Rapid determination of anti-tuberculosis drug resistance from whole-genome sequences

Genotyping of Mycobacterium tuberculosis spreading in Hanoi, Vietnam using conventional and whole genome sequencing methods

Hanoi is the capital of Vietnam, one of the 30 countries with a high tuberculosis (TB) burden. Fundamental data on the molecular epidemiology of the disease is required for future TB management. To identify lineages and genotypes of Mycobacterium tuberculosis (Mtb), conventional genotyping data from clinical isolates of the Hanoi area was compared with whole genome sequencing (WGS) analysis from 332 of 470 samples. It was obtained from lineage-specific single nucleotide variants (SNVs), large sequence polymorphisms, spoligotyping, and variable number of tandem repeats (VNTR) analysis using mycobacterial interspersed repetitive unit (MIRU) and Japan anti-tuberculosis association (JATA) locus sets. This information was directly compared with results obtained from WGS. Mini-satellite repeat unit variants were identified using BLAST search against concatenated short read sequences, the RepUnitTyping tool. WGS analysis revealed that the Mtb strains tested are diverse and classified into lineage (L) 1, 2 and 4 (24.7, 57.2 and 18.1% respectively). The majority of the L2 strains were further divided into ancient and modern Beijing genotypes, and most of the L1 group were EAI4_VNM strains. Although conventional PCR-based genotyping results were mostly consistent with information obtained through WGS analysis, in-depth analysis identified aberrant deletions and spacers that may cause discordance. JATA-VNTR sets, including hypervariable loci, separated large Beijing genotypic clusters generated by MIRU15 into smaller groups. The distribution of repeat unit variants observed within 33 VNTR loci showed clear variation depending on the three lineages. WGS-based pairwise-SNV differences within VNTR-defined genotypic clusters were greater in L1 than in L2 and L4 (P = .001). Direct comparisons between results of PCR-based genotyping and in silico analysis of WGS data would bridge a gap between classical and modern technologies during this transition period, and provide further information on Mtb genotypes in specific geographical areas.

Recent Developments in Tuberculous Meningitis Pathogenesis and Diagnostics

The pathogenesis of Tuberculous meningitis (TBM) is poorly understood, but contemporary molecular biology technologies have allowed for recent improvements in our understanding of TBM. For instance, neutrophils appear to play a significant role in the immunopathogenesis of TBM, and either a paucity or an excess of inflammation can be detrimental in TBM. Further, severity of HIV-associated immunosuppression is an important determinant of inflammatory response; patients with the advanced immunosuppression (CD4+ T-cell count of <150 cells/μL) having higher CSF neutrophils, greater CSF cytokine concentrations and higher mortality than those with CD4+ T-cell counts > 150 cells/μL. Host genetics may also influence outcomes with LT4AH genotype predicting inflammatory phenotype, steroid responsiveness and survival in Vietnamese adults with TBM. Whist in Indonesia, CSF tryptophan level was a predictor of survival, suggesting tryptophan metabolism may be important in TBM pathogenesis. These varying responses mean that we must consider whether a "one-size-fits-all" approach to anti-bacillary or immunomodulatory treatment in TBM is truly the best way forward. Of course, to allow for proper treatment, early and rapid diagnosis of TBM must occur. Diagnosis has always been a challenge but the field of TB diagnosis is evolving, with sensitivities of at least 70% now possible in less than two hours with GeneXpert MTB/Rif Ultra. In addition, advanced molecular techniques such as CRISPR-MTB and metagenomic next generation sequencing may hold promise for TBM diagnosis. Host-based biomarkers and signatures are being further evaluated in childhood and adult TBM as adjunctive biomarkers as even with improved molecular assays, cases are still missed. A better grasp of host and pathogen behaviour may lead to improved diagnostics, targeted immunotherapy, and possibly biomarker-based, patient-specific treatment regimens.

Improved Resistance Prediction in Mycobacterium tuberculosis by Better Handling of Insertions and Deletions, Premature Stop Codons, and Filtering of Non-informative Sites

Resistance in Mycobacterium tuberculosis is a major obstacle for effective treatment of tuberculosis. Multiple studies have shown promising results for predicting drug resistance in M. tuberculosis based on whole genome sequencing (WGS) data, however, these tools are often limited to this single species. We have previously developed a common platform for resistance prediction in multiple species. This platform detects acquired resistance genes (ResFinder) and species-specific chromosomal mutations (PointFinder) associated with resistance, all based on WGS data. In this study, we present a new version of PointFinder together with an updated M. tuberculosis database. PointFinder now includes predictions based on insertions and deletions, and it explicitly reports frameshift mutations and premature stop codons. We found that premature stop codons in four resistance-associated genes (katG, ethA, pncA, and gidB) were over-represented in resistant strains, and we saw an increased prediction performance when including premature stop codons in these genes as resistance markers. Different M. tuberculosis resistance prediction tools vary in performance mostly due to the mutation library used. We found that a well-established mutation library included non-predictive linage markers, and through forward feature selection we eliminated those from the mutation library. Compared to other similar web-based tools, PointFinder performs equally good. The advantages of PointFinder is that together with ResFinder it serves as a common web-based and downloadable platform for resistance detection in multiple species. It is easy to use for clinicians and already widely used in the research community.

Whole genome sequencing, analyses of drug resistance-conferring mutations, and correlation with transmission of Mycobacterium tuberculosis carrying katG-S315T in Hanoi, Vietnam

Drug-resistant tuberculosis (TB) is a serious global problem, and pathogen factors involved in the transmission of isoniazid (INH)-resistant TB have not been fully investigated. We performed whole genome sequencing of 332 clinical Mycobacterium tuberculosis (Mtb) isolates collected from patients newly diagnosed with smear-positive pulmonary TB in Hanoi, Vietnam. Using a bacterial genome-wide approach based on linear mixed models, we investigated the associations between 31-bp k-mers and clustered strains harboring katG-S315T, a major INH-resistance mutation in the present cohort and in the second panel previously published in South Africa. Five statistically significant genes, namely, PPE18/19, gid, emrB, Rv1588c, and pncA, were shared by the two panels. We further identified variants of the genes responsible for these k-mers, which are relevant to the spread of INH-resistant strains. Phylogenetic convergence test showed that variants relevant to PPE46/47-like chimeric genes were significantly associated with the same phenotype in Hanoi. The associations were further confirmed after adjustment for the confounders. These findings suggest that genomic variations of the pathogen facilitate the expansion of INH-resistance TB, at least in part, and our study provides a new insight into the mechanisms by which drug-resistant Mtb maintains fitness and spreads in Asia and Africa.

Whole genome sequencing of drug resistant Mycobacterium tuberculosis isolates from a high burden tuberculosis region of North West Pakistan

Tuberculosis (TB), caused by Mycobacterium tuberculosis bacteria, is a leading infectious cause of mortality worldwide, including in Pakistan. Drug resistant M. tuberculosis is an emerging threat for TB control, making it important to detect the underlying genetic mutations, and thereby inform treatment decision making and prevent transmission. Whole genome sequencing has emerged as the new diagnostic to reliably predict drug resistance within a clinically relevant time frame, and its deployment will have the greatest impact on TB control in highly endemic regions. To evaluate the mutations leading to drug resistance and to assess for evidence of the transmission of resistant strains, 81 M. tuberculosis samples from Khyber Pakhtunkhwa province (North West Pakistan) were subjected to whole genome sequencing and standard drug susceptibility testing for eleven anti-TB drugs. We found the majority of M. tuberculosis isolates were the CAS/Delhi strain-type (lineage 3; n = 57; 70.4%) and multi-drug resistant (MDR; n = 62; 76.5%). The most frequent resistance mutations were observed in the katG and rpoB genes, conferring resistance to isoniazid and rifampicin respectively. Mutations were also observed in genes conferring resistance to other first and second-line drugs, including in pncA (pyrazinamide), embB (ethambutol), gyrA (fluoroquinolones), rrs (aminoglycosides), rpsL, rrs and giB (streptomycin) loci. Whilst the majority of mutations have been reported in global datasets, we describe unreported putative resistance markers in katG, ethA (ethionamide), gyrA and gyrB (fluoroquinolones), and pncA. Analysis of the mutations revealed that acquisition of rifampicin resistance often preceded isoniazid in our isolates. We also observed a high proportion (17.6%) of pre-MDR isolates with fluoroquinolone resistance markers, potentially due to unregulated anti-TB drug use. Our isolates were compared to previously sequenced strains from Pakistan in a combined phylogenetic tree analysis. The presence of lineage 2 was only observed in our isolates. Using a cut-off of less than ten genome-wide mutation differences between isolates, a transmission analysis revealed 18 M. tuberculosis isolates clustering within eight networks, thereby providing evidence of drug-resistant TB transmission in the Khyber Pakhtunkhwa province. Overall, we have demonstrated that drug-resistant TB isolates are circulating and transmitted in North West Pakistan. Further, we have shown the usefulness of whole genome sequencing as a diagnostic tool for characterizing M. tuberculosis isolates, which will assist future epidemiological studies and disease control activities in Pakistan.

