MOTIVATION:The primary mechanism for transcription termination in bacteria is intrinsic terminators. These terminators influence transcript stability and play key roles in gene regulation. Existing computational methods for genome-wide terminator identification have been designed and evaluated based on a small number of experimentally evinced terminators often from only one or two organisms. RESULTS:We present TerminatorNet, a system for identifying intrinsic transcription terminators throughout bacteria. TerminatorNet uses a neural network model trained on a large set of experimentally characterized transcription terminators from a variety of bacterial genomes. TerminatorNet identifies 98% of terminators and has a false positive rate of 3%, substantially better than existing approaches. TerminatorNet commonly identifies terminators at the ends of operons. We applied TerminatorNet to thousands of genomes across the taxonomic spectrum of prokaryotes, creating a repository of tens of millions of terminators. We observe heavy use of intrinsic termination in some groups, such as Bacillota, and rare use in other groups such as archaea. We also observe a wealth of instances of DNA uptake signal sequences, important components of transformation specificity for some competent bacteria, in terminators identified in Neisseriaceae and Pasteurellaceae. AVAILABILITY AND IMPLEMENTATION:TerminatorNet and its repository of identifications are available for use via a webserver: https://cs.wellesley.edu/∼btjaden/TermNet. The source code is available at GitHub https://github.com/btjaden/TerminatorNet and Zenodo https://doi.org/10.5281/zenodo.18406126.

SUMMARY:The most challenging prokaryotic genes to identify often correspond to short ORFs (sORFs) encoding small proteins or to noncoding RNAs. RNA-seq experiments commonly evince small transcripts that do not correspond to annotated genes and are candidates for novel coding sORFs or small regulatory RNAs, but it can be difficult to accurately assess whether the numerous small transcripts are coding or not. We present Popcorn (PrOkaryotic Prediction of Coding OR Noncoding), a novel machine learning method for determining whether prokaryotic sequences are coding or noncoding. We find that Popcorn is effective in distinguishing coding from noncoding sequences, including coding sORFs and noncoding RNAs. AVAILABILITY AND IMPLEMENTATION:Freely available for use on the web at https://cs.wellesley.edu/∼btjaden/Popcorn. Source code available at https://github.com/btjaden/Popcorn and https://doi.org/10.5281/zenodo.15120075.

Small regulatory RNAs (sRNAs) are widespread in bacteria. However, characterizing the targets of sRNA regulation in a way that scales with the increasing number of identified sRNAs has proven challenging. Computational methods offer one means for efficient characterization of sRNA targets, but the sensitivity and precision of such computational methods is limited. Here, we investigate whether publicly available expression data from RNA-seq experiments can improve the accuracy of computational prediction of sRNA regulatory targets. Using compendia of 2143 Escherichia coli RNA-seq samples and 177 Salmonella RNA-seq samples, we identify groups of co-expressed genes in each organism and incorporate this expression information into computational prediction of sRNA targets based on machine learning methods. We find that integrating expression information significantly improves the accuracy of computational results. Further, we observe that computational methods perform better when trained on smaller, higher quality sets of targets rather than on larger, noisier sets of targets identified by high-throughput methods.

How organisms evolve under extreme environmental changes is a critical question in the face of global climate change. Genetic accommodation is an evolutionary process by which natural selection acts on novel phenotypes generated through repeated encounters with extreme environments. In this study, polyphenic and monophenic strains of the black mutant tobacco hornworm, Manduca sexta, were evolved via genetic accommodation of heat stress-induced phenotypes, and the molecular differences between the two strains were explored. Transcriptomic analyses showed that epigenetic and hormonal differences underlie the differences between the two strains and their distinct responses to temperature. DNA methylation had diverged between the two strains potentially mediating genetic assimilation. Juvenile hormone (JH) signaling in the polyphenic strain was temperature sensitive, whereas in the monophenic strain, JH signaling remained low at all temperatures. Although 20-hydroxyecdysone titers were elevated under heat shock conditions in both strains, the strains did not differ in the titers. Tyrosine hydroxylase was also found to differ between the two strains at different temperatures, and its expression could be modulated by topical application of a JH analog. Finally, heat shock of unselected black mutants demonstrated that the expression of the JH-response gene, Krüppel-homolog 1 (Kr-h1), increased within the first 30 min of heat shock, suggesting that JH levels respond readily to thermal stress. Our study highlights the critical role that hormones and epigenetics play during genetic accommodation and potentially in the evolution of populations in the face of climate change.

