Ten simple rules for the sharing of bacterial genotype-Phenotype data on antimicrobial resistance.

The increasing availability of high-throughput sequencing (frequently termed next-generation sequencing (NGS)) data has created opportunities to gain deeper insights into the mechanisms of a number of diseases and is already impacting many areas of medicine and public health. The area of infectious diseases stands somewhat apart from other human diseases insofar as the relevant genomic data comes from the microbes rather than their human hosts. A particular concern about the threat of antimicrobial resistance (AMR) has driven the collection and reporting of large-scale datasets containing information from microbial genomes together with antimicrobial susceptibility test (AST) results. Unfortunately, the lack of clear standards or guiding principles for the reporting of such data is hampering the field's advancement. We therefore present our recommendations for the publication and sharing of genotype and phenotype data on AMR, in the form of 10 simple rules. The adoption of these recommendations will enhance AMR data interoperability and help enable its large-scale analyses using computational biology tools, including mathematical modelling and machine learning. We hope that these rules can shed light on often overlooked but nonetheless very necessary aspects of AMR data sharing and enhance the field's ability to address the problems of understanding AMR mechanisms, tracking their emergence and spread in populations, and predicting microbial susceptibility to antimicrobials for diagnostic purposes.

Evolution of Salmonella enterica serotype Typhimurium driven by anthropogenic selection and niche adaptation.

Salmonella enterica serotype Typhimurium (S. Typhimurium) is a leading cause of gastroenteritis and bacteraemia worldwide, and a model organism for the study of host-pathogen interactions. Two S. Typhimurium strains (SL1344 and ATCC14028) are widely used to study host-pathogen interactions, yet genotypic variation results in strains with diverse host range, pathogenicity and risk to food safety. The population structure of diverse strains of S. Typhimurium revealed a major phylogroup of predominantly sequence type 19 (ST19) and a minor phylogroup of ST36. The major phylogroup had a population structure with two high order clades (α and ß) and multiple subclades on extended internal branches, that exhibited distinct signatures of host adaptation and anthropogenic selection. Clade α contained a number of subclades composed of strains from well characterized epidemics in domesticated animals, while clade ß contained multiple subclades associated with wild avian species. The contrasting epidemiology of strains in clade α and ß was reflected by the distinct distribution of antimicrobial resistance (AMR) genes, accumulation of hypothetically disrupted coding sequences (HDCS), and signatures of functional diversification. These observations were consistent with elevated anthropogenic selection of clade α lineages from adaptation to circulation in populations of domesticated livestock, and the predisposition of clade ß lineages to undergo adaptation to an invasive lifestyle by a process of convergent evolution with of host adapted Salmonella serotypes. Gene flux was predominantly driven by acquisition and recombination of prophage and associated cargo genes, with only occasional loss of these elements. The acquisition of large chromosomally-encoded genetic islands was limited, but notably, a feature of two recent pandemic clones (DT104 and monophasic S. Typhimurium ST34) of clade α (SGI-1 and SGI-4).

Rapid Genomic Characterization and Global Surveillance of Klebsiella Using Pathogenwatch.

BACKGROUND: Klebsiella species, including the notable pathogen K. pneumoniae, are increasingly associated with antimicrobial resistance (AMR). Genome-based surveillance can inform interventions aimed at controlling AMR. However, its widespread implementation requires tools to streamline bioinformatic analyses and public health reporting. METHODS: We developed the web application Pathogenwatch, which implements analytics tailored to Klebsiella species for integration and visualization of genomic and epidemiological data. We populated Pathogenwatch with 16 537 public Klebsiella genomes to enable contextualization of user genomes. We demonstrated its features with 1636 genomes from 4 low- and middle-income countries (LMICs) participating in the NIHR Global Health Research Unit (GHRU) on AMR. RESULTS: Using Pathogenwatch, we found that GHRU genomes were dominated by a small number of epidemic drug-resistant clones of K. pneumoniae. However, differences in their distribution were observed (eg, ST258/512 dominated in Colombia, ST231 in India, ST307 in Nigeria, ST147 in the Philippines). Phylogenetic analyses including public genomes for contextualization enabled retrospective monitoring of their spread. In particular, we identified hospital outbreaks, detected introductions from abroad, and uncovered clonal expansions associated with resistance and virulence genes. Assessment of loci encoding O-antigens and capsule in K. pneumoniae, which represent possible vaccine candidates, showed that 3 O-types (O1-O3) represented 88.9% of all genomes, whereas capsule types were much more diverse. CONCLUSIONS: Pathogenwatch provides a free, accessible platform for real-time analysis of Klebsiella genomes to aid surveillance at local, national, and global levels. We have improved representation of genomes from GHRU participant countries, further facilitating ongoing surveillance.