Machine Learning Predicts Accurately Mycobacterium tuberculosis Drug Resistance From Whole Genome Sequencing Data

Background: Tuberculosis disease, caused by Mycobacterium tuberculosis, is a major public health problem. The emergence of M. tuberculosis strains resistant to existing treatments threatens to derail control efforts. Resistance is mainly conferred by mutations in genes coding for drug targets or converting enzymes, but our knowledge of these mutations is incomplete. Whole genome sequencing (WGS) is an increasingly common approach to rapidly characterize isolates and identify mutations predicting antimicrobial resistance and thereby providing a diagnostic tool to assist clinical decision making. Methods: We applied machine learning approaches to 16,688 M. tuberculosis isolates that have undergone WGS and laboratory drug-susceptibility testing (DST) across 14 antituberculosis drugs, with 22.5% of samples being multidrug resistant and 2.1% being extensively drug resistant. We used non-parametric classification-tree and gradient-boosted-tree models to predict drug resistance and uncover any associated novel putative mutations. We fitted separate models for each drug, with and without "co-occurrent resistance" markers known to be causing resistance to drugs other than the one of interest. Predictive performance was measured using sensitivity, specificity, and the area under the receiver operating characteristic curve, assuming DST results as the gold standard. Results: The predictive performance was highest for resistance to first-line drugs, amikacin, kanamycin, ciprofloxacin, moxifloxacin, and multidrug-resistant tuberculosis (area under the receiver operating characteristic curve above 96%), and lowest for third-line drugs such as D-cycloserine and Para-aminosalisylic acid (area under the curve below 85%). The inclusion of co-occurrent resistance markers led to improved performance for some drugs and superior results when compared to similar models in other large-scale studies, which had smaller sample sizes. Overall, the gradient-boosted-tree models performed better than the classification-tree models. The mutation-rank analysis detected no new single nucleotide polymorphisms linked to drug resistance. Discordance between DST and genotypically inferred resistance may be explained by DST errors, novel rare mutations, hetero-resistance, and nongenomic drivers such as efflux-pump upregulation. Conclusion: Our work demonstrates the utility of machine learning as a flexible approach to drug resistance prediction that is able to accommodate a much larger number of predictors and to summarize their predictive ability, thus assisting clinical decision making and single nucleotide polymorphism detection in an era of increasing WGS data generation.

Whole-genome sequencing-based epidemiological analysis of anti-tuberculosis drug resistance genes in Japan in 2007: Application of the Genome Research for Asian Tuberculosis (GReAT) database

We investigated the lineages of Mycobacterium tuberculosis (Mtb) isolates from the RYOKEN study in Japan in 2007 and the usefulness of genotypic drug susceptibility testing (DST) using the Genome Research for Asian Tuberculosis (GReAT) database. In total, 667 isolates were classified into lineage 1 (4.6%), lineage 2 (0.8%), lineage 2/Beijing (72.1%), lineage 3 (0.5%), and lineage 4 (22.0%). The nationality, gender, and age groups associated with the isolates assigned to lineage 1 were significantly different from those associated with other lineages. In particular, isolates of lineage 1.2.1 (EAI2) formed sub-clusters and included a 2,316-bp deletion in the genome. The proportion of the isolates resistant to at least one anti-tuberculosis (TB) drug was 10.8%, as determined by either the genotypic or phenotypic method of DST. However, the sensitivities to isoniazid, streptomycin, and ethambutol determined by the genotypic method were low. Thus, unidentified mutations in the genome responsible for drug resistance were explored, revealing previously unreported mutations in the katG, gid, and embB genes. This is the first nationwide report of whole-genome analysis of TB in Japan.

COMBAT-TB-NeoDB: fostering tuberculosis research through integrative analysis using graph database technologies

MOTIVATION

Recent advancements in genomic technologies have enabled high throughput cost-effective generation of 'omics' data from M.tuberculosis (M.tb) isolates, which then gets shared via a number of heterogeneous publicly available biological databases. Albeit useful, fragmented curation negatively impacts the researcher's ability to leverage the data via federated queries.

RESULTS

We present Combat-TB-NeoDB, an integrated M.tb 'omics' knowledge-base. Combat-TB-NeoDB is based on Neo4j and was created by binding the labeled property graph model to a suitable ontology namely Chado. Combat-TB-NeoDB enables researchers to execute complex federated queries by linking prominent biological databases, and supplementary M.tb variants data from published literature.

AVAILABILITY AND IMPLEMENTATION

The Combat-TB-NeoDB (https://neodb.sanbi.ac.za) repository and all tools mentioned in this manuscript are freely available at https://github.com/COMBAT-TB.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Direct transmission of within-host Mycobacterium tuberculosis diversity to secondary cases can lead to variable between-host heterogeneity without de novo mutation: A genomic investigation

BACKGROUND

Whole genome sequencing (WGS) has enabled the development of new approaches to track Mycobacterium tuberculosis (Mtb) transmission between tuberculosis (TB) cases but its utility may be challenged by the discovery that Mtb diversifies within hosts. Nevertheless, there is limited data on the presence and degree of within-host evolution.

METHODS

We profiled a well-documented Mtb transmission cluster with three pulmonary TB cases to investigate within-host evolution and describe its impact on recent transmission estimates. We used deep sequencing to track minority allele frequencies (<50·0% abundance) during transmission and standard treatment.

FINDINGS

Pre-treatment (n = 3) and serial samples collected over 2 months of antibiotic treatment (n = 16) from all three cases were analysed. Consistent with the epidemiological data, zero fixed SNP separated all genomes. However, we identified six subclones between the three cases with an allele frequency ranging from 35·0% to 100·0% across sampling intervals. Five subclones were identified within the index case pre-treatment and shared with one secondary case, while only the dominant clone was observed in the other secondary case. By tracking the frequency of these heterogeneous alleles over the two-month therapy, we observed distinct signatures of drift and negative selection, but limited evidence for de novo mutations, even under drug pressure.

INTERPRETATION

We document within-host Mtb diversity in an index case, which led to transmission of minority alleles to a secondary case. Incorporating data on heterogeneous alleles may refine our understanding of Mtb transmission dynamics. However, more evidence is needed on the role of transmission bottleneck on observed heterogeneity between cases.

Xpert MTB/RIF performance to diagnose tuberculosis and rifampicin resistance in a reference centre in southern Brazil

Effective treatment of tuberculosis (TB) remains a serious public health problem in many countries, including Brazil, especially when considering drug-resistant disease. Xpert MTB/RIF has been implemented in many countries to reduce the time to TB diagnosis and to rapidly detect rifampicin resistance. The study aimed to describe and evaluate Xpert MTB/RIF performance in diagnosing pulmonary TB and rifampicin resistance in a tertiary healthcare facility in Brazil. A cross-sectional study was performed, which included all isolates of confirmed pulmonary TB patients from 2015 to 2018. Both Xpert MTB/RIF and GenoType MTBDRplus assays were performed to detect rifampicin and isoniazid resistance. In addition, isolates with detected resistance to rifampicin and/or isoniazid were analysed by phenotypic testing using MGIT-960 SIRE kit and whole-genome sequencing (WGS) using Illumina MiSeq Sequencing System. 2148 respiratory specimens tested with Xpert MTB/RIF were included: n=1556 sputum, n=348 bronchoalveolar lavage and n=244 gastric washing. The overall Xpert MTB/RIF sensitivity in sputum was 94% and the overall specificity was 98%. The negative predictive value in sputum of all the patients was 99% with a positive predictive value of 89%. The concordance between Xpert MTB/RIF and phenotypic susceptibility test was 94.1%, while its concordance with WGS was 78.9%. Xpert MTB/RIF is a rapid and accurate diagnostic strategy for pulmonary TB, which can contribute to improvement in TB control. However, detection of rifampicin resistance might be associated with false-positive results.

