A dose-response model for statistical analysis of chemical genetic interactions in CRISPRi screens.

An important application of CRISPR interference (CRISPRi) technology is for identifying chemical-genetic interactions (CGIs). Discovery of genes that interact with exposure to antibiotics can yield insights to drug targets and mechanisms of action or resistance. The objective is to identify CRISPRi mutants whose relative abundance is suppressed (or enriched) in the presence of a drug when the target protein is depleted, reflecting synergistic behavior. Different sgRNAs for a given target can induce a wide range of protein depletion and differential effects on growth rate. The effect of sgRNA strength can be partially predicted based on sequence features. However, the actual growth phenotype depends on the sensitivity of cells to depletion of the target protein. For essential genes, sgRNA efficiency can be empirically measured by quantifying effects on growth rate. We observe that the most efficient sgRNAs are not always optimal for detecting synergies with drugs. sgRNA efficiency interacts in a non-linear way with drug sensitivity, producing an effect where the concentration-dependence is maximized for sgRNAs of intermediate strength (and less so for sgRNAs that induce too much or too little target depletion). To capture this interaction, we propose a novel statistical method called CRISPRi-DR (for Dose-Response model) that incorporates both sgRNA efficiencies and drug concentrations in a modified dose-response equation. We use CRISPRi-DR to re-analyze data from a recent CGI experiment in Mycobacterium tuberculosis to identify genes that interact with antibiotics. This approach can be generalized to non-CGI datasets, which we show via an CRISPRi dataset for E. coli growth on different carbon sources. The performance is competitive with the best of several related analytical methods. However, for noisier datasets, some of these methods generate far more significant interactions, likely including many false positives, whereas CRISPRi-DR maintains higher precision, which we observed in both empirical and simulated data.

Weak links: Advancing target-based drug discovery by identifying the most vulnerable targets.

Mycobacterium tuberculosis remains the most common infectious killer worldwide despite decades of antitubercular drug development. Effectively controlling the tuberculosis (TB) pandemic will require innovation in drug discovery. In this review, we provide a brief overview of the two main approaches to discovering new TB drugs-phenotypic screens and target-based drug discovery-and outline some of the limitations of each method. We then explore recent advances in genetic tools that aim to overcome some of these limitations. In particular, we highlight a novel metric to prioritize essential targets, termed vulnerability. Stratifying targets based on their vulnerability presents new opportunities for future target-based drug discovery campaigns.

Compensatory evolution in NusG improves fitness of drug-resistant M. tuberculosis.

Drug-resistant bacteria are emerging as a global threat, despite frequently being less fit than their drug-susceptible ancestors1-8. Here we sought to define the mechanisms that drive or buffer the fitness cost of rifampicin resistance (RifR) in the bacterial pathogen Mycobacterium tuberculosis (Mtb). Rifampicin inhibits RNA polymerase (RNAP) and is a cornerstone of modern short-course tuberculosis therapy9,10. However, RifR Mtb accounts for one-quarter of all deaths due to drug-resistant bacteria11,12. We took a comparative functional genomics approach to define processes that are differentially vulnerable to CRISPR interference (CRISPRi) inhibition in RifR Mtb. Among other hits, we found that the universally conserved transcription factor NusG is crucial for the fitness of RifR Mtb. In contrast to its role in Escherichia coli, Mtb NusG has an essential RNAP pro-pausing function mediated by distinct contacts with RNAP and the DNA13. We find this pro-pausing NusG-RNAP interface to be under positive selection in clinical RifR Mtb isolates. Mutations in the NusG-RNAP interface reduce pro-pausing activity and increase fitness of RifR Mtb. Collectively, these results define excessive RNAP pausing as a molecular mechanism that drives the fitness cost of RifR in Mtb, identify a new mechanism of compensation to overcome this cost, suggest rational approaches to exacerbate the fitness cost, and, more broadly, could inform new therapeutic approaches to develop drug combinations to slow the evolution of RifR in Mtb.

Beyond antibiotic resistance: The whiB7 transcription factor coordinates an adaptive response to alanine starvation in mycobacteria.