The democratization of machine learning is a popular and growing movement. In a world with a wealth of publicly available data, it is important that algorithms for analysis of data are accessible and usable by everyone. We present MLpronto, a system for machine learning analysis that is designed to be easy to use so as to facilitate engagement with machine learning algorithms. With its web interface, MLpronto requires no computer programming or machine learning background, and it normally returns results in a matter of seconds. As input, MLpronto takes a file of data to be analyzed. MLpronto then executes some of the more commonly used supervised machine learning algorithms on the data and reports the results of the analyses. As part of its execution, MLpronto generates computer programming code corresponding to its machine learning analysis, which it also supplies as output. Thus, MLpronto can be used as a no-code solution for citizen data scientists with no machine learning or programming background, as an educational tool for those learning about machine learning, and as a first step for those who prefer to engage with programming code in order to facilitate rapid development of machine learning projects. MLpronto is freely available for use at https://mlpronto.org/.

Ligand-binding RNAs (RNA aptamers) are widespread in the three domains of life, serving as sensors of metabolites and other small molecules. When aptamers are embedded within RNA transcripts as components of riboswitches, they can regulate gene expression upon binding their ligands. Previous methods for biochemical validation of computationally predicted aptamers are not well-suited for rapid screening of large numbers of RNA aptamers. Therefore, we utilized DRaCALA (Differential Radial Capillary Action of Ligand Assay), a technique designed originally to study protein-ligand interactions, to examine RNA-ligand binding, permitting rapid screening of dozens of RNA aptamer candidates concurrently. Using this method, which we call RNA-DRaCALA, we screened 30 ykkC family subtype 2a RNA aptamers that were computationally predicted to bind (p)ppGpp. Most of the aptamers bound both ppGpp and pppGpp, but some strongly favored only ppGpp or pppGpp, and some bound neither. Expansion of the number of biochemically verified sites allowed construction of more accurate secondary structure models and prediction of key features in the aptamers that distinguish a ppGpp from a pppGpp binding site. To demonstrate that the method works with other ligands, we also used RNA DRaCALA to analyze aptamer binding by thiamine pyrophosphate.

Small regulatory RNAs pervade prokaryotes, with the best-studied family of these non-coding genes corresponding to trans-acting regulators that bind via base pairing to their message targets. Given the increasing frequency with which these genes are being identified, it is important that methods for illuminating their regulatory targets keep pace. Using a machine learning approach, we investigate thousands of interactions between small RNAs and their targets, and we interrogate more than a hundred features indicative of these interactions. We present a new method, TargetRNA3, for predicting targets of small RNA regulators and show that it outperforms existing approaches. TargetRNA3 is available at https://cs.wellesley.edu/~btjaden/TargetRNA3 .

Owing to the complexities of bacterial RNA biology, the transcriptomes of even the best studied bacteria are not fully understood. To help elucidate the transcriptional landscape of E. coli, we compiled a compendium of 3,376 RNA-seq data sets composed of more than 7 trillion sequenced bases, which we evaluate with a transcript assembly pipeline. We report expression profiles for all annotated E. coli genes as well as 5,071 other transcripts. Additionally, we observe hundreds of instances of co-transcribed genes that are novel with respect to existing operon databases. By integrating data from a large number of sequencing experiments corresponding to a wide range of conditions, we are able to obtain a comprehensive view of the E. coli transcriptome.