Deciphering drug resistance in Mycobacterium tuberculosis using whole-genome sequencing: progress, promise, and challenges

Tuberculosis (TB) is a global infectious threat that is intensified by an increasing incidence of highly drug-resistant disease. Whole-genome sequencing (WGS) studies of Mycobacterium tuberculosis, the causative agent of TB, have greatly increased our understanding of this pathogen. Since the first M. tuberculosis genome was published in 1998, WGS has provided a more complete account of the genomic features that cause resistance in populations of M. tuberculosis, has helped to fill gaps in our knowledge of how both classical and new antitubercular drugs work, and has identified specific mutations that allow M. tuberculosis to escape the effects of these drugs. WGS studies have also revealed how resistance evolves both within an individual patient and within patient populations, including the important roles of de novo acquisition of resistance and clonal spread. These findings have informed decisions about which drug-resistance mutations should be included on extended diagnostic panels. From its origins as a basic science technique, WGS of M. tuberculosis is becoming part of the modern clinical microbiology laboratory, promising rapid and improved detection of drug resistance, and detailed and real-time epidemiology of TB outbreaks. We review the successes and highlight the challenges that remain in applying WGS to improve the control of drug-resistant TB through monitoring its evolution and spread, and to inform more rapid and effective diagnostic and therapeutic strategies.

Personalized Approach as a Basis for the Future Diagnosis of Tuberculosis (Literature Review)

The global spread of tuberculosis remains one of actual problems of public health despite of introduction of public health safety programs. Early, rapid and accurate identification of M. tuberculosis and determination of drug susceptibility are essential for treatment and management of this disease. Delay in delivering results prolongs potentially inappropriate antituberculosis therapy, contributing to emergence of drug resistance, reducing treatment options and increasing treatment duration and associated costs, resulting in increased mortality and morbidity. Faster, more comprehensive diagnostics will enable earlier use of the most appropriate drug regimen, thus improving patient outcomes and reducing overall healthcare costs. The treatment of infection based on the using of massive antimicrobial therapy with analysis of bacterial strains resistance to first line drugs (FLD) isoniazid (INH), rifampin (RIF), pyrazinamide (PZA), ethambutol (EMB) and streptomycin (SM). However, the public health practitioners pay no attention to functional activity of human immune system genes. The interaction of bacterial genomes and immune system genes plays the major role in infection progress. There is growing evidence that, together with human and environmental factors, Mycobacterium tuberculosis complex strain diversity contributes to the variable outcome of infection and disease in human TB. We suppose that the future of diagnosis and treatment of tuberculosis lies in the field of personal medicine with comprehensive analysis of host and pathogen genes.

Identification and Characterization of Genetic Determinants of Isoniazid and Rifampicin Resistance in Mycobacterium tuberculosis in Southern India

Drug-resistant tuberculosis (TB), one of the leading causes of death worldwide, arises mainly from spontaneous mutations in the genome of Mycobacterium tuberculosis. There is an urgent need to understand the mechanisms by which the mutations confer resistance in order to identify new drug targets and to design new drugs. Previous studies have reported numerous mutations that confer resistance to anti-TB drugs, but there has been little systematic analysis to understand their genetic background and the potential impacts on the drug target stability and/or interactions. Here, we report the analysis of whole-genome sequence data for 98 clinical M. tuberculosis isolates from a city in southern India. The collection was screened for phenotypic resistance and sequenced to mine the genetic mutations conferring resistance to isoniazid and rifampicin. The most frequent mutation among isoniazid and rifampicin isolates was S315T in katG and S450L in rpoB respectively. The impacts of mutations on protein stability, protein-protein interactions and protein-ligand interactions were analysed using both statistical and machine-learning approaches. Drug-resistant mutations were predicted not only to target active sites in an orthosteric manner, but also to act through allosteric mechanisms arising from distant sites, sometimes at the protein-protein interface.

Whole genome sequencing provides additional insights into recurrent tuberculosis classified as endogenous reactivation by IS6110 DNA fingerprinting

Recurrent tuberculosis (TB) after successful TB treatment occurs due to endogenous reactivation (relapse) or exogenous reinfection. We revisited the conclusions of relapse in a high TB incidence setting that were drawn on the basis of IS6110 restriction fragment length polymorphism (RFLP) analysis in a large retrospective cohort study in suburban Cape Town, South Africa. Using whole genome sequencing (WGS), we undertook pair-wise genome comparison of Mycobacterium tuberculosis strains cultured from diagnostic sputum samples collected at the index and recurrent TB episode for 25 recurrent TB cases who had been classified as relapse based on identical DNA fingerprint patterns in the earlier study. We found that paired strain genome sequences were identical or showed minimal variant differences in 22 of 25 recurrent TB cases, consistent with relapse. One showed 20 variant differences, suggestive of exogenous reinfection. Two of the 25 had mixed infections, each with the index episode strain detected as the dominant strain at recurrence in one of these patients, the minority strain harboured drug-resistance conferring mutations (rpoB, katG). In conclusion, our study highlights the additional value of WGS for investigating recurrent TB in settings with high infection pressure and closely related circulating strains, where the extent of re- and mixed infection may be underestimated.

Application of machine learning techniques to tuberculosis drug resistance analysis

MOTIVATION

Timely identification of Mycobacterium tuberculosis (MTB) resistance to existing drugs is vital to decrease mortality and prevent the amplification of existing antibiotic resistance. Machine learning methods have been widely applied for timely predicting resistance of MTB given a specific drug and identifying resistance markers. However, they have been not validated on a large cohort of MTB samples from multi-centers across the world in terms of resistance prediction and resistance marker identification. Several machine learning classifiers and linear dimension reduction techniques were developed and compared for a cohort of 13 402 isolates collected from 16 countries across 6 continents and tested 11 drugs.

RESULTS

Compared to conventional molecular diagnostic test, area under curve of the best machine learning classifier increased for all drugs especially by 23.11%, 15.22% and 10.14% for pyrazinamide, ciprofloxacin and ofloxacin, respectively (P < 0.01). Logistic regression and gradient tree boosting found to perform better than other techniques. Moreover, logistic regression/gradient tree boosting with a sparse principal component analysis/non-negative matrix factorization step compared with the classifier alone enhanced the best performance in terms of F1-score by 12.54%, 4.61%, 7.45% and 9.58% for amikacin, moxifloxacin, ofloxacin and capreomycin, respectively, as well increasing area under curve for amikacin and capreomycin. Results provided a comprehensive comparison of various techniques and confirmed the application of machine learning for better prediction of the large diverse tuberculosis data. Furthermore, mutation ranking showed the possibility of finding new resistance/susceptible markers.

AVAILABILITY AND IMPLEMENTATION

The source code can be found at http://www.robots.ox.ac.uk/ davidc/code.php.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Mycobacterium tuberculosis whole genome sequencing provides insights into the Manila strain and drug-resistance mutations in the Philippines

The Philippines has a high incidence of tuberculosis disease (TB), with an increasing prevalence of multidrug-resistant Mycobacterium tuberculosis (MDR-TB) strains making its control difficult. Although the M. tuberculosis "Manila" ancient lineage 1 strain-type is thought to be prevalent in the country, with evidence of export to others, little is known about the genetic diversity of circulating strains. By whole genome sequencing (WGS) 178 isolates from the Philippines National Drug Resistance Survey, we found the majority (143/178; 80.3%) belonged to the lineage 1 Manila clade, with the minority belonging to lineages 4 (European-American; n = 33) and 2 (East Asian; n = 2). A high proportion were found to be multidrug-resistant (34/178; 19.1%), established through highly concordant laboratory drug susceptibility testing and in silico prediction methods. Some MDR-TB isolates had near identical genomic variation, providing potential evidence of transmission. By placing the Philippine isolates within a phylogeny of global M. tuberculosis (n > 17,000), we established that they are genetically similar to those observed outside the country, including a clade of Manila-like strain-types in Thailand. An analysis of the phylogeny revealed a set of ~200 SNPs that are specific for the Manila strain-type, and a subset can be used within a molecular barcode. Sixty-eight mutations known to be associated with 10 anti-TB drug resistance were identified in the Philippine strains, and all have been observed in other populations. Whilst nine putative streptomycin resistance conferring markers in gid (8) and rrs (1) genes appear to be novel and with functional consequences. Overall, this study provides an important baseline characterisation of M. tuberculosis genetic diversity for the Philippines, and will fill a gap in global datasets and aid the development of a nation-wide database for epidemiological studies and clinical decision making. Further, by establishing a molecular barcode for detecting Manila strains it will assist with the design of diagnostic tools for disease control activities.