Pathogenic mycobacteria are a significant cause of morbidity and mortality worldwide. The conserved whiB7 stress response reduces the effectiveness of antibiotic therapy by activating several intrinsic antibiotic resistance mechanisms. Despite our comprehensive biochemical understanding of WhiB7, the complex set of signals that induce whiB7 expression remain less clear. We employed a reporter-based, genome-wide CRISPRi epistasis screen to identify a diverse set of 150 mycobacterial genes whose inhibition results in constitutive whiB7 expression. We show that whiB7 expression is determined by the amino acid composition of the 5' regulatory uORF, thereby allowing whiB7 to sense amino acid starvation. Although deprivation of many amino acids can induce whiB7, whiB7 specifically coordinates an adaptive response to alanine starvation by engaging in a feedback loop with the alanine biosynthetic enzyme, aspC. These findings describe a metabolic function for whiB7 and help explain its evolutionary conservation across mycobacterial species occupying diverse ecological niches.

A dose-response model for statistical analysis of chemical genetic interactions in CRISPRi screens.

An important application of CRISPR interference (CRISPRi) technology is for identifying chemical-genetic interactions (CGIs). Discovery of genes that interact with exposure to antibiotics can yield insights to drug targets and mechanisms of action or resistance. The objective is to identify CRISPRi mutants whose relative abundance is suppressed (or enriched) in the presence of a drug when the target protein is depleted, reflecting synergistic behavior. Different sgRNAs for a given target can induce a wide range of protein depletion and differential effects on growth rate. The effect of sgRNA strength can be partially predicted based on sequence features. However, the actual growth phenotype depends on the sensitivity of cells to depletion of the target protein. For essential genes, sgRNA efficiency can be empirically measured by quantifying effects on growth rate. We observe that the most efficient sgRNAs are not always optimal for detecting synergies with drugs. sgRNA efficiency interacts in a non-linear way with drug sensitivity, producing an effect where the concentration-dependence is maximized for sgRNAs of intermediate strength (and less so for sgRNAs that induce too much or too little target depletion). To capture this interaction, we propose a novel statistical method called CRISPRi-DR (for Dose-Response model) that incorporates both sgRNA efficiencies and drug concentrations in a modified dose-response equation. We use CRISPRi-DR to re-analyze data from a recent CGI experiment in Mycobacterium tuberculosis to identify genes that interact with antibiotics. This approach can be generalized to non-CGI datasets, which we show via an CRISPRi dataset for E. coli growth on different carbon sources. The performance is competitive with the best of several related analytical methods. However, for noisier datasets, some of these methods generate far more significant interactions, likely including many false positives, whereas CRISPRi-DR maintains higher precision, which we observed in both empirical and simulated data.

Beyond antibiotic resistance: the whiB7 transcription factor coordinates an adaptive response to alanine starvation in mycobacteria.

Pathogenic mycobacteria are a significant cause of morbidity and mortality worldwide. These bacteria are highly intrinsically drug resistant, making infections challenging to treat. The conserved whiB7 stress response is a key contributor to mycobacterial intrinsic drug resistance. Although we have a comprehensive structural and biochemical understanding of WhiB7, the complex set of signals that activate whiB7 expression remain less clear. It is believed that whiB7 expression is triggered by translational stalling in an upstream open reading frame (uORF) within the whiB7 5' leader, leading to antitermination and transcription into the downstream whiB7 ORF. To define the signals that activate whiB7, we employed a genome-wide CRISPRi epistasis screen and identified a diverse set of 150 mycobacterial genes whose inhibition results in constitutive whiB7 activation. Many of these genes encode amino acid biosynthetic enzymes, tRNAs, and tRNA synthetases, consistent with the proposed mechanism for whiB7 activation by translational stalling in the uORF. We show that the ability of the whiB7 5' regulatory region to sense amino acid starvation is determined by the coding sequence of the uORF. The uORF shows considerable sequence variation among different mycobacterial species, but it is universally and specifically enriched for alanine. Providing a potential rationalization for this enrichment, we find that while deprivation of many amino acids can activate whiB7 expression, whiB7 specifically coordinates an adaptive response to alanine starvation by engaging in a feedback loop with the alanine biosynthetic enzyme, aspC. Our results provide a holistic understanding of the biological pathways that influence whiB7 activation and reveal an extended role for the whiB7 pathway in mycobacterial physiology, beyond its canonical function in antibiotic resistance. These results have important implications for the design of combination drug treatments to avoid whiB7 activation, as well as help explain the conservation of this stress response across a wide range of pathogenic and environmental mycobacteria.

