidly as possible. In the case of unfinished data, it was suggested that sequence assemblies of 1–2 kb in size should be released within 24 hr of generation. Finished data should be submitted to the public sequence databases on a similarly rapid time scale. Most of the funding agencies engaged in supporting the Human Genome Project have adopted policies that reflect the importance of rapid dissemination of genomic sequence data. Similarly the nucleotide sequence databases have made provision for the acceptance of the unfinished data as well as, of course, finished sequence data. At the 1998 meeting, the question was raised about similar data dissemination practices for genomic sequence data from other organisms. The attendees unanimously agreed that this was also critical and adopted the following statement as individual scientists. ‘‘As extensive determination of the genomic DNA sequence of several organisms proceeds, it is increasingly clear that sequence information has enormous and immediate scientific value, even prior to its final assembly and completion. Delaying the release of either unfinished or finished genomic DNA sequence data serves no useful purpose and actually has the effect of slowing the progress of research. Therefore, the attendees at the Third International Strategy Meeting on Human Genome Sequencing (Bermuda, Feb. 27–28, 1998) agreed unanimously to support, as individual scientists, the view that all publicly funded large-scale DNA sequencing projects, regardless of the organism, should deposit data immediately into the public domain, following the same guidelines that have previously been adopted by this group for human genomic sequence [see D.R. Bentley, Science 274: 533–534 (1996); http:// hugo.gdb.org/bermuda.htm]. The scientists attending this meeting will continue to adhere to these principles and urge all other scientists and policymaking groups involved in large-scale sequencing to adopt them as well.’’ This statement does not yet reflect, nor is it binding on, the policy of any funding agency. The attendees at the Third International Strategy Meeting on Human Genome Sequencing were Mark Adams, David Bentley, Bruce Birren, Martin Bobrow, Elbert Branscomb, Graham Cameron, Anthony V. Carrano, Ellson Chen, Francis Collins, David Cox, Mike Davies, Pieter de Jong, Richard Durbin, Glen Evans, Asao Fujiyama, Richard Gibbs, Eric Green, Phil Green, Mark Guyer, LaDeana Hillier, Ursula Hurtenbach, Elke Jordan, Jill Kent, Fumihiko Kikuchi, Steven Koonin, Jane Lamerdin, Eric Lander, Chris Martin, Richard McCombie, John McPherson, Michael Morgan, Richard Myers, James Ostell, Ari Patrinos, Jane Peterson, Bruce Roe, Jane Rogers, Andre Rosenthal, Yoshiyuki Sakaki, Barbara Skene, Edwin Southern, Jilly Steward, John Sulston, Mark Vaudin, Jeffrey Wadsworth, Robert Waterson, James Watson, Jean Weissenbach, Richard Wilson, and MarieLaure Yaspo.

Objective To study the effects of different pre-sequencing sample processing modes on the results of whole genome sequencing with high-throughput sequencing (HTS) by taking the largest RNA virus (human coronavirus, HCoV) as the representative. Methods Cell-cultured human coronavirus HCoV-OC43 strains were used as the representative samples and divided into different groups based on pre-sequencing processing modes as follows: untreated group, DNase and RNase treatment before nucleic acid extraction group, DNase treatment after nucleic acid extraction group, and DNase and RNase treatment before nucleic acid extraction and DNase treatment after nucleic acid extraction group. Nucleic acid samples of each group were analyzed by direct RNA sequencing (without amplification) and DNA sequencing after sequence independent single primer amplification (SISPA), respectively. Results No significant difference in viral genome coverage rates was observed between different groups. The highest genome coverage and sequencing accuracy were obtained in DNase treatment after nucleic acid extraction group by direct RNA sequencing, and the ratio of viral reads and the sequencing depth of each locus were effectively improved by SISPA amplification. Conclusions This study provided an optimized technical strategy for whole genome sequencing of RNA viruses such as coronavirus. Key words: Human coronavirus; Sequence independent single primer amplification; Whole genome sequencing; High-throughput sequencing