An operon is a set of neighboring genes in a genome that is transcribed as a single polycistronic message. Genes that are part of the same operon often have related functional roles or participate in the same metabolic pathways. The majority of all bacterial genes are co-transcribed with one or more other genes as part of a multi-gene operon. Thus, accurate identification of operons is important in understanding co-regulation of genes and their functional relationships. Here, we present a computational system that uses RNA-seq data to determine operons throughout a genome. The system takes the name of a genome and one or more files of RNA-seq data as input. Our method combines primary genomic sequence information with expression data from the RNA-seq files in a unified probabilistic model in order to identify operons. We assess our method's ability to accurately identify operons in a range of species through comparison to external databases of operons, both experimentally confirmed and computationally predicted, and through focused experiments that confirm new operons identified by our method. Our system is freely available at https://cs.wellesley.edu/~btjaden/Rockhopper/.

Recent emergence of antimicrobial resistance of Neisseria gonorrhoeae worldwide has resulted in limited therapeutic choices for treatment of infections caused by this organism. We performed global transcriptomic analysis of N. gonorrhoeae in subjects with gonorrhea who attended a Nanjing, China, sexually transmitted infection (STI) clinic, where antimicrobial resistance of N. gonorrhoeae is high and increasing. We found that N. gonorrhoeae transcriptional responses to infection differed in genital specimens taken from men and women, particularly antibiotic resistance gene expression, which was increased in men. These sex-specific findings may provide a new approach to guide therapeutic interventions and preventive measures that are also sex specific while providing additional insight to address antimicrobial resistance of N. gonorrhoeae .

One key to the success of Mycobacterium tuberculosis as a pathogen is its ability to reside in the hostile environment of the human macrophage. Bacteria adapt to stress through a variety of mechanisms, including the use of small regulatory RNAs (sRNAs), which posttranscriptionally regulate bacterial gene expression. However, very little is currently known about mycobacterial sRNA-mediated riboregulation. To date, mycobacterial sRNA discovery has been performed primarily in log-phase growth, and no direct interaction between any mycobacterial sRNA and its targets has been validated. Here, we performed large-scale sRNA discovery and expression profiling in M. tuberculosis during exposure to five pathogenically relevant stresses. From these data, we identified a subset of sRNAs that are highly induced in multiple stress conditions. We focused on one of these sRNAs, ncRv11846, here renamed mycobacterial regulatory sRNA in iron (MrsI). We characterized the regulon of MrsI and showed in mycobacteria that it regulates one of its targets, bfrA, through a direct binding interaction. MrsI mediates an iron-sparing response that is required for optimal survival of M. tuberculosis under iron-limiting conditions. However, MrsI is induced by multiple host-like stressors, which appear to trigger MrsI as part of an anticipatory response to impending iron deprivation in the macrophage environment.

Isolates of a given bacterial pathogen often display phenotypic variation, and this can negatively impact public health, for example, by reducing the efficacy of preventative measures. Here, we identify that the human pathogen group A Streptococcus (GAS; Streptococcus pyogenes) expresses pili on its cell surface in a serotype-specific manner. Specifically, we show that serotype M3 GAS isolates, which are nonrandomly associated with causing particularly severe and lethal invasive infections, produce negligible amounts of pili relative to serotype M1 and M49 isolates. Performance of an interserotype transcriptome comparison (serotype M1 versus serotype M3) was instrumental in this discovery. We also identified that the transcriptional regulator Nra positively regulates pilus expression in M3 GAS isolates and that the low level of pilus expression of these isolates correlates with a low level of nra transcription. Finally, we discovered that the phenotypic consequences of low levels of pilus expression by M3 GAS isolates are a reduced ability to adhere to host cells and an increased ability to survive and proliferate in human blood. We propose that an enhanced ability to survive in human blood, in part due to reduced pilus expression, is a contributing factor in the association of serotype M3 isolates with highly invasive infections. In conclusion, our data show that GAS isolates express pili in a serotype-dependent manner and may inform vaccine development, given that pilus proteins are being discussed as possible GAS vaccine antigens.