Mutations in rv0678 Confer Low-Level Resistance to Benzothiazinone DprE1 Inhibitors in Mycobacterium tuberculosis.

Tuberculosis (TB) is the leading cause of death from any bacterial infection, causing 1.5 million deaths worldwide each year. Due to the emergence of drug-resistant strains of Mycobacterium tuberculosis (Mtb) there have been significant efforts aimed at developing novel drugs to treat TB. One promising drug target in Mtb is the arabinogalactan biosynthetic enzyme DprE1, and there have been over a dozen unique chemical scaffolds identified which inhibit the activity of this protein. Among the most promising lead compounds are the benzothiazinones BTZ043 and PBTZ169, both of which are currently in or have completed phase IIa clinical trials. Due to the potential clinical utility of these drugs, we sought to identify potential synergistic interactions and new mechanisms of resistance using a genome-scale CRISPRi chemical-genetic screen with PBTZ169. We found that knockdown of rv0678, the negative regulator of the mmpS5/L5 drug efflux pump, confers resistance to PBTZ169. Mutations in rv0678 are the most common form of resistance to bedaquiline and there is already abundant evidence of these mutations emerging in bedaquiline-treated patients. We confirmed that rv0678 mutations from clinical isolates confer low level cross-resistance to BTZ043 and PBTZ169. While it is yet unclear whether rv0678 mutations would render benzothiazinones ineffective in treating TB, these results highlight the importance of monitoring for clinically prevalent rv0678 mutations during ongoing BTZ043 and PBTZ169 clinical trials.

CRISPRi chemical genetics and comparative genomics identify genes mediating drug potency in Mycobacterium tuberculosis.

Mycobacterium tuberculosis (Mtb) infection is notoriously difficult to treat. Treatment efficacy is limited by Mtb's intrinsic drug resistance, as well as its ability to evolve acquired resistance to all antituberculars in clinical use. A deeper understanding of the bacterial pathways that influence drug efficacy could facilitate the development of more effective therapies, identify new mechanisms of acquired resistance, and reveal overlooked therapeutic opportunities. Here we developed a CRISPR interference chemical-genetics platform to titrate the expression of Mtb genes and quantify bacterial fitness in the presence of different drugs. We discovered diverse mechanisms of intrinsic drug resistance, unveiling hundreds of potential targets for synergistic drug combinations. Combining chemical genetics with comparative genomics of Mtb clinical isolates, we further identified several previously unknown mechanisms of acquired drug resistance, one of which is associated with a multidrug-resistant tuberculosis outbreak in South America. Lastly, we found that the intrinsic resistance factor whiB7 was inactivated in an entire Mtb sublineage endemic to Southeast Asia, presenting an opportunity to potentially repurpose the macrolide antibiotic clarithromycin to treat tuberculosis. This chemical-genetic map provides a rich resource to understand drug efficacy in Mtb and guide future tuberculosis drug development and treatment.

CRISPR Interference Reveals That All-Trans-Retinoic Acid Promotes Macrophage Control of Mycobacterium tuberculosis by Limiting Bacterial Access to Cholesterol and Propionyl Coenzyme A.