Integrating informatics tools and portable sequencing technology for rapid detection of resistance to anti-tuberculous drugs

Background

Mycobacterium tuberculosis resistance to anti-tuberculosis drugs is a major threat to global public health. Whole genome sequencing (WGS) is rapidly gaining traction as a diagnostic tool for clinical tuberculosis settings. To support this informatically, previous work led to the development of the widely used TBProfiler webtool, which predicts resistance to 14 drugs from WGS data. However, for accurate and rapid high throughput of samples in clinical or epidemiological settings, there is a need for a stand-alone tool and the ability to analyse data across multiple WGS platforms, including Oxford Nanopore MinION.

Results

We present a new command line version of the TBProfiler webserver, which includes hetero-resistance calling and will facilitate the batch processing of samples. The TBProfiler database has been expanded to incorporate 178 new markers across 16 anti-tuberculosis drugs. The predictive performance of the mutation library has been assessed using > 17,000 clinical isolates with WGS and laboratory-based drug susceptibility testing (DST) data. An integrated MinION analysis pipeline was assessed by performing WGS on 34 replicates across 3 multi-drug resistant isolates with known resistance mutations. TBProfiler accuracy varied by individual drug. Assuming DST as the gold standard, sensitivities for detecting multi-drug-resistant TB (MDR-TB) and extensively drug-resistant TB (XDR-TB) were 94% (95%CI 93–95%) and 83% (95%CI 79–87%) with specificities of 98% (95%CI 98–99%) and 96% (95%CI 95–97%) respectively. Using MinION data, only one resistance mutation was missed by TBProfiler, involving an insertion in the tlyA gene coding for capreomycin resistance. When compared to alternative platforms (e.g. Mykrobe predictor TB, the CRyPTIC library), TBProfiler demonstrated superior predictive performance across first- and second-line drugs.

Conclusions

The new version of TBProfiler can rapidly and accurately predict anti-TB drug resistance profiles across large numbers of samples with WGS data. The computing architecture allows for the ability to modify the core bioinformatic pipelines and outputs, including the analysis of WGS data sourced from portable technologies. TBProfiler has the potential to be integrated into the point of care and WGS diagnostic environments, including in resource-poor settings.

Electronic supplementary material

The online version of this article (10.1186/s13073-019-0650-x) contains supplementary material, which is available to authorized users.

Machine learning for classifying tuberculosis drug-resistance from DNA sequencing data

Motivation

Correct and rapid determination of Mycobacterium tuberculosis (MTB) resistance against available tuberculosis (TB) drugs is essential for the control and management of TB. Conventional molecular diagnostic test assumes that the presence of any well-studied single nucleotide polymorphisms is sufficient to cause resistance, which yields low sensitivity for resistance classification.

Summary

Given the availability of DNA sequencing data from MTB, we developed machine learning models for a cohort of 1839 UK bacterial isolates to classify MTB resistance against eight anti-TB drugs (isoniazid, rifampicin, ethambutol, pyrazinamide, ciprofloxacin, moxifloxacin, ofloxacin, streptomycin) and to classify multi-drug resistance.

Results

Compared to previous rules-based approach, the sensitivities from the best-performing models increased by 2-4% for isoniazid, rifampicin and ethambutol to 97% (P < 0.01), respectively; for ciprofloxacin and multi-drug resistant TB, they increased to 96%. For moxifloxacin and ofloxacin, sensitivities increased by 12 and 15% from 83 and 81% based on existing known resistance alleles to 95% and 96% (P < 0.01), respectively. Particularly, our models improved sensitivities compared to the previous rules-based approach by 15 and 24% to 84 and 87% for pyrazinamide and streptomycin (P < 0.01), respectively. The best-performing models increase the area-under-the-ROC curve by 10% for pyrazinamide and streptomycin (P < 0.01), and 4–8% for other drugs (P < 0.01).

Availability and implementation

The details of source code are provided at http://www.robots.ox.ac.uk/~davidc/code.php.

Supplementary information

Supplementary data are available at Bioinformatics online.

Whole genomic sequencing as a tool for diagnosis of drug and multidrug-resistance tuberculosis in an endemic region in Mexico

Background

Whole genome sequencing (WGS) has been proposed as a tool for diagnosing drug resistance in tuberculosis. However, reports of its effectiveness in endemic countries with important numbers of drug resistance are scarce. The goal of this study was to evaluate the effectiveness of this procedure in isolates from a tuberculosis endemic region in Mexico.

Methods

WGS analysis was performed in 81 tuberculosis positive clinical isolates with a known phenotypic profile of resistance against first-line drugs (isoniazid, rifampin, ethambutol, pyrazinamide and streptomycin). Mutations related to drug resistance were identified for each isolate; drug resistant genotypes were predicted and compared with the phenotypic profile. Genotypes and transmission clusters based on genetic distances were also characterized.

Findings

Prediction by WGS analysis of resistance against isoniazid, rifampicin, ethambutol, pyrazinamide and streptomycin showed sensitivity values of 84%, 96%, 71%, 75% and 29%, while specificity values were 100%, 94%, 90%, 90% and 98%, respectively. Prediction of multidrug resistance showed a sensitivity of 89% and specificity of 97%. Moreover, WGS analysis revealed polymorphisms related to second-line drug resistance, enabling classification of eight and two clinical isolates as pre- and extreme drug-resistant cases, respectively. Lastly, four lineages were identified in the population (L1, L2, L3 and L4). The most frequent of these was L4, which included 90% (77) of the isolates. Six transmission clusters were identified; the most frequent was TC6, which included 13 isolates with a L4.1.1 and a predominantly multidrug-resistant condition.

Conclusions

The results illustrate the utility of WGS for establishing the potential for prediction of resistance against first and second line drugs in isolates of tuberculosis from the region. They also demonstrate the feasibility of this procedure for use as a tool to support the epidemiological surveillance of drug- and multidrug-resistant tuberculosis.

Empirical ways to identify novel Bedaquiline resistance mutations in AtpE

Clinical resistance against Bedaquiline, the first new anti-tuberculosis compound with a novel mechanism of action in over 40 years, has already been detected in Mycobacterium tuberculosis. As a new drug, however, there is currently insufficient clinical data to facilitate reliable and timely identification of genomic determinants of resistance. Here we investigate the structural basis for M. tuberculosis associated bedaquiline resistance in the drug target, AtpE. Together with the 9 previously identified resistance-associated variants in AtpE, 54 non-resistance-associated mutations were identified through comparisons of bedaquiline susceptibility across 23 different mycobacterial species. Computational analysis of the structural and functional consequences of these variants revealed that resistance associated variants were mainly localized at the drug binding site, disrupting key interactions with bedaquiline leading to reduced binding affinity. This was used to train a supervised predictive algorithm, which accurately identified likely resistance mutations (93.3% accuracy). Application of this model to circulating variants present in the Asia-Pacific region suggests that current circulating variants are likely to be susceptible to bedaquiline. We have made this model freely available through a user-friendly web interface called SUSPECT-BDQ, StrUctural Susceptibility PrEdiCTion for bedaquiline (http://biosig.unimelb.edu.au/suspect_bdq/). This tool could be useful for the rapid characterization of novel clinical variants, to help guide the effective use of bedaquiline, and to minimize the spread of clinical resistance.

Whole genome sequencing Mycobacterium tuberculosis directly from sputum identifies more genetic diversity than sequencing from culture

Background

Repeated culture reduces within-sample Mycobacterium tuberculosis genetic diversity due to selection of clones suited to growth in culture and/or random loss of lineages, but it is not known to what extent omitting the culture step altogether alters genetic diversity. We compared M. tuberculosis whole genome sequences generated from 33 paired clinical samples using two methods. In one method DNA was extracted directly from sputum then enriched with custom-designed SureSelect (Agilent) oligonucleotide baits and in the other it was extracted from mycobacterial growth indicator tube (MGIT) culture.