In eukaryotes, chromatin provides a way to compact the genetic material into the confined space of a nucleus. It is also a means to store the same genetic information in different chromatin states. Alteration of these states is enabled by chromatin modifying and remodeling machineries – enzymes that utilize a diverse range of structural changes to chromatin. Despite their apparent importance in gene regulation, it is unclear how they facilitate the transition between chromatin states. Within two distinct projects, we aimed to (1) decipher how chromatin modifying complexes, namely the Polycomb group proteins, are targeted to chromatin and (2) how chromatin remodelers, specifically the ISWI remodeling complexes, change chromatin structure. Polycomb group proteins assemble as chromatin-modifying complexes that maintain the memory of the silent transcriptional state, in part through methylation of lysine 27 on histone H3. Despite their established importance during development, it is largely unclear how these complexes are recruited to specific target genes and how they impair transcription. In flies, Polycomb is recruited by Polycomb response elements that are abundant in various DNA-binding factor motifs. However, the contribution of individual motifs is not yet resolved. In mammals, equivalents of Polycomb response elements are not yet characterized. Here, we aimed to dissect Polycomb-mediated silencing in the mouse genome by identifying DNA determinants of Polycomb recruitment and investigating the role of Polycomb recruitment in transcriptional silencing. More specifically, we developed an assay to test many DNA sequences with various sequence properties for their ability to drive PRC2 recruitment in mouse embryonic stem cells. The assay enabled integration of hundreds of sequences into a defined genomic location in parallel. We found that high density of unmethylated CG motifs within a synthetic backbone sequence is sufficient to recruit PRC2. Furthermore, to link PRC2 recruitment with transcriptional repression, we used CRISPR/Cas9 technology to delete the core PRC2 (Eed) component and monitored the transcriptional response by RNA-seq. Upon depletion of global H3K27me3 levels, we observed no significant changes in gene expression in mouse embryonic stem cells but global deregulation of PRC2 targets during differentiation into neuronal progenitors. These results indicate that recruitment of PRC2 and subsequent H3K27 methylation is important for cell-fate transition, but not required for gene repression in mouse embryonic stem cells. For the second project, we were interested in chromatin remodelers (ISWI) and their role in regulating chromatin structure. Chromatin remodelers are known to use the energy of ATP hydrolysis to evict, slide and reposition nucleosomes, yet we do not fully understand how nucleosome positioning and occupancy affects transcription factor binding. To this aim, we deleted Snf2h, the ATPase subunit of the ISWI chromatin remodeling family, in mouse embryonic stem cells. The Snf2h knockout mouse embryonic stem cells are viable with unchanged expression of pluripotency markers, which is exciting as this is the first viable knockout of an ATPase remodeler subunit. To determine global changes upon deletion of Snf2h, we monitored nucleosome positioning, chromatin accessibility and transcriptional response in Snf2h knockout cells using MNase, ATAC and RNA sequencing, respectively. Extensive data analysis revealed global changes in nucleosome positioning proximal to transcription start sites and transcription factor motifs. Analyzing nucleosome positioning and chromatin accessibility data, we identified transcription factors that require Snf2h to bind their target sites, such as CTCF. It seems that in the absence of Snf2h, nucleosomes cannot be evicted from CTCF motifs, which in turn results in loss of CTCF binding. Taken together, these results indicate that ISWI complexes enable transcription factor binding, at both promoters and distal regulatory regions, by sliding of motif-bound nucleosomes.

The sequencing of the human genome has revealed the existence of a large number of new members of the various cytokine families and has raised the perennial issue of how cytokines function to benefit human evolution. Usually duplication or redundancy of a gene suggests that it functions to assist the survival of the species. In the case of the proinflammatory cytokines, their role in the survival of the host is less clear since this class of cytokines is clearly implicated in host defense mechanisms against infection as well as the pathogenic processes of several diseases. The interleukin-1 (IL-1) and tumor necrosis factor (TNF) families of proinflammatory cytokines have been studied in great detail for their roles in disease; however, studies on their roles in host defense against infection are limited to a few organisms and models. However, it remains likely that these cytokines contribute to host defense mechanisms using the very same molecular and cellular pathways involved in chronic inflammation, tissue destruction, and tissue remodeling. How does one reconcile the existence of two opposing outcomes for the host—one clearly not in the interest of survival and another vital to the host in order to survive even a minor invasion by a microbe? In each case, identical molecular mechanisms are at work. For example, the processes for emigration of neutrophils into tissues to engulf microorganisms via an increase in endothelial adhesion molecules and chemokine production are the same for the host whether in infected tissue or an inflamed joint. On closer examination, the production and activity of proinflammatory cytokines reveal several levels of regulation, which may explain these seemingly discordant roles for the host. For routine infection or injury, cytokines such as IL-1β and TNF-α are transiently expressed and secreted, their activities are modulated by coproduction of naturally occurring antiinflammatory cytokines, and the levels of production fall rapidly. Thus, IL-1β and TNF-α turn on and off rapidly during infection. In contrast, the production of these same cytokines in chronic inflammatory diseases such as rheumatoid arthritis and inflammatory bowel disease fails to cease. Cytokine production in host defense is regulated, whereas in chronic disease it is dysregulated.