Macrophages are a protective replicative niche for Mycobacterium tuberculosis (Mtb) but can kill the infecting bacterium when appropriately activated. To identify mechanisms of clearance, we compared levels of bacterial restriction by human macrophages after treatment with 26 compounds, including some currently in clinical trials for tuberculosis. All-trans-retinoic acid (ATRA), an active metabolite of vitamin A, drove the greatest increase in Mtb control. Bacterial clearance was transcriptionally and functionally associated with changes in macrophage cholesterol trafficking and lipid metabolism. To determine how these macrophage changes affected bacterial control, we performed the first Mtb CRISPR interference screen in an infection model, identifying Mtb genes specifically required to survive in ATRA-activated macrophages. These data showed that ATRA treatment starves Mtb of cholesterol and the downstream metabolite propionyl coenzyme A (propionyl-CoA). Supplementation with sources of propionyl-CoA, including cholesterol, abrogated the restrictive effect of ATRA. This work demonstrates that targeting the coupled metabolism of Mtb and the macrophage improves control of infection and that it is possible to genetically map the mode of bacterial death using CRISPR interference. IMPORTANCE Tuberculosis, caused by the bacterium Mycobacterium tuberculosis, is a leading cause of death due to infectious disease. Improving the immune response to tuberculosis holds promise for fighting the disease but is limited by our lack of knowledge as to how the immune system kills M. tuberculosis. Our research identifies a potent way to make relevant immune cells more effective at fighting M. tuberculosis and then uses paired human and bacterial genomic methods to determine the mechanism of that improved bacterial clearance.

An improved statistical method to identify chemical-genetic interactions by exploiting concentration-dependence.

Chemical-genetics (C-G) experiments can be used to identify interactions between inhibitory compounds and bacterial genes, potentially revealing the targets of drugs, or other functionally interacting genes and pathways. C-G experiments involve constructing a library of hypomorphic strains with essential genes that can be knocked-down, treating it with an inhibitory compound, and using high-throughput sequencing to quantify changes in relative abundance of individual mutants. The hypothesis is that, if the target of a drug or other genes in the same pathway are present in the library, such genes will display an excessive fitness defect due to the synergy between the dual stresses of protein depletion and antibiotic exposure. While assays at a single drug concentration are susceptible to noise and can yield false-positive interactions, improved detection can be achieved by requiring that the synergy between gene and drug be concentration-dependent. We present a novel statistical method based on Linear Mixed Models, called CGA-LMM, for analyzing C-G data. The approach is designed to capture the dependence of the abundance of each gene in the hypomorph library on increasing concentrations of drug through slope coefficients. To determine which genes represent candidate interactions, CGA-LMM uses a conservative population-based approach in which genes with negative slopes are considered significant only if they are outliers with respect to the rest of the population (assuming that most genes in the library do not interact with a given inhibitor). We applied the method to analyze 3 independent hypomorph libraries of M. tuberculosis for interactions with antibiotics with anti-tubercular activity, and we identify known target genes or expected interactions for 7 out of 9 drugs where relevant interacting genes are known.

Genome-wide gene expression tuning reveals diverse vulnerabilities of M. tuberculosis.

Antibacterial agents target the products of essential genes but rarely achieve complete target inhibition. Thus, the all-or-none definition of essentiality afforded by traditional genetic approaches fails to discern the most attractive bacterial targets: those whose incomplete inhibition results in major fitness costs. In contrast, gene "vulnerability" is a continuous, quantifiable trait that relates the magnitude of gene inhibition to the effect on bacterial fitness. We developed a CRISPR interference-based functional genomics method to systematically titrate gene expression in Mycobacterium tuberculosis (Mtb) and monitor fitness outcomes. We identified highly vulnerable genes in various processes, including novel targets unexplored for drug discovery. Equally important, we identified invulnerable essential genes, potentially explaining failed drug discovery efforts. Comparison of vulnerability between the reference and a hypervirulent Mtb isolate revealed incomplete conservation of vulnerability and that differential vulnerability can predict differential antibacterial susceptibility. Our results quantitatively redefine essential bacterial processes and identify high-value targets for drug development.

Statistical analysis of variability in TnSeq data across conditions using zero-inflated negative binomial regression.