Machine learning identifies signatures of host adaptation in the bacterial pathogen Salmonella enterica.

Emerging pathogens are a major threat to public health, however understanding how pathogens adapt to new niches remains a challenge. New methods are urgently required to provide functional insights into pathogens from the massive genomic data sets now being generated from routine pathogen surveillance for epidemiological purposes. Here, we measure the burden of atypical mutations in protein coding genes across independently evolved Salmonella enterica lineages, and use these as input to train a random forest classifier to identify strains associated with extraintestinal disease. Members of the species fall along a continuum, from pathovars which cause gastrointestinal infection and low mortality, associated with a broad host-range, to those that cause invasive infection and high mortality, associated with a narrowed host range. Our random forest classifier learned to perfectly discriminate long-established gastrointestinal and invasive serovars of Salmonella. Additionally, it was able to discriminate recently emerged Salmonella Enteritidis and Typhimurium lineages associated with invasive disease in immunocompromised populations in sub-Saharan Africa, and within-host adaptation to invasive infection. We dissect the architecture of the model to identify the genes that were most informative of phenotype, revealing a common theme of degradation of metabolic pathways in extraintestinal lineages. This approach accurately identifies patterns of gene degradation and diversifying selection specific to invasive serovars that have been captured by more labour-intensive investigations, but can be readily scaled to larger analyses.

Evaluation of parameters affecting performance and reliability of machine learning-based antibiotic susceptibility testing from whole genome sequencing data.

Prediction of antibiotic resistance phenotypes from whole genome sequencing data by machine learning methods has been proposed as a promising platform for the development of sequence-based diagnostics. However, there has been no systematic evaluation of factors that may influence performance of such models, how they might apply to and vary across clinical populations, and what the implications might be in the clinical setting. Here, we performed a meta-analysis of seven large Neisseria gonorrhoeae datasets, as well as Klebsiella pneumoniae and Acinetobacter baumannii datasets, with whole genome sequence data and antibiotic susceptibility phenotypes using set covering machine classification, random forest classification, and random forest regression models to predict resistance phenotypes from genotype. We demonstrate how model performance varies by drug, dataset, resistance metric, and species, reflecting the complexities of generating clinically relevant conclusions from machine learning-derived models. Our findings underscore the importance of incorporating relevant biological and epidemiological knowledge into model design and assessment and suggest that doing so can inform tailored modeling for individual drugs, pathogens, and clinical populations. We further suggest that continued comprehensive sampling and incorporation of up-to-date whole genome sequence data, resistance phenotypes, and treatment outcome data into model training will be crucial to the clinical utility and sustainability of machine learning-based molecular diagnostics.

A profile-based method for identifying functional divergence of orthologous genes in bacterial genomes.

MOTIVATION: Next generation sequencing technologies have provided us with a wealth of information on genetic variation, but predicting the functional significance of this variation is a difficult task. While many comparative genomics studies have focused on gene flux and large scale changes, relatively little attention has been paid to quantifying the effects of single nucleotide polymorphisms and indels on protein function, particularly in bacterial genomics. RESULTS: We present a hidden Markov model based approach we call delta-bitscore (DBS) for identifying orthologous proteins that have diverged at the amino acid sequence level in a way that is likely to impact biological function. We benchmark this approach with several widely used datasets and apply it to a proof-of-concept study of orthologous proteomes in an investigation of host adaptation in Salmonella enterica We highlight the value of the method in identifying functional divergence of genes, and suggest that this tool may be a better approach than the commonly used dN/dS metric for identifying functionally significant genetic changes occurring in recently diverged organisms. AVAILABILITY AND IMPLEMENTATION: A program implementing DBS for pairwise genome comparisons is freely available at: https://github.com/UCanCompBio/deltaBS CONTACT: nicole.wheeler@pg.canterbury.ac.nz or lars.barquist@uni-wuerzburg.deSupplementary information: Supplementary data are available at Bioinformatics online.

Why so narrow: Distribution of anti-sense regulated, type I toxin-antitoxin systems compared with type II and type III systems.