Results

DNA directly sequenced from sputum showed significantly more within-sample diversity than that from MGIT culture (median 5.0 vs 4.5 heterozygous alleles per sample, p = 0.04). Resistance associated variants present as HAs occurred in four patients, and in two cases may provide a genotypic explanation for phenotypic resistance.

Conclusions

Culture-free M. tuberculosis whole genome sequencing detects more within-sample diversity than a leading culture-based method and may allow detection of mycobacteria that are not actively replicating.

Electronic supplementary material

The online version of this article (10.1186/s12864-019-5782-2) contains supplementary material, which is available to authorized users.

Whole-Genome Sequencing in Relation to Resistance of Mycobacterium Tuberculosis

Tuberculosis, a disease caused by Mycobacterium tuberculosis, represents one of the deadliest infections worldwide. The incidence of resistant forms is increasing year by year; therefore, it is necessary to involve new methods for rapid diagnostics and treatment. One of the possible solutions is the use of whole-genome sequencing (WGS).

The WGS provides an identification of complete genome of the microorganism, including all genes responsible for resistance, in comparison with other genotypic methods (eg. Xpert MTB / RIF or Hain line-probes) that are capable to detect only basic genes. WGS data are available in 1-9 days and several online software tools (TBProfiler, CASTB, Mykrobe PredictorTB) are used for their interpretation and analysis, compared to 3-8 weeks in the case of classic phenotypic evaluation.

Furthermore, WGS predicts resistance to the first-line antituberculotics with a sensitivity of 85-100% and a specificity of 85-100%.

This review elucidates the importance and summarizes the current knowledge about the possible use of WGS in diagnosis and treatment of resistant forms of tuberculosis elucidates.

WGS of M. tuberculosis brings new possibilities for rapid and accurate diagnostics of resistant forms of tuberculosis. Introducing WGS into routine practice can help to reduce the spread of resistant forms of tuberculosis as well as to increase the success rate of the treatment, especially through an appropriate combination of antituberculotics ATs. Introduction of WGS into routine diagnostics can, in spite of the financial difficulty, significantly improve patient care.

A large scale evaluation of TBProfiler and Mykrobe for antibiotic resistance prediction in Mycobacterium tuberculosis

Recent years saw a growing interest in predicting antibiotic resistance from whole-genome sequencing data, with promising results obtained for Staphylococcus aureus and Mycobacterium tuberculosis. In this work, we gathered 6,574 sequencing read datasets of M. tuberculosis public genomes with associated antibiotic resistance profiles for both first and second-line antibiotics. We performed a systematic evaluation of TBProfiler and Mykrobe, two widely recognized softwares allowing to predict resistance in M. tuberculosis. The size of the dataset allowed us to obtain confident estimations of their overall predictive performance, to assess precisely the individual predictive power of the markers they rely on, and to study in addition how these softwares behave across the major M. tuberculosis lineages. While this study confirmed the overall good performance of these tools, it revealed that an important fraction of the catalog of mutations they embed is of limited predictive power. It also revealed that these tools offer different sensitivity/specificity trade-offs, which is mainly due to the different sets of mutation they embed but also to their underlying genotyping pipelines. More importantly, it showed that their level of predictive performance varies greatly across lineages for some antibiotics, therefore suggesting that the predictions made by these softwares should be deemed more or less confident depending on the lineage inferred and the predictive performance of the marker(s) actually detected. Finally, we evaluated the relevance of machine learning approaches operating from the set of markers detected by these softwares and show that they present an attractive alternative strategy, allowing to reach better performance for several drugs while significantly reducing the number of candidate mutations to consider.

Comparisons of whole-genome sequencing and phenotypic drug susceptibility testing for Mycobacterium tuberculosis causing MDR-TB and XDR-TB in Thailand

Drug-resistant tuberculosis (TB) is a major public health problem. There is little information regarding the genotypic-phenotypic association of anti-TB drugs, especially for second-line drugs. This study compared phenotypic drug susceptibility testing (DST) with predictions based on whole-genome sequencing (WGS) data for 266 Mycobacterium tuberculosis isolates. Phenotypic DST used the standard proportional method. Clinical isolates of M. tuberculosis collected in Thailand between 1998 and 2013 comprised 51 drug-sensitive strains, six mono-resistant strains, two multiple-resistant strains, 88 multi-drug-resistant strains, 95 pre-extensively drug-resistant strains and 24 extensively drug-resistant strains. WGS analysis was performed using the computer programs PhyResSE and TB-Profiler. TB-Profiler had higher average concordance with phenotypic DST than PhyResSE for both first-line (91.96% vs. 91.4%) and second-line (79.67% vs. 78.20%) anti-TB drugs. The average sensitivity for all anti-TB drugs was also higher (83.13% vs. 72.08%) with slightly lower specificity (83.50% vs. 86.68%). Regardless of the program used, isoniazid, rifampicin and amikacin had the highest concordance with phenotypic DST (96.2%, 93.5% and 95.6%, respectively). Ethambutol, ethionamide and fluoroquinolones had the lowest concordance (87.34%, 81.44% and 73.85%, respectively). Concordance rates of ofloxacin (a second-generation fluoroquinolone), levofloxacin, moxifloxacin and gatifloxacin (third- and fourth-generation fluoroquinolones) were 91.79%, 76.62%, 72.64% and 57.35%, respectively. Discordance between phenotypic and WGS-based DSTs may be due, in part, to the choice of critical concentration and variable reproducibility of the phenotypic tests. It may also be due to limitations of the mutation databases (especially for the second-line drugs) and the analysis program used. Mutations related to fluoroquinolone resistance, especially the later generations, need to be identified.

Genome-wide analysis of Mycobacterium tuberculosis polymorphisms reveals lineage-specific associations with drug resistance

Background

Continuing evolution of the Mycobacterium tuberculosis (Mtb) complex genomes associated with resistance to anti-tuberculosis drugs is threatening tuberculosis disease control efforts. Both multi- and extensively drug resistant Mtb (MDR and XDR, respectively) are increasing in prevalence, but the full set of Mtb genes involved are not known. There is a need for increased sensitivity of genome-wide approaches in order to elucidate the genetic basis of anti-microbial drug resistance and gain a more detailed understanding of Mtb genome evolution in a context of widespread antimicrobial therapy. Population structure within the Mtb complex, due to clonal expansion, lack of lateral gene transfer and low levels of recombination between lineages, may be reducing statistical power to detect drug resistance associated variants.

Results

To investigate the effect of lineage-specific effects on the identification of drug resistance associations, we applied genome-wide association study (GWAS) and convergence-based (PhyC) methods to multiple drug resistance phenotypes of a global dataset of Mtb lineages 2 and 4, using both lineage-wise and combined approaches. We identify both well-established drug resistance variants and novel associations; uniquely identifying associations for both lineage-specific and -combined GWAS analyses. We report 17 potential novel associations between antimicrobial resistance phenotypes and Mtb genomic variants.

Conclusions

For GWAS, both lineage-specific and -combined analyses are useful, whereas PhyC may perform better in contexts of greater diversity. Unique associations with XDR in lineage-specific analyses provide evidence of diverging evolutionary trajectories between lineages 2 and 4 in response to antimicrobial drug therapy.

Electronic supplementary material

The online version of this article (10.1186/s12864-019-5615-3) contains supplementary material, which is available to authorized users.

Whole-Genome Sequencing for Drug Resistance Profile Prediction in Mycobacterium tuberculosis

Whole-genome sequencing allows rapid detection of drug-resistant Mycobacterium tuberculosis isolates. However, the availability of high-quality data linking quantitative phenotypic drug susceptibility testing (DST) and genomic data have thus far been limited. We determined drug resistance profiles of 176 genetically diverse clinical M. tuberculosis isolates from the Democratic Republic of the Congo, Ivory Coast, Peru, Thailand, and Switzerland by quantitative phenotypic DST for 11 antituberculous drugs using the BD Bactec MGIT 960 system and 7H10 agar dilution to generate a cross-validated phenotypic DST readout. We compared DST results with predicted drug resistance profiles inferred by whole-genome sequencing. Classification of strains by the two phenotypic DST methods into resistotype/wild-type populations was concordant in 73 to 99% of cases, depending on the drug. Our data suggest that the established critical concentration (5 mg/liter) for ethambutol resistance (MGIT 960 system) is too high and misclassifies strains as susceptible, unlike 7H10 agar dilution. Increased minimal inhibitory concentrations were explained by mutations identified by whole-genome sequencing. Using whole-genome sequences, we were able to predict quantitative drug resistance levels for the majority of drug resistance mutations. Predicting quantitative levels of drug resistance by whole-genome sequencing was partially limited due to incompletely understood drug resistance mechanisms. The overall sensitivity and specificity of whole-genome-based DST were 86.8% and 94.5%, respectively. Despite some limitations, whole-genome sequencing has the potential to infer resistance profiles without the need for time-consuming phenotypic methods.