BACKGROUND: Deep sequencing of transposon mutant libraries (or TnSeq) is a powerful method for probing essentiality of genomic loci under different environmental conditions. Various analytical methods have been described for identifying conditionally essential genes whose tolerance for insertions varies between two conditions. However, for large-scale experiments involving many conditions, a method is needed for identifying genes that exhibit significant variability in insertions across multiple conditions. RESULTS: In this paper, we introduce a novel statistical method for identifying genes with significant variability of insertion counts across multiple conditions based on Zero-Inflated Negative Binomial (ZINB) regression. Using likelihood ratio tests, we show that the ZINB distribution fits TnSeq data better than either ANOVA or a Negative Binomial (in a generalized linear model). We use ZINB regression to identify genes required for infection of M. tuberculosis H37Rv in C57BL/6 mice. We also use ZINB to perform a analysis of genes conditionally essential in H37Rv cultures exposed to multiple antibiotics. CONCLUSIONS: Our results show that, not only does ZINB generally identify most of the genes found by pairwise resampling (and vastly out-performs ANOVA), but it also identifies additional genes where variability is detectable only when the magnitudes of insertion counts are treated separately from local differences in saturation, as in the ZINB model.

Chemical Genetic Interaction Profiling Reveals Determinants of Intrinsic Antibiotic Resistance in Mycobacterium tuberculosis.

Chemotherapy for tuberculosis (TB) is lengthy and could benefit from synergistic adjuvant therapeutics that enhance current and novel drug regimens. To identify genetic determinants of intrinsic antibiotic susceptibility in Mycobacterium tuberculosis, we applied a chemical genetic interaction (CGI) profiling approach. We screened a saturated transposon mutant library and identified mutants that exhibit altered fitness in the presence of partially inhibitory concentrations of rifampin, ethambutol, isoniazid, vancomycin, and meropenem, antibiotics with diverse mechanisms of action. This screen identified the M. tuberculosis cell envelope to be a major determinant of antibiotic susceptibility but did not yield mutants whose increase in susceptibility was due to transposon insertions in genes encoding efflux pumps. Intrinsic antibiotic resistance determinants affecting resistance to multiple antibiotics included the peptidoglycan-arabinogalactan ligase Lcp1, the mycolic acid synthase MmaA4, the protein translocase SecA2, the mannosyltransferase PimE, the cell envelope-associated protease CaeA/Hip1, and FecB, a putative iron dicitrate-binding protein. Characterization of a deletion mutant confirmed FecB to be involved in the intrinsic resistance to every antibiotic analyzed. In contrast to its predicted function, FecB was dispensable for growth in low-iron medium and instead functioned as a critical mediator of envelope integrity.

Statistical analysis of genetic interactions in Tn-Seq data.

Tn-Seq is an experimental method for probing the functions of genes through construction of complex random transposon insertion libraries and quantification of each mutant's abundance using next-generation sequencing. An important emerging application of Tn-Seq is for identifying genetic interactions, which involves comparing Tn mutant libraries generated in different genetic backgrounds (e.g. wild-type strain versus knockout strain). Several analytical methods have been proposed for analyzing Tn-Seq data to identify genetic interactions, including estimating relative fitness ratios and fitting a generalized linear model. However, these have limitations which necessitate an improved approach. We present a hierarchical Bayesian method for identifying genetic interactions through quantifying the statistical significance of changes in enrichment. The analysis involves a four-way comparison of insertion counts across datasets to identify transposon mutants that differentially affect bacterial fitness depending on genetic background. Our approach was applied to Tn-Seq libraries made in isogenic strains of Mycobacterium tuberculosis lacking three different genes of unknown function previously shown to be necessary for optimal fitness during infection. By analyzing the libraries subjected to selection in mice, we were able to distinguish several distinct classes of genetic interactions for each target gene that shed light on their functions and roles during infection.

Comprehensive Essentiality Analysis of the Mycobacterium tuberculosis Genome via Saturating Transposon Mutagenesis.