Toxin-antitoxin (TA) systems are gene modules that appear to be horizontally mobile across a wide range of prokaryotes. It has been proposed that type I TA systems, with an antisense RNA-antitoxin, are less mobile than other TAs that rely on direct toxin-antitoxin binding but no direct comparisons have been made. We searched for type I, II and III toxin families using iterative searches with profile hidden Markov models across phyla and replicons. The distribution of type I toxin families were comparatively narrow, but these patterns weakened with recently discovered families. We discuss how the function and phenotypes of TA systems as well as biases in our search methods may account for differences in their distribution.

Therapeutic Vaccination With Dendritic Cells Loaded With Autologous HIV Type 1-Infected Apoptotic Cells.

BACKGROUND: We report the results of a phase I/II, open-label, single-arm clinical trial to evaluate the safety and anti-human immunodeficiency virus type 1 (HIV-1) efficacy of an autologous dendritic cell (DC)-based HIV-1 vaccine loaded with autologous HIV-1-infected apoptotic cells. METHODS: Antiretroviral therapy (ART)-naive individuals were enrolled, and viremia was suppressed by ART prior to delivery of 4 doses of DC-based vaccine. Participants underwent treatment interruption 6 weeks after the third vaccine dose. The plasma HIV-1 RNA level 12 weeks after treatment interruption was compared to the pre-ART (ie, baseline) level. RESULTS: The vaccine was safe and well tolerated but did not prevent viral rebound during treatment interruption. Vaccination resulted in a modest but significant decrease in plasma viremia from the baseline level (from 4.53 log10 copies/mL to 4.27 log10 copies/mL;P= .05). Four of 10 participants had a >0.70 log10 increase in the HIV-1 RNA load in plasma following vaccination, despite continuous ART. Single-molecule sequencing of HIV-1 RNA in plasma before and after vaccination revealed increases in G>A hypermutants in gag and pol after vaccination, which suggests cytolysis of infected cells. CONCLUSIONS: A therapeutic HIV-1 vaccine based on DCs loaded with apoptotic bodies was safe and induced T-cell activation and cytolysis, including HIV-1-infected cells, in a subset of study participants. CLINICAL TRIALS REGISTRATION: NCT00510497.

Robust identification of noncoding RNA from transcriptomes requires phylogenetically-informed sampling.

Noncoding RNAs are integral to a wide range of biological processes, including translation, gene regulation, host-pathogen interactions and environmental sensing. While genomics is now a mature field, our capacity to identify noncoding RNA elements in bacterial and archaeal genomes is hampered by the difficulty of de novo identification. The emergence of new technologies for characterizing transcriptome outputs, notably RNA-seq, are improving noncoding RNA identification and expression quantification. However, a major challenge is to robustly distinguish functional outputs from transcriptional noise. To establish whether annotation of existing transcriptome data has effectively captured all functional outputs, we analysed over 400 publicly available RNA-seq datasets spanning 37 different Archaea and Bacteria. Using comparative tools, we identify close to a thousand highly-expressed candidate noncoding RNAs. However, our analyses reveal that capacity to identify noncoding RNA outputs is strongly dependent on phylogenetic sampling. Surprisingly, and in stark contrast to protein-coding genes, the phylogenetic window for effective use of comparative methods is perversely narrow: aggregating public datasets only produced one phylogenetic cluster where these tools could be used to robustly separate unannotated noncoding RNAs from a null hypothesis of transcriptional noise. Our results show that for the full potential of transcriptomics data to be realized, a change in experimental design is paramount: effective transcriptomics requires phylogeny-aware sampling.

Monitoring residual solvents in pharmaceutical products using a portable pre-concentration GC-PID.

Pharmaceutical manufacturing utilizes solvents at different stages of production. Some of the harmful solvent residuals may be retained in the final product; therefore, they need to be monitored for quality control and to meet the regulation requirement. Here, a novel method capable of rapidly analyzing residual solvents in pharmaceutical products was developed using a compact-portable gas chromatography with a photoionization detector (GC-PID). The method consists of modified Tedlar® bag sampling, online pre-concentration, separation of volatiles by miniaturized GC, and micro-PID detection. The method detection limits of selected residual solvents were in the range of 26.00 - 52.03 pg/mL which is much lower than the pharmaceutical compliance concentration limits. Limits of detection > 520 pg of analyte per grams of sample was also determined for the over-the-counter drugs. The method performance showed rapid speed (5 min), linear calibration (r2 < 0.99), and repeatable retention time (RSD < 0.4 %). Direct analysis of residual solvents in solid samples was conducted without the need for complex sample preparation. The method validation using over-the-counter pharmaceutical products yielded excellent accuracy (recovery > 91.2 %) and precision (RSD < 6.5 %) for the selected residual solvents, including 1,4-dioxane, benzene, chlorobenzene, cyclohexane, xylenes, and toluene. This portable and rapid method could be deployed during the pharmaceutical drug manufacturing processes for quality control.