Genomic analysis of the emergence of drug-resistant strains of Mycobacterium tuberculosis in the Middle East

Tuberculosis (TB) represents a significant challenge to public health authorities, especially with the emergence of drug-resistant (DR) and multidrug-resistant (MDR) isolates of Mycobacterium tuberculosis. We sought to examine the genomic variations among recently isolated strains of M. tuberculosis in two closely related countries with different population demography in the Middle East. Clinical isolates of M. tuberculosis from both Egypt and Saudi Arabia were subjected to phenotypic and genotypic analysis on gene and genome-wide levels. Isolates with MDR phenotypes were highly prevalent in Egypt (up to 35%) despite its relatively stable population structure (sympatric pattern). MDR-TB isolates were not identified in the isolates from Saudi Arabia despite its active guest worker program (allopatric pattern). However, tuberculosis isolates from Saudi Arabia, where lineage 4 was more prevalent (>65%), showed more diversity than isolates from Egypt, where lineage 3 was the most prevalent (>75%). Phylogenetic and molecular dating analyses indicated that lineages from Egypt were recently diverged (~78 years), whereas those from Saudi Arabia were diverged by over 200 years. Interestingly, DR isolates did not appear to cluster together or spread more widely than drug-sensitive isolates, suggesting poor treatment as the main cause for emergence of drug resistance rather than more virulence or more capacity to persist.

SNP-IT Tool for Identifying Subspecies and Associated Lineages of Mycobacterium tuberculosis Complex

The clinical phenotype of zoonotic tuberculosis and its contribution to the global burden of disease are poorly understood and probably underestimated. This shortcoming is partly because of the inability of currently available laboratory and in silico tools to accurately identify all subspecies of the Mycobacterium tuberculosis complex (MTBC). We present SNPs to Identify TB (SNP-IT), a single-nucleotide polymorphism-based tool to identify all members of MTBC, including animal clades. By applying SNP-IT to a collection of clinical genomes from a UK reference laboratory, we detected an unexpectedly high number of M. orygis isolates. M. orygis is seen at a similar rate to M. bovis, yet M. orygis cases have not been previously described in the United Kingdom. From an international perspective, it is possible that M. orygis is an underestimated zoonosis. Accurate identification will enable study of the clinical phenotype, host range, and transmission mechanisms of all subspecies of MTBC in greater detail.

Overcoming the pitfalls of automatic interpretation of whole genome sequencing data by online tools for the prediction of pyrazinamide resistance in Mycobacterium tuberculosis

OBJECTIVES

Automated online software tools that analyse whole genome sequencing (WGS) data without the need for bioinformatics expertise can motivate the implementation of WGS-based molecular drug susceptibility testing (DST) in routine diagnostic settings for tuberculosis (TB). Pyrazinamide (PZA) is a key drug for current and future TB treatment regimens; however, it was reported that predictive power for PZA resistance by the available tools is low. Therefore, this low predictive power may make users hesitant to use the tools. This study aimed to elucidate why and to uncover the real performance of the tools when taking into account their variation calling lists (manual inspection), not just their automated reporting system (default setting) that was evaluated by previous studies.

METHODS

WGS data from 191 datasets comprising 108 PZA-resistant and 83 susceptible strains were used to evaluate the potential performance of the available online tools (TB Profiler, TGS-TB, PhyResSE, and CASTB) for predicting phenotypic PZA resistance.

RESULTS

When taking into consideration the variation calling lists, 73 variants in total (47 non-synonymous mutations and 26 indels) in pncA were detected by TGS-TB and PhyResSE, covering all mutations for the 108 PZA-resistant strains. The 73 variants were confirmed by Sanger sequencing. TB Profiler also detected all but three complete loss, two large deletion at the 3'-end, and one relatively large insertion of pncA. On the other hand, many of the 73 variants were lacking in the automated reporting systems except by TGS-TB; of these variants, CASTB detected only 20. By applying the 'non-wild type sequence' approach for predicting PZA resistance, accuracy of the results significantly improved compared with that of the automated results obtained by each tool.

CONCLUSION

Users can obtain more accurate predictions for PZA resistance than previously reported by manually checking the results and applying the 'non-wild type sequence' approach.

Genome-Based Prediction of Bacterial Antibiotic Resistance

Clinical microbiology has long relied on growing bacteria in culture to determine antimicrobial susceptibility profiles, but the use of whole-genome sequencing for antibiotic susceptibility testing (WGS-AST) is now a powerful alternative. This review discusses the technologies that made this possible and presents results from recent studies to predict resistance based on genome sequences.

Clinical microbiology has long relied on growing bacteria in culture to determine antimicrobial susceptibility profiles, but the use of whole-genome sequencing for antibiotic susceptibility testing (WGS-AST) is now a powerful alternative. This review discusses the technologies that made this possible and presents results from recent studies to predict resistance based on genome sequences. We examine differences between calling antibiotic resistance profiles by the simple presence or absence of previously known genes and single-nucleotide polymorphisms (SNPs) against approaches that deploy machine learning and statistical models. Often, the limitations to genome-based prediction arise from limitations of accuracy of culture-based AST in addition to an incomplete knowledge of the genetic basis of resistance. However, we need to maintain phenotypic testing even as genome-based prediction becomes more widespread to ensure that the results do not diverge over time. We argue that standardization of WGS-AST by challenge with consistently phenotyped strain sets of defined genetic diversity is necessary to compare the efficacy of methods of prediction of antibiotic resistance based on genome sequences.

Evaluation of a gene-by-gene approach for prospective whole-genome sequencing-based surveillance of multidrug resistant Mycobacterium tuberculosis

Whole-genome sequencing (WGS) offers unprecedented resolution for tracking Mycobacterium tuberculosis transmission and antibiotic-resistance spread. Still, the establishment of standardized WGS-based pipelines and the definition of epidemiological clusters based on genetic relatedness are under discussion. We aimed to implement a dynamic gene-by-gene approach, fully relying on freely available software, for prospective WGS-based tuberculosis surveillance, demonstrating its application for detecting transmission chains by retrospectively analysing all M/XDR strains isolated in 2013-2017 in Portugal. We observed a good correlation between genetic relatedness and epidemiological links, with strongly epilinked clusters displaying mean pairwise allele differences (AD) always below 0.3% (ratio of mean AD over the total number of shared loci between same-cluster strains). This data parallels the genetic distances acquired by the core-SNV analysis, while providing higher resolution and epidemiological concordance than MIRU-VNTR genotyping. The dynamic analysis of strain sub-sets (i.e., increasing the number of shared loci within each sub-set) also strengthens the confidence in detecting epilinked clusters. This gene-by-gene strategy also offers several practical benefits (e.g., reliance on freely-available software, scalability and low computational requirements) that further consolidated its suitability for a timely and robust prospective WGS-based laboratory surveillance of M/XDR-TB cases.

Whole genome sequencing for drug resistance determination in Mycobacterium tuberculosis

South Africa remains challenged with a high tuberculosis burden accompanied by an increase in drug resistant cases. We assessed the use of the Illumina MiSeq, a next-generation sequencing platform for whole genome sequencing, followed by bioinformatic analysis using a commercial software package to determine resistance to selected drugs used for Mycobacterium tuberculosis treatment in our setting. Whole genome sequencing shows potential as a diagnostic platform for the detection of drug resistance in Mycobacterium tuberculosis with the provision of information for several drugs simultaneously.

Genomic prediction of tuberculosis drug-resistance: benchmarking existing databases and prediction algorithms

Background

It is possible to predict whether a tuberculosis (TB) patient will fail to respond to specific antibiotics by sequencing the genome of the infecting Mycobacterium tuberculosis (Mtb) and observing whether the pathogen carries specific mutations at drug-resistance sites. This advancement has led to the collation of TB databases such as PATRIC and ReSeqTB that possess both whole genome sequences and drug resistance phenotypes of infecting Mtb isolates. Bioinformatics tools have also been developed to predict drug resistance from whole genome sequencing (WGS) data.