For decades, identifying the regions of a bacterial chromosome that are necessary for viability has relied on mapping integration sites in libraries of random transposon mutants to find loci that are unable to sustain insertion. To date, these studies have analyzed subsaturated libraries, necessitating the application of statistical methods to estimate the likelihood that a gap in transposon coverage is the result of biological selection and not the stochasticity of insertion. As a result, the essentiality of many genomic features, particularly small ones, could not be reliably assessed. We sought to overcome this limitation by creating a completely saturated transposon library in Mycobacterium tuberculosis In assessing the composition of this highly saturated library by deep sequencing, we discovered that a previously unknown sequence bias of the Himar1 element rendered approximately 9% of potential TA dinucleotide insertion sites less permissible for insertion. We used a hidden Markov model of essentiality that accounted for this unanticipated bias, allowing us to confidently evaluate the essentiality of features that contained as few as 2 TA sites, including open reading frames (ORF), experimentally identified noncoding RNAs, methylation sites, and promoters. In addition, several essential regions that did not correspond to known features were identified, suggesting uncharacterized functions that are necessary for growth. This work provides an authoritative catalog of essential regions of the M. tuberculosis genome and a statistical framework for applying saturating mutagenesis to other bacteria. IMPORTANCE: Sequencing of transposon-insertion mutant libraries has become a widely used tool for probing the functions of genes under various conditions. The Himar1 transposon is generally believed to insert with equal probabilities at all TA dinucleotides, and therefore its absence in a mutant library is taken to indicate biological selection against the corresponding mutant. Through sequencing of a saturated Himar1 library, we found evidence that TA dinucleotides are not equally permissive for insertion. The insertion bias was observed in multiple prokaryotes and influences the statistical interpretation of transposon insertion (TnSeq) data and characterization of essential genomic regions. Using these insights, we analyzed a fully saturated TnSeq library for M. tuberculosis, enabling us to generate a comprehensive catalog of in vitro essentiality, including ORFs smaller than those found in any previous study, small (noncoding) RNAs (sRNAs), promoters, and other genomic features.

Trehalose-6-Phosphate-Mediated Toxicity Determines Essentiality of OtsB2 in Mycobacterium tuberculosis In Vitro and in Mice.

Trehalose biosynthesis is considered an attractive target for the development of antimicrobials against fungal, helminthic and bacterial pathogens including Mycobacterium tuberculosis. The most common biosynthetic route involves trehalose-6-phosphate (T6P) synthase OtsA and T6P phosphatase OtsB that generate trehalose from ADP/UDP-glucose and glucose-6-phosphate. In order to assess the drug target potential of T6P phosphatase, we generated a conditional mutant of M. tuberculosis allowing the regulated gene silencing of the T6P phosphatase gene otsB2. We found that otsB2 is essential for growth of M. tuberculosis in vitro as well as for the acute infection phase in mice following aerosol infection. By contrast, otsB2 is not essential for the chronic infection phase in mice, highlighting the substantial remodelling of trehalose metabolism during infection by M. tuberculosis. Blocking OtsB2 resulted in the accumulation of its substrate T6P, which appears to be toxic, leading to the self-poisoning of cells. Accordingly, blocking T6P production in a ΔotsA mutant abrogated otsB2 essentiality. T6P accumulation elicited a global upregulation of more than 800 genes, which might result from an increase in RNA stability implied by the enhanced neutralization of toxins exhibiting ribonuclease activity. Surprisingly, overlap with the stress response caused by the accumulation of another toxic sugar phosphate molecule, maltose-1-phosphate, was minimal. A genome-wide screen for synthetic lethal interactions with otsA identified numerous genes, revealing additional potential drug targets synergistic with OtsB2 suitable for combination therapies that would minimize the emergence of resistance to OtsB2 inhibitors.

Normalization of transposon-mutant library sequencing datasets to improve identification of conditionally essential genes.