Developing a Common Global Baseline for Nucleic Acid Synthesis Screening.

Introduction: Nucleic acid synthesis is a powerful tool that has revolutionized the life sciences. However, the misuse of synthetic nucleic acids could pose a serious threat to public health and safety. There is a need for international standards for nucleic acid synthesis screening to help prevent the misuse of this technology. Methods: We outline current barriers to the adoption of screening, which include the cost of developing screening tools and resources, adapting to existing commercial practices, internationalizing screening, and adapting screening to benchtop nucleic acid synthesis devices. To address these challenges, we then introduce the Common Mechanism for DNA Synthesis Screening, which was developed in consultation with a technical consortium of experts in DNA synthesis, synthetic biology, biosecurity, and policy, with the aim of addressing current barriers. The Common Mechanism software uses a variety of methods to identify sequences of concern, identify taxonomic best matches to regulated pathogens, and identify benign genes that can be cleared for synthesis. Finally, we describe outstanding challenges in the development of screening practices. Results: The Common Mechanism is a step toward ensuring the safe and responsible use of synthetic nucleic acids. It provides a baseline capability that overcomes challenges to nucleic acid synthesis screening and provides a solution for broader international adoption of screening practices. Conclusion: The Common Mechanism is a valuable tool for preventing the misuse of synthetic nucleic acids. It is a critical step toward ensuring the safe and responsible use of this powerful technology.

Progress and Prospects for a Nucleic Acid Screening Test Set.

Objective: DNA synthesis companies screen orders to detect controlled sequences with misuse risks. Assessing screening accuracy is challenging owing to the breadth of biological risks and ambiguities in risk definitions. Here, we detail an International Gene Synthesis Consortium working group's rationale and process to develop a prototype DNA synthesis screening test dataset, aiming to establish a baseline of screening system accuracy to compare with various screening approaches. Methodology: Construction of the prototype test dataset involved four tool developers screening nucleic acid sequences from three taxonomic clusters of controlled organisms (Orbivirus, Francisella tularensis, and Coccidioides). Results were mapped onto predefined, comparable categories, checking for consensus or conflicts. Conflicts were grouped based on gene annotation and resolved through discussion. Results: The process highlighted several long-standing challenges in DNA synthesis screening, including the qualitative differences in approaches taken by screening tools. Our findings highlight the lack of clarity in assessing pathogen sequences with respect to regulatory control language, compounded by scientific uncertainty. We illustrate the current degree of consensus and existing challenges using classification statistics and specific examples. Conclusions and Next Steps: This prototype underscores the necessity of expert-regulator coordination in assessing gene-associated risks, offering a template for creating test sets across all taxonomic groups on international control lists. Expanding the working group would enrich dataset comprehensiveness, enabling a transition from species-focused to function-focused regulatory controls. This sets the foundation for quality control, certification, and improved risk assessment in DNA synthesis screening.

Gauge your phage: benchmarking of bacteriophage identification tools in metagenomic sequencing data.

BACKGROUND: The prediction of bacteriophage sequences in metagenomic datasets has become a topic of considerable interest, leading to the development of many novel bioinformatic tools. A comparative analysis of ten state-of-the-art phage identification tools was performed to inform their usage in microbiome research. METHODS: Artificial contigs generated from complete RefSeq genomes representing phages, plasmids, and chromosomes, and a previously sequenced mock community containing four phage species, were used to evaluate the precision, recall, and F1 scores of the tools. We also generated a dataset of randomly shuffled sequences to quantify false-positive calls. In addition, a set of previously simulated viromes was used to assess diversity bias in each tool's output. RESULTS: VIBRANT and VirSorter2 achieved the highest F1 scores (0.93) in the RefSeq artificial contigs dataset, with several other tools also performing well. Kraken2 had the highest F1 score (0.86) in the mock community benchmark by a large margin (0.3 higher than DeepVirFinder in second place), mainly due to its high precision (0.96). Generally, k-mer-based tools performed better than reference similarity tools and gene-based methods. Several tools, most notably PPR-Meta, called a high number of false positives in the randomly shuffled sequences. When analysing the diversity of the genomes that each tool predicted from a virome set, most tools produced a viral genome set that had similar alpha- and beta-diversity patterns to the original population, with Seeker being a notable exception. CONCLUSIONS: This study provides key metrics used to assess performance of phage detection tools, offers a framework for further comparison of additional viral discovery tools, and discusses optimal strategies for using these tools. We highlight that the choice of tool for identification of phages in metagenomic datasets, as well as their parameters, can bias the results and provide pointers for different use case scenarios. We have also made our benchmarking dataset available for download in order to facilitate future comparisons of phage identification tools. Video Abstract.