Here, we evaluate the performance of four popular tools (TBProfiler, MyKrobe, KvarQ, PhyResSE) with 6746 isolates compiled from publicly available databases, and subsequently identify highly probable phenotyping errors in the databases by genetically predicting the drug phenotypes using all four software.

Results

Our results show that these bioinformatics tools generally perform well in predicting the resistance status for two key first-line agents (isoniazid, rifampicin), but the accuracy is lower for second-line injectables and fluoroquinolones. The error rates in the databases are also non-trivial, reaching as high as 31.1% for prothionamide, and that phenotypes from ReSeqTB are more susceptible to errors.

Conclusions

The good performance of the automated software for drug resistance prediction from TB WGS data shown in this study further substantiates the usefulness and promise of utilising genetic data to accurately profile TB drug resistance, thereby reducing misdiagnoses arising from error-prone culture-based drug susceptibility testing.

Electronic supplementary material

The online version of this article (10.1186/s12859-019-2658-z) contains supplementary material, which is available to authorized users.

Investigation of false-positive Mycobacterium tuberculosis culture tests using whole genome sequencing

Although accurate identification of Mycobacterium tuberculosis is the gold standard for tuberculosis (TB) diagnosis, there have been several reports of false-positive results. After identifying a sudden increase in extensively drug-resistant TB, false-positive mycobacterial culture tests were suspected, and we contacted the supranational reference center for molecular typing. In silico genotyping tests showed that isolates from all five patients had an identical genotype pattern, and all harbored the same Beijing strain based on sequence-based phylogenic analysis and drug-resistant single nucleotide polymorphisms (SNPs) analysis. We also used whole genome sequencing (WGS) to compare the SNPs of all isolates with a reference genome, and all were identical. We adapted WGS to efficiently detect false-positive MTB culture tests.

First insights on the genetic diversity of MDR Mycobacterium tuberculosis in Lebanon

Background

Lebanon hosts a heterogeneous population coming from underdeveloped and developing countries, resulting in increasing incidences of tuberculosis over the past years. The genetic heterogeneity and lineages associated with tuberculosis, along with their resistance determinants have not been studied at the genomic level previously in the region.

Methods

Isolates were recovered from the American University of Beirut Medical Center (AUBMC). Antimicrobial susceptibility profiles were determined using the MGIT automated system for the first-line drugs at AUBMC, while second-line drug susceptibility was tested at Mayo Clinic Laboratories. Whole Genome Sequencing (WGS) was performed to classify mycobacterial lineages and highlight single nucleotide mutations causing resistance to both 1st line and 2nd line antimicrobials. wgSNP analysis provided insights on the phylogeny of the isolates along with spoligotyping and core genomic SNVs, IS6110 insertion sites, and variable number tandem repeats (VNTR).

Results

The analyzed isolates carry distinct resistance determinants to isoniazid, rifampicin, ethambutol, quinolones, and streptomycin. The isolates belonged to different lineages including the Euro/American lineage (Lineage 4) (53.8%), M. bovis (15.4%) and Delhi/Central Asia (Lineage 1) (15.4%), Beijing/East Asia (Lineage 2) (7.7%), and East Africa/Indian Ocean lineage (Lineage 3) (7.7%) showing great phylogenetic differences at the genomic level.

Conclusions

The population diversity in Lebanon holds an equally diverse and uncharacterized population of drug resistant mycobacteria. To achieve the WHO “END-TB” milestones of 2025 and 2035, Lebanon must decrease TB incidences by 95% in the next decade. This can only be done through WGS-based patient centered diagnosis with higher throughput and genomic resolution to improve treatment outcomes and to monitor transmission patterns.

Electronic supplementary material

The online version of this article (10.1186/s12879-018-3626-3) contains supplementary material, which is available to authorized users.

Transmission of drug-resistant tuberculosis in HIV-endemic settings

The emergence and expansion of the multidrug-resistant tuberculosis epidemic is a threat to the global control of tuberculosis. Multidrug-resistant tuberculosis is the result of the selection of resistance-conferring mutations during inadequate antituberculosis treatment. However, HIV has a profound effect on the natural history of tuberculosis, manifesting in an increased rate of disease progression, leading to increased transmission and amplification of multidrug-resistant tuberculosis. Interventions specific to HIV-endemic areas are urgently needed to block tuberculosis transmission. These interventions should include a combination of rapid molecular diagnostics and improved chemotherapy to shorten the duration of infectiousness, implementation of infection control measures, and active screening of multidrug-resistant tuberculosis contacts, with prophylactic regimens for individuals without evidence of disease. Development and improvement of the efficacy of interventions will require a greater understanding of the factors affecting the transmission of multidrug-resistant tuberculosis in HIV-endemic settings, including population-based molecular epidemiology studies. In this Series article, we review what we know about the transmission of multidrug-resistant tuberculosis in settings with high burdens of HIV and define the research priorities required to develop more effective interventions, to diminish ongoing transmission and the amplification of drug resistance.

Mycobacterium tuberculosis Next-Generation Whole Genome Sequencing: Opportunities and Challenges

Mycobacterium tuberculosis drug resistance is a threat to global tuberculosis (TB) control. Comprehensive and timely drug susceptibility determination is critical to inform appropriate treatment of drug-resistant tuberculosis (DR-TB). Phenotypic drug susceptibility testing (DST) is the gold standard for M. tuberculosis drug resistance determination. M. tuberculosis whole genome sequencing (WGS) has the potential to be a one-stop method for both comprehensive DST and epidemiological investigations. We discuss in this review the tremendous opportunities that next-generation WGS presents in terms of understanding the molecular epidemiology of tuberculosis and mechanisms of drug resistance. The potential clinical value and public health impact in the areas of DST for patient management and tracing of transmission chains for timely public health intervention are also discussed. We present the current challenges for the implementation of WGS in low and middle-income settings. WGS analysis has already been adapted routinely in laboratories to inform patient management and public health interventions in low burden high-income settings such as the United Kingdom. We predict that the technology will be adapted similarly in high burden settings where the impact on the epidemic will be greatest.

From Theory to Practice: Translating Whole-Genome Sequencing (WGS) into the Clinic

Hospitals worldwide are facing an increasing incidence of hard-to-treat infections. Limiting infections and providing patients with optimal drug regimens require timely strain identification as well as virulence and drug-resistance profiling. Additionally, prophylactic interventions based on the identification of environmental sources of recurrent infections (e.g., contaminated sinks) and reconstruction of transmission chains (i.e., who infected whom) could help to reduce the incidence of nosocomial infections. WGS could hold the key to solving these issues. However, uptake in the clinic has been slow. Some major scientific and logistical challenges need to be solved before WGS fulfils its potential in clinical microbial diagnostics. In this review we identify major bottlenecks that need to be resolved for WGS to routinely inform clinical intervention and discuss possible solutions.

Advances in the molecular diagnosis of tuberculosis: From probes to genomes

Tuberculosis, disease caused by Mycobacterium tuberculosis, is currently the leading cause of death by a single infectious agent worldwide. Early, rapid and accurate identification of M. tuberculosis and the determination of drug susceptibility is essential for the treatment and management of this disease. Tuberculosis diagnosis is mainly based on chest radiography, smear microscopy and bacteriological culture. Smear microscopy has variable sensitivity, mainly in patients co-infected with the human immunodeficiency virus (HIV). Conventional culture for M. tuberculosis isolation, identification and drug susceptibility testing requires several weeks owning to the slow growth of M. tuberculosis. The delay in the time to results drives the prolongation of potentially inappropriate antituberculosis therapy contributing to the emergence of drug resistance, reducing treatment options and increasing treatment duration and associated costs, resulting in increased mortality and morbidity. For these reasons, novel diagnostic methods are need for timely identification of M. tuberculosis and determination of the antibiotic susceptibility profile of the infecting strain. Molecular methods offer enhanced sensitivity and specificity, early detection and the capacity to detect mixed infections. These technologies have improved turnaround time, cost effectiveness and are amenable for point-of-care testing. However, although these methods produce results within hours from sample collection, the phenotypic susceptibility testing is still needed for the determination of drug susceptibility and quantify the susceptibility levels of a given strain towards individual antibiotics. This review presents the history, advances and forthcoming promises in the molecular diagnosis of tuberculosis. An overview on the general principles, diagnostic value and the main advantages and disadvantages of the molecular methods used for the detection and identification of M. tuberculosis and its associated disease, is provided. It will be also discussed how the current phenotypic methods should be used in combination with the genotypic methods for rapid antituberculosis susceptibility testing.