During environmental adaptation bacteria use small regulatory RNAs (sRNAs) to repress or activate expression of a large fraction of their proteome. We extended the use of the in vivo RNA proximity ligation method toward probing global sRNA interactions with their targets in Pseudomonas aeruginosa and verified the method with a known regulon controlled by the PrrF1 sRNA. We also identified two sRNAs (Sr0161 and ErsA) that interact with the mRNA encoding the major porin OprD responsible for the uptake of carbapenem antibiotics. These two sRNAs base pair with the 5 UTR of oprD leading to increase in resistance of the bacteria to meropenem. Additional proximity ligation experiments and enrichment for Sr0161 targets identified the mRNA for the regulator of type III secretion system. Interaction between the exsA mRNA and Sr0161 leads to a block in the synthesis of a component of the T3SS apparatus and an effector. Another sRNA, Sr006, positively regulates, without Hfq, the expression of PagL, an enzyme responsible for deacylation of lipid A, reducing its pro-inflammatory property and resulting in polymyxin resistance. Therefore, an analysis of global sRNA-mRNA interactions can lead to discoveries of novel pathways controlling gene expression that are likely integrated into larger regulatory networks.

The first step in the post-transcriptional regulatory function of most bacterial small non-coding RNAs (sRNAs) is base pairing with partially complementary sequences of targeted transcripts. We present a simple method for identifying sRNA targets in vivo and defining processing sites of the regulated transcripts. The technique, referred to as global small non-coding RNA target identification by ligation and sequencing (GRIL-seq), is based on preferential ligation of sRNAs to the ends of base-paired targets in bacteria co-expressing T4 RNA ligase, followed by sequencing to identify the chimaeras. In addition to the RNA chaperone Hfq, the GRIL-seq method depends on the activity of the pyrophosphorylase RppH. Using PrrF1, an iron-regulated sRNA in Pseudomonas aeruginosa , we demonstrated that direct regulatory targets of this sRNA can readily be identified. Therefore, GRIL-seq represents a powerful tool not only for identifying direct targets of sRNAs in a variety of environments, but also for uncovering novel roles for sRNAs and their targets in complex regulatory networks.

Transcriptome assays are increasingly being performed by high-throughput RNA sequencing (RNA-seq). For organisms whose genomes have not been sequenced and annotated, transcriptomes must be assembled de novo from the RNA-seq data. Here, we present novel algorithms, specific to bacterial gene structures and transcriptomes, for analysis of bacterial RNA-seq data and de novo transcriptome assembly. The algorithms are implemented in an open source software system called Rockhopper 2. We find that Rockhopper 2 outperforms other de novo transcriptome assemblers and offers accurate and efficient analysis of bacterial RNA-seq data. Rockhopper 2 is available at http://cs.wellesley.edu/~btjaden/Rockhopper .

Gonorrhea is a highly prevalent disease resulting in significant morbidity worldwide, with an estimated 106 cases reported annually. Neisseria gonorrhoeae, the causative agent of gonorrhea, colonizes and infects the human genital tract and often evades host immune mechanisms until successful antibiotic treatment is used. The alarming increase in antibiotic-resistant strains of N. gonorrhoeae, the often asymptomatic nature of this disease in women and the lack of a vaccine directed at crucial virulence determinants have prompted us to perform transcriptome analysis to understand gonococcal gene expression patterns during natural infection. We sequenced RNA extracted from cervico-vaginal lavage samples collected from women recently exposed to infected male partners and determined the complete N. gonorrhoeae transcriptome during infection of the lower genital tract in women. On average, 3.19% of total RNA isolated from female samples aligned to the N. gonorrhoeae NCCP11945 genome and 1750 gonococcal ORFs (65% of all protein-coding genes) were transcribed. High expression in vivo was observed in genes encoding antimicrobial efflux pumps, iron response, phage production, pilin structure, outer membrane structures and hypothetical proteins. A parallel analysis was performed using the same strains grown in vitro in a chemically defined media (CDM). A total of 140 genes were increased in expression during natural infection compared to growth in CDM, and 165 genes were decreased in expression. Large differences were found in gene expression profiles under each condition, particularly with genes involved in DNA and RNA processing, iron, transposase, pilin and lipoproteins. We specifically interrogated genes encoding DNA binding regulators and iron-scavenging proteins, and identified increased expression of several iron-regulated genes, including tbpAB and fbpAB, during infection in women as compared to growth in vitro, suggesting that during infection of the genital tract in women, the gonococcus is exposed to an iron deplete environment. Collectively, we demonstrate that a large portion of the gonococcal genome is expressed and regulated during mucosal infection including genes involved in regulatory functions and iron scavenging.