Safety by design: Biosafety and biosecurity in the age of synthetic genomics.

Technologies to profoundly engineer biology are becoming increasingly affordable, powerful, and accessible to a widening group of actors. While offering tremendous potential to fuel biological research and the bioeconomy, this development also increases the risk of inadvertent or deliberate creation and dissemination of pathogens. Effective regulatory and technological frameworks need to be developed and deployed to manage these emerging biosafety and biosecurity risks. Here, we review digital and biological approaches of a range of technology readiness levels suited to address these challenges. Digital sequence screening technologies already are used to control access to synthetic DNA of concern. We examine the current state of the art of sequence screening, challenges and future directions, and environmental surveillance for the presence of engineered organisms. As biosafety layer on the organism level, we discuss genetic biocontainment systems that can be used to created host organisms with an intrinsic barrier against unchecked environmental proliferation.

Innovations in genomic antimicrobial resistance surveillance.

Whole-genome sequencing of antimicrobial-resistant pathogens is increasingly being used for antimicrobial resistance (AMR) surveillance, particularly in high-income countries. Innovations in genome sequencing and analysis technologies promise to revolutionise AMR surveillance and epidemiology; however, routine adoption of these technologies is challenging, particularly in low-income and middle-income countries. As part of a wider series of workshops and online consultations, a group of experts in AMR pathogen genomics and computational tool development conducted a situational analysis, identifying the following under-used innovations in genomic AMR surveillance: clinical metagenomics, environmental metagenomics, gene or plasmid tracking, and machine learning. The group recommended developing cost-effective use cases for each approach and mapping data outputs to clinical outcomes of interest to justify additional investment in capacity, training, and staff required to implement these technologies. Harmonisation and standardisation of methods, and the creation of equitable data sharing and governance frameworks, will facilitate successful implementation of these innovations.

Evidence review and recommendations for the implementation of genomics for antimicrobial resistance surveillance: reports from an international expert group.

Nearly a century after the beginning of the antibiotic era, which has been associated with unparalleled improvements in human health and reductions in mortality associated with infection, the dwindling pipeline for new antibiotic classes coupled with the inevitable spread of antimicrobial resistance (AMR) poses a major global challenge. Historically, surveillance of bacteria with AMR typically relied on phenotypic analysis of isolates taken from infected individuals, which provides only a low-resolution view of the epidemiology behind an individual infection or wider outbreak. Recent years have seen increasing adoption of powerful new genomic technologies with the potential to revolutionise AMR surveillance by providing a high-resolution picture of the AMR profile of the bacteria causing infections and providing real-time actionable information for treating and preventing infection. However, many barriers remain to be overcome before genomic technologies can be adopted as a standard part of routine AMR surveillance around the world. Accordingly, the Surveillance and Epidemiology of Drug-resistant Infections Consortium convened an expert working group to assess the benefits and challenges of using genomics for AMR surveillance. In this Series, we detail these discussions and provide recommendations from the working group that can help to realise the massive potential benefits for genomics in surveillance of AMR.

Population-based hospitalization burden estimates for respiratory viruses, 2015-2019.