Endowing computers with the ability to apply commonsense knowledge with human- level performance is a primary challenge for computer science, comparable in importance to past great challenges in other fields of science such as the sequencing of the human genome. The right approach to this problem is still under debate. Here we shall discuss and attempt to justify one ap- proach, that of knowledge infusion. This approach is based on the view that the fundamental objective that needs to be achieved is robustness in the following sense: a framework is needed in which a computer system can represent pieces of knowledge about the world, each piece having some un- certainty, and the interactions among the pieces having even more uncertainty, such that the system can nevertheless reason from these pieces so that the uncertainties in its conclusions are at least controlled. In knowledge infusion rules are learned from the world in a principled way so that sub- sequent reasoning using these rules will also be principled, and subject only to errors that can be bounded in terms of the inverse of the effort invested in the learning process.

Objective To study the structural features and different types of the genomic molecular organization of human bocavirus in infected infants in Guangzhou. Methods Primers were designed on the basis of the genomic sequence of human bocavirus BJ3064 strain (DQ988933) in the GenBank, and amplify HBOV genome by PCR.The PCR-products were directly sequenced. The genomic nucleotide sequences were analyzed with the programs of CLUSTAL W/X and MEGA5.0.Results The strain GZ2010-1 genome is 5299bp in length. The genomic sequences of GZ2010-1 strain were compared with those sequences of HBOV1, HBOV2, HBOV3, and HBOV4 genotypes in GenBank, revealed that the homology with HBOV1 genotype ranged from 98% to 99%, homology with genotypes HBOV2, HBOV3, HBOV4 were 76%, 81% ,and 75% respectively. The F genes of GZ2010-1 strain was compared with F genes of HBOV1 genotype, showed that the homology was also between 98%-99%. Conclusions The complete nucleotide sequence of human bocavirus GZ2010-1 strain isolated from Guangzhou is 5299 bp in length,GenBank accession NO.is JN 128956. Sequence and phylogenetic analysis of the GZ2010-1 strain reveales that this isolate belongs to HBOV1. Key words: Human bocavirus; Genome; Sequence analysis; Guangzhou

Faced with the immensity of new biological data spurred by the Human Genome Project, bio-ontologies have been generated to provide models of biological concepts to form a semantic framework. Such a semantic framework can be used to underpin a knowledge management (KM) mechanism to facilitate effective knowledge storage, sharing, and communication across computers for biological research. With regards to DNA sequences, the research process usually starts when biologists acquire new sequences. We propose a knowledge organization mechanism using fuzzy measure similarity with Gene Ontology (GO) to assist researchers by alleviating the difficulties of parsing through an immense amount of data and providing insight into how various organisms function. Based on particular evolving ontologies, a value-added knowledge base and semantic-based knowledge agent were established to return precise results tailored for the user.