Highlights on the Application of Genomics and Bioinformatics in the Fight Against Infectious Diseases: Challenges and Opportunities in Africa

Genomics and bioinformatics are increasingly contributing to our understanding of infectious diseases caused by bacterial pathogens such as Mycobacterium tuberculosis and parasites such as Plasmodium falciparum. This ranges from investigations of disease outbreaks and pathogenesis, host and pathogen genomic variation, and host immune evasion mechanisms to identification of potential diagnostic markers and vaccine targets. High throughput genomics data generated from pathogens and animal models can be combined with host genomics and patients’ health records to give advice on treatment options as well as potential drug and vaccine interactions. However, despite accounting for the highest burden of infectious diseases, Africa has the lowest research output on infectious disease genomics. Here we review the contributions of genomics and bioinformatics to the management of infectious diseases of serious public health concern in Africa including tuberculosis (TB), dengue fever, malaria and filariasis. Furthermore, we discuss how genomics and bioinformatics can be applied to identify drug and vaccine targets. We conclude by identifying challenges to genomics research in Africa and highlighting how these can be overcome where possible.

MTBseq: a comprehensive pipeline for whole genome sequence analysis of Mycobacterium tuberculosis complex isolates

Analyzing whole-genome sequencing data of Mycobacterium tuberculosis complex (MTBC) isolates in a standardized workflow enables both comprehensive antibiotic resistance profiling and outbreak surveillance with highest resolution up to the identification of recent transmission chains. Here, we present MTBseq, a bioinformatics pipeline for next-generation genome sequence data analysis of MTBC isolates. Employing a reference mapping based workflow, MTBseq reports detected variant positions annotated with known association to antibiotic resistance and performs a lineage classification based on phylogenetic single nucleotide polymorphisms (SNPs). When comparing multiple datasets, MTBseq provides a joint list of variants and a FASTA alignment of SNP positions for use in phylogenomic analysis, and identifies groups of related isolates. The pipeline is customizable, expandable and can be used on a desktop computer or laptop without any internet connection, ensuring mobile usage and data security. MTBseq and accompanying documentation is available from https://github.com/ngs-fzb/MTBseq_source.

A fast and agnostic method for bacterial genome-wide association studies: Bridging the gap between k-mers and genetic events

Genome-wide association study (GWAS) methods applied to bacterial genomes have shown promising results for genetic marker discovery or detailed assessment of marker effect. Recently, alignment-free methods based on k-mer composition have proven their ability to explore the accessory genome. However, they lead to redundant descriptions and results which are sometimes hard to interpret. Here we introduce DBGWAS, an extended k-mer-based GWAS method producing interpretable genetic variants associated with distinct phenotypes. Relying on compacted De Bruijn graphs (cDBG), our method gathers cDBG nodes, identified by the association model, into subgraphs defined from their neighbourhood in the initial cDBG. DBGWAS is alignment-free and only requires a set of contigs and phenotypes. In particular, it does not require prior annotation or reference genomes. It produces subgraphs representing phenotype-associated genetic variants such as local polymorphisms and mobile genetic elements (MGE). It offers a graphical framework which helps interpret GWAS results. Importantly it is also computationally efficient-experiments took one hour and a half on average. We validated our method using antibiotic resistance phenotypes for three bacterial species. DBGWAS recovered known resistance determinants such as mutations in core genes in Mycobacterium tuberculosis, and genes acquired by horizontal transfer in Staphylococcus aureus and Pseudomonas aeruginosa-along with their MGE context. It also enabled us to formulate new hypotheses involving genetic variants not yet described in the antibiotic resistance literature. An open-source tool implementing DBGWAS is available at https://gitlab.com/leoisl/dbgwas.

Predicting bacterial resistance from whole-genome sequences using k-mers and stability selection

Background

Several studies demonstrated the feasibility of predicting bacterial antibiotic resistance phenotypes from whole-genome sequences, the prediction process usually amounting to detecting the presence of genes involved in antibiotic resistance mechanisms, or of specific mutations, previously identified from a training panel of strains, within these genes. We address the problem from the supervised statistical learning perspective, not relying on prior information about such resistance factors. We rely on a k-mer based genotyping scheme and a logistic regression model, thereby combining several k-mers into a probabilistic model. To identify a small yet predictive set of k-mers, we rely on the stability selection approach (Meinshausen et al., J R Stat Soc Ser B 72:417–73, 2010), that consists in penalizing logistic regression models with a Lasso penalty, coupled with extensive resampling procedures.

Results

Using public datasets, we applied the resulting classifiers to two bacterial species and achieved predictive performance equivalent to state of the art. The models are extremely sparse, involving 1 to 8 k-mers per antibiotic, hence are remarkably easy and fast to evaluate on new genomes (from raw reads to assemblies).

Conclusion

Our proof of concept therefore demonstrates that stability selection is a powerful approach to investigate bacterial genotype-phenotype relationships.

Electronic supplementary material

The online version of this article (10.1186/s12859-018-2403-z) contains supplementary material, which is available to authorized users.

Global expansion of Mycobacterium tuberculosis lineage 4 shaped by colonial migration and local adaptation

On the basis of population genomic and phylogeographic analyses of 1669 Mycobacterium tuberculosis lineage 4 (L4) genomes, we find that dispersal of L4 has been completely dominated by historical migrations out of Europe. We demonstrate an intimate temporal relationship between European colonial expansion into Africa and the Americas and the spread of L4 tuberculosis (TB). Markedly, in the age of antibiotics, mutations conferring antimicrobial resistance overwhelmingly emerged locally (at the level of nations), with minimal cross-border transmission of resistance. The latter finding was found to reflect the relatively recent emergence of these mutations, as a similar degree of local restriction was observed for susceptible variants emerging on comparable time scales. The restricted international transmission of drug-resistant TB suggests that containment efforts at the level of individual countries could be successful.

Drug-resistant tuberculosis: challenges and opportunities for diagnosis and treatment

With an estimated incidence of 490000 cases in 2016, multidrug resistant tuberculosis (TB), against which key first-line anti-tuberculars are less efficacious, presents major challenges for global health. Poor treatment outcomes coupled with a yawning treatment gap between those in need of second-line therapy and those who receive it, underscore the urgent need for new approaches to tackle the scourge of drug-resistant TB. Against this background, significant progress has been made in understanding the complex biology of TB drug resistance and disease pathogenesis, and in establishing a pipeline for delivering new drugs and drug combinations. In this review, we highlight the challenges of drug-resistant TB and the ways in which new advances could be harnessed to improve treatment outcomes.

Genetics and roadblocks of drug resistant tuberculosis

Considering the extensive evolutionary history of Mycobacterium tuberculosis, anti-Tuberculosis (TB) drug therapy exerts a recent selective pressure. However, in a microorganism devoid of horizontal gene transfer and with a strictly clonal populational structure such as M. tuberculosis the usual, but not sole, path to overcome drug susceptibility is through de novo mutations on a relatively strict set of genes. The possible allelic diversity that can be associated with drug resistance through several mechanisms such as target alteration or target overexpression, will dictate how these genes can become associated with drug resistance. The success demonstrated by this pathogenic microbe in this latter process and its ability to spread is currently one of the major obstacles to an effective TB elimination. This article reviews the action mechanism of the more important anti-TB drugs, including bedaquiline and delamanid, along with new findings on specific resistance mechanisms. With the development, validation and endorsement of new in vitro molecular tests for drug resistance, knowledge on these resistance mechanisms and microevolutionary dynamics leading to the emergence and fixation of drug resistance mutations within the host is highly important. Additionally, the fitness toll imposed by resistance development is also herein discussed together with known compensatory mechanisms. By elucidating the possible mechanisms that enable one strain to reacquire the original fitness levels, it will be theoretically possible to make more informed decisions and develop novel strategies that can force M. tuberculosis microevolutionary trajectory down through a path of decreasing fitness levels.