BACKGROUND: Acute respiratory infections (ARIs) result in millions of illnesses and hundreds of thousands of hospitalizations annually in the United States. The responsible viruses include influenza, parainfluenza, human metapneumovirus, coronaviruses, respiratory syncytial virus (RSV), and human rhinoviruses. This study estimated the population-based hospitalization burden of those respiratory viruses (RVs) over 4 years, from July 1, 2015 to June 30, 2019, among adults ≥18 years of age for Allegheny County (Pittsburgh), Pennsylvania. METHODS: We used population-based statewide hospital discharge data, health system electronic medical record (EMR) data for RV tests, census data, and a published method to calculate burden. RESULTS: Among 26,211 eligible RV tests, 67.6% were negative for any virus. The viruses detected were rhinovirus/enterovirus (2552; 30.1%), influenza A (2,299; 27.1%), RSV (1082; 12.7%), human metapneumovirus (832; 9.8%), parainfluenza (601; 7.1%), influenza B (565; 6.7%), non-SARS-CoV-2 coronavirus (420; 4.9% 1.5 years of data available), and adenovirus (136; 1.6%). Most tests were among female (58%) and White (71%) patients with 60% of patients ≥65 years, 24% 50-64 years, and 16% 18-49 years. The annual burden ranged from 137-174/100,000 population for rhinovirus/enterovirus; 99-182/100,000 for influenza A; and 56-81/100,000 for RSV. Among adults <65 years, rhinovirus/enterovirus hospitalization burden was higher than influenza A; whereas the reverse was true for adults ≥65 years. RV hospitalization burden increased with increasing age. CONCLUSIONS: These virus-specific ARI population-based hospital burden estimates showed significant non-influenza burden. These estimates can serve as the basis for several areas of research that are essential for setting funding priorities and guiding public health policy.

Increased power from conditional bacterial genome-wide association identifies macrolide resistance mutations in Neisseria gonorrhoeae.

The emergence of resistance to azithromycin complicates treatment of Neisseria gonorrhoeae, the etiologic agent of gonorrhea. Substantial azithromycin resistance remains unexplained after accounting for known resistance mutations. Bacterial genome-wide association studies (GWAS) can identify novel resistance genes but must control for genetic confounders while maintaining power. Here, we show that compared to single-locus GWAS, conducting GWAS conditioned on known resistance mutations reduces the number of false positives and identifies a G70D mutation in the RplD 50S ribosomal protein L4 as significantly associated with increased azithromycin resistance (p-value = 1.08 × 10-11). We experimentally confirm our GWAS results and demonstrate that RplD G70D and other macrolide binding site mutations are prevalent (present in 5.42% of 4850 isolates) and widespread (identified in 21/65 countries across two decades). Overall, our findings demonstrate the utility of conditional associations for improving the performance of microbial GWAS and advance our understanding of the genetic basis of macrolide resistance.

Improved Prediction of Bacterial Genotype-Phenotype Associations Using Interpretable Pangenome-Spanning Regressions.

Discovery of genetic variants underlying bacterial phenotypes and the prediction of phenotypes such as antibiotic resistance are fundamental tasks in bacterial genomics. Genome-wide association study (GWAS) methods have been applied to study these relations, but the plastic nature of bacterial genomes and the clonal structure of bacterial populations creates challenges. We introduce an alignment-free method which finds sets of loci associated with bacterial phenotypes, quantifies the total effect of genetics on the phenotype, and allows accurate phenotype prediction, all within a single computationally scalable joint modeling framework. Genetic variants covering the entire pangenome are compactly represented by extended DNA sequence words known as unitigs, and model fitting is achieved using elastic net penalization, an extension of standard multiple regression. Using an extensive set of state-of-the-art bacterial population genomic data sets, we demonstrate that our approach performs accurate phenotype prediction, comparable to popular machine learning methods, while retaining both interpretability and computational efficiency. Compared to those of previous approaches, which test each genotype-phenotype association separately for each variant and apply a significance threshold, the variants selected by our joint modeling approach overlap substantially.IMPORTANCE Being able to identify the genetic variants responsible for specific bacterial phenotypes has been the goal of bacterial genetics since its inception and is fundamental to our current level of understanding of bacteria. This identification has been based primarily on painstaking experimentation, but the availability of large data sets of whole genomes with associated phenotype metadata promises to revolutionize this approach, not least for important clinical phenotypes that are not amenable to laboratory analysis. These models of phenotype-genotype association can in the future be used for rapid prediction of clinically important phenotypes such as antibiotic resistance and virulence by rapid-turnaround or point-of-care tests. However, despite much effort being put into adapting genome-wide association study (GWAS) approaches to cope with bacterium-specific problems, such as strong population structure and horizontal gene exchange, current approaches are not yet optimal. We describe a method that advances methodology for both association and generation of portable prediction models.