In the last several years, bacterial gene regulation via small RNAs (sRNAs) has been recognized as an important mechanism controlling expression of essential proteins that are critical to bacterial growth and metabolism. Technologies such as RNA-seq are rapidly expanding the field of sRNAs and are enabling a global view of the "sRNAome" of several bacterial species. While numerous sRNAs have been identified in a variety of both Gram-negative and Gram-positive bacteria, only a very small number have been fully characterized in the human pathogen Neisseria gonorrhoeae, the etiological agent of the STD gonorrhea. Here we present the first analysis of N. gonorrhoeae specifically focused on the identification of sRNAs through RNA-seq analysis of the organism cultured under different in vitro growth conditions. Using a new computational program, Rockhopper, to analyze prokaryotic RNA-seq data obtained from N. gonorrhoeae we identified several putative sRNAs and confirmed their expression and size through Northern blot analysis. In addition, RNA was collected from four different growth conditions (iron replete and deplete, as well as with and without co-culture with human endocervical cells). Many of the putative sRNAs identified shoed varying expression levels relative to the different growth conditions examine or were detected only under certain conditions but not others. Comparisons of identified sRNAs with the regulatory pattern of putative mRNA targets revealed possible functional roles for these sRNAs. These studies are the first to carry out a global analysis of N. gonorrhoeae specifically focused on sRNAs and show that RNA-mediated regulation may be an important mechanism of gene control in this human pathogen.

Many small, noncoding RNAs (sRNAs) in bacteria act as posttranscriptional regulators of messenger RNAs. TargetRNA2 is a web server that identifies mRNA targets of sRNA regulatory action in bacteria. As input, TargetRNA2 takes the sequence of an sRNA and the name of a sequenced bacterial replicon. When searching for targets of RNA regulation, TargetRNA2 uses a variety of features, including conservation of the sRNA in other bacteria, the secondary structure of the sRNA, the secondary structure of each candidate mRNA target and the hybridization energy between the sRNA and each candidate mRNA target. TargetRNA2 outputs a ranked list of likely regulatory targets for the input sRNA. When evaluated on a comprehensive set of sRNA-target interactions, TargetRNA2 was found to be both accurate and efficient in identifying targets of sRNA regulatory action. Furthermore, TargetRNA2 has the ability to integrate RNA-seq data, if available. If an sRNA is differentially expressed in two or more RNA-seq experiments, TargetRNA2 considers co-differential gene expression when searching for regulatory targets, significantly improving the accuracy of target identifications. The TargetRNA2 web server is freely available for use at http://cs.wellesley.edu/∼btjaden/TargetRNA2.

Recent advances in high-throughput RNA sequencing (RNA-seq) have enabled tremendous leaps forward in our understanding of bacterial transcriptomes. However, computational methods for analysis of bacterial transcriptome data have not kept pace with the large and growing data sets generated by RNA-seq technology. Here, we present new algorithms, specific to bacterial gene structures and transcriptomes, for analysis of RNA-seq data. The algorithms are implemented in an open source software system called Rockhopper that supports various stages of bacterial RNA-seq data analysis, including aligning sequencing reads to a genome, constructing transcriptome maps, quantifying transcript abundance, testing for differential gene expression, determining operon structures and visualizing results. We demonstrate the performance of Rockhopper using 2.1 billion sequenced reads from 75 RNA-seq experiments conducted with Escherichia coli, Neisseria gonorrhoeae, Salmonella enterica, Streptococcus pyogenes and Xenorhabdus nematophila. We find that the transcriptome maps generated by our algorithms are highly accurate when compared with focused experimental data from E. coli and N. gonorrhoeae, and we validate our system’s ability to identify novel small RNAs, operons and transcription start sites. Our results suggest that Rockhopper can be used for efficient and accurate analysis of bacterial RNA-seq data, and that it can aid with elucidation of bacterial transcriptomes.