Sequencing of transposon-mutant libraries using next-generation sequencing (TnSeq) has become a popular method for determining which genes and non-coding regions are essential for growth under various conditions in bacteria. For methods that rely on quantitative comparison of counts of reads at transposon insertion sites, proper normalization of TnSeq datasets is vitally important. Real TnSeq datasets are often noisy and exhibit a significant skew that can be dominated by high counts at a small number of sites (often for non-biological reasons). If two datasets that are not appropriately normalized are compared, it might cause the artifactual appearance of Differentially Essential (DE) genes in a statistical test, constituting type I errors (false positives). In this paper, we propose a novel method for normalization of TnSeq datasets that corrects for the skew of read-count distributions by fitting them to a Beta-Geometric distribution. We show that this read-count correction procedure reduces the number of false positives when comparing replicate datasets grown under the same conditions (for which no genuine differences in essentiality are expected). We compare these results to results obtained with other normalization procedures, and show that it results in greater reduction in the number of false positives. In addition we investigate the effects of normalization on the detection of DE genes.

TRANSIT--A Software Tool for Himar1 TnSeq Analysis.

TnSeq has become a popular technique for determining the essentiality of genomic regions in bacterial organisms. Several methods have been developed to analyze the wealth of data that has been obtained through TnSeq experiments. We developed a tool for analyzing Himar1 TnSeq data called TRANSIT. TRANSIT provides a graphical interface to three different statistical methods for analyzing TnSeq data. These methods cover a variety of approaches capable of identifying essential genes in individual datasets as well as comparative analysis between conditions. We demonstrate the utility of this software by analyzing TnSeq datasets of M. tuberculosis grown on glycerol and cholesterol. We show that TRANSIT can be used to discover genes which have been previously implicated for growth on these carbon sources. TRANSIT is written in Python, and thus can be run on Windows, OSX and Linux platforms. The source code is distributed under the GNU GPL v3 license and can be obtained from the following GitHub repository: https://github.com/mad-lab/transit.

Capturing Uncertainty by Modeling Local Transposon Insertion Frequencies Improves Discrimination of Essential Genes.

Transposon mutagenesis experiments enable the identification of essential genes in bacteria. Deep-sequencing of mutant libraries provides a large amount of high-resolution data on essentiality. Statistical methods developed to analyze this data have traditionally assumed that the probability of observing a transposon insertion is the same across the genome. This assumption, however, is inconsistent with the observed insertion frequencies from transposon mutant libraries of M. tuberculosis. We propose a modified Binomial model of essentiality that can characterize the insertion probability of individual genes in which we allow local variation in the background insertion frequency in different non-essential regions of the genome. Using the Metropolis-Hastings algorithm, samples of the posterior insertion probabilities were obtained for each gene, and the probability of each gene being essential is estimated. We compared our predictions to those of previous methods and show that, by taking into consideration local insertion frequencies, our method is capable of making more conservative predictions that better match what is experimentally known about essential and non-essential genes.

A Hidden Markov Model for identifying essential and growth-defect regions in bacterial genomes from transposon insertion sequencing data.

BACKGROUND: Knowledge of which genes are essential to the survival of an organism is critical to understanding the function of genes, and for the identification of potential drug targets for antimicrobial treatment. Previous statistical methods for assessing essentiality based on sequencing of tranposon libraries have usually limited their assessment to strict 'essential' or 'non-essential' categories. However, this binary view of essentiality does not accurately represent the more nuanced ways the growth of an organism might be affected by the disruption of its genes. In addition, these methods often limit their analysis to open-reading frames. We propose a novel method for analyzing sequence data from transposon mutant libraries using a Hidden Markov Model (HMM), along with formulas to adapt the parameters of the model to different datasets for robustness. This approach allows for the clustering of insertion sites into distinct regions of essentiality across the entire genome in a statistically rigorous manner, while also allowing for the detection of growth-defect and growth-advantage regions. RESULTS: We evaluate the performance of a 4-state HMM on a sequence dataset of M. tuberculosis transposon mutants. We also test the HMM on several synthetic datasets representing different levels of transposon insertion density and sequence coverage. We show that the HMM produces results that are highly correlated with previous assignments of essentiality for this organism. We also show that it detects growth-defect and growth-advantage genes previously shown to impair or enhance growth when disrupted. CONCLUSIONS: A 4-state HMM provides an improved way of analyzing Tn-seq data and assessing different levels of essentiality that enables not only the characterization of essential and non-essential genes, but also genes whose disruption leads to impairment (or enhancement) of growth.