Motivation: Modern sequencing technology now permits the sequencing of entire genomes, leading to thousands of new gene sequences in need of detailed annotation. It is too time consuming to predict the properties of each protein sequence manually and to organize the results of many prediction tools by hand. The prediction process must be automated so the predictions can be automatically organized, but the predictions must also be transparent. That is, the rationale for each prediction should be easily examinable by anyone that wishes to use the prediction. Results: Proteome Analyst (PA) is a web-based system for predicting the properties of each protein in a proteome. PA has three interesting features. First, it provides a single web-based tool that allows the user to select a wide range of analytic tools and automatically apply them to each protein in a proteome. In essence, PA provides one-stop automatic high-throughput analysis. Second, PA has the ability to explain its predictions to users. PA is based on established machine learning techniques, but makes every prediction transparent to its users. Third, PA allows users to easily create their own transparent custom predictors without programming. Availability: http://www.cs.ualberta.ca/~bioinfo/PA Supplementary Information: http://www.cs.ualberta.ca/~bioinfo/PA/Walkthrough http://www.cs.ualberta.ca/~bioinfo/PA/Experiments Contact: bioinfo@cs.ualberta.ca INTRODUCTION High-throughput sequencing technology has made it possible for even the smallest laboratory to sequence the genome of organelles, viruses and bacteria. Indeed there are now more than 1200 genome sequences deposited in public databases (http://www.ebi.ac.uk/genomes/) and this number is growing rapidly. Given the size and complexity of these data sets, most researchers are compelled to use automated annotation systems to identify or classify individual genes and proteins in their genomic data. A number of excellent systems have been developed over the past few years that permit automated genome-wide or proteome-wide annotation. These include GeneQuiz (Andrade et al., 1999), GeneAtlas (Kitson et al., 2002), EnsEMBL (Hubbard et al., 2002), PEDANT (Frishman et al., 2001), Genotator (Harris, 1997), Magpie (Gaasterland and Sensen, 1996) and GAIA (Overton et al., 1998). These use web-based tools to identify genes, parse data, translate sequences, search against public databases, identify domains or motifs and perform predictive analyses of protein sequences. Many of these packages provide user-customizable search schemas and graphical, hyperlinked output. The level of interpretation or inference offered by these annotation systems varies widely, with some offering only raw data (lists of homologues, calculated properties, etc.) in a consolidated format and others inferring function or ontology through detailed lexical analysis. Our PA system focuses on the task of prediction or classification. Our results show that classification/prediction can be used for many annotations. The most obvious is general-function, but we are also working on sub-cellular localization, specific function and protein-protein interaction are others we are working on. To use classification, we require an ontology or “dictionary” of class terms. Consequently, a number of controlled vocabularies or ontologies of protein function have started appearing. Among the first was the enzyme classification scheme (E.C. number) developed by the IUBMB and first employed in the ENZYME database (Bairoch, 1993). Later, Riley (1993) developed functional classification schemas for E. coli and other microbes. Variations on these functional vocabularies were added as other genomes and genome annotation systems were

Human tuberculosis disease (TB) is caused by bacteria within the Mycobacterium tuberculosis complex, including M. tuberculosis (Mtb). Genetic variation within the pathogen can lead to drug resistance, affect virulence and transmissibility. I have analysed Mtb whole genome sequence data to improve the understanding of global genetic variation, and the resulting insights could ultimately assist the development of TB control measures. Whole genome sequencing platforms are being used to infer drug resistance profiles, and thereby could assist clinical management. I investigated the reproducibility of sequence data from two platforms (Illumina MiSeq, Ion Torrent PGM™) and two rapid analytic pipelines (TBProfiler, Mykrobe predictor). DNA replicates from the reference strain (H37Rv) and 10 drug-resistant strains were sequenced, and inferred drug resistance genotypes were compared to drug susceptibility testing phenotypes. Genome-wide association study (GWAS) can be used to detect mutations associated with Mtb drug resistance. A first GWAS (n=127) attempted to identify mutations associated with minimum inhibitory concentrations for first-line anti-tuberculosis drugs. A second GWAS was applied to a large global set (n>6400) to identify mutations associated with first- and secondline drug resistance. M. aurum is an environmental mycobacteria that has been proposed as a model for the development of anti-TB drugs. I have assembled and annotated its draft genome, and identified copy number variants in known drug resistance targets. Approximately 10% of the Mtb genome consists of two gene families (pe/ppe) that are poorly characterised, and are hypothesised to be important virulence factors. Using a de novo assembly approach, I characterised these genes and their diversity across a global collection of clinical isolates with high depth short-read sequence data (n=518). A follow-up study using a long-read sequence technology (n=18, diverse stain types) confirmed the findings. This work also generated new annotated reference genomes and characterised methylation sites, which may affect transmissibility, pathogenicity and virulence. A future direction of the TB genomics field is to identify genetic check points in host-pathogen interactions using both human and Mtb genotypes. I analysed the genomes of ~720 TB case–Mtb pairs and identified susceptibility markers, which are promising targets for future control measures.