Prediction of antimicrobial resistance based on whole-genome sequencing and machine learning.

MOTIVATION: Antimicrobial resistance (AMR) is one of the biggest global problems threatening human and animal health. Rapid and accurate AMR diagnostic methods are thus very urgently needed. However, traditional antimicrobial susceptibility testing (AST) is time-consuming, low throughput and viable only for cultivable bacteria. Machine learning methods may pave the way for automated AMR prediction based on genomic data of the bacteria. However, comparing different machine learning methods for the prediction of AMR based on different encodings and whole-genome sequencing data without previously known knowledge remains to be done. RESULTS: In this study, we evaluated logistic regression (LR), support vector machine (SVM), random forest (RF) and convolutional neural network (CNN) for the prediction of AMR for the antibiotics ciprofloxacin, cefotaxime, ceftazidime and gentamicin. We could demonstrate that these models can effectively predict AMR with label encoding, one-hot encoding and frequency matrix chaos game representation (FCGR encoding) on whole-genome sequencing data. We trained these models on a large AMR dataset and evaluated them on an independent public dataset. Generally, RFs and CNNs perform better than LR and SVM with AUCs up to 0.96. Furthermore, we were able to identify mutations that are associated with AMR for each antibiotic. AVAILABILITY AND IMPLEMENTATION: Source code in data preparation and model training are provided at GitHub website (https://github.com/YunxiaoRen/ML-iAMR). SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Genomic plasticity and adaptive capacity of the quaternary alkyl-ammonium compound and copper tolerant Acinetobacter bohemicus strain QAC-21b isolated from pig manure.

Here, we present the genomic characterization of an Acinetobacter bohemicus strain QAC-21b which was isolated in the presence of a quaternary alky-ammonium compound (QAAC) from manure of a conventional German pig farm. The genetic determinants for QAAC, heavy metal and antibiotic resistances are reported based of the whole genome shotgun sequence and physiological growth tests. A. bohemicus QAC-21b grew in a species typical manner well at environmental temperatures but not at 37 °C. The strain showed tolerance to QAACs and copper but was susceptible to antibiotics relevant for Acinetobacter treatments. The genome of QAC-21b contained several Acinetobacter typical QAAC and heavy metal transporting efflux pumps coding genes, but no key genes for acquired antimicrobial resistances. The high genomic content of transferable genetic elements indicates that this bacterium can be involved in the transmission of antimicrobial resistances, if it is released with manure as organic fertilizer on agricultural fields. The genetic content of the strain was compared to that of two other A. bohemicus strains, the type strain ANC 3994T, isolated from forest soil, and KCTC 42081, originally described as A. pakistanensis, a metal resistant strain isolated from a wastewater treatment pond. In contrast to the forest soil strain, both strains from anthropogenically impacted sources showed genetic features indicating their evolutionary adaptation to the anthropogenically impacted environments. Strain QAC-21b will be used as model strain to study the transmission of antimicrobial resistance to environmentally adapted Acinetobacter in agricultural environments receiving high content of pollutants with organic fertilizers from livestock husbandry.

Pseudocitrobacter corydidari sp. nov., isolated from the Asian emerald cockroach Corydidarum magnifica.

A Gram-negative bacterial strain, G163CMT, was isolated from the gut of the Asian emerald cockroach Corydidarum magnifica. The 16S rRNA gene sequence (1416 bp) of strain G163CMT showed 99.22% similarity to Pseudocitrobacter faecalis CCM 8479T and Pseudocitrobacter vendiensis CPO20170097T. The average nucleotide identity, digital DNA-DNA hybridization and average amino acid identity values of strain G163CMT were 92.4, 48.8 and 95.7% to P. faecalis CCM 8479T, and 93.3, 52.4 and 95.7% to P. vendiensis CPO20170097T. This strongly supports the designation of G163CMT as representing a new species in the genus Pseudocitrobacter. Phylogenetic trees based on the alignment of 16S rRNA, multilocus sequence analysis of six single-copy genes (fusA, pyrG, leuS, rpoB, recN and mnmE) and 107 core protein sequences consistently showed G163CMT to be a member of the genus Pseudocitrobacter, closely related to P. vendiensis CPO20170097T. In contrast to P. faecalis CCM 8479T and P. vendiensis CPO20170097T, the genome of G163CMT did not encode for proteins conferring resistance to antibiotics. However, all three genomes encoded a similar number of virulence factors and specialized metabolite biosynthetic proteins. The major fatty acids of strain G163CMT were C16:0 (31.5 %), C18:1 ω7c (22.6 %), C17:0 cyclo (15.3 %) and C14:0 (6.5 %). Based on the polyphasic results, we conclude that strain G163CMT represents a novel species of the genus Pseudocitrobacter and we propose the name Pseudocitrobacter corydidari sp. nov. with the type strain G163CMT (=DSM 112648T=CCM 9160T).

ASA3P: An automatic and scalable pipeline for the assembly, annotation and higher-level analysis of closely related bacterial isolates.

Whole genome sequencing of bacteria has become daily routine in many fields. Advances in DNA sequencing technologies and continuously dropping costs have resulted in a tremendous increase in the amounts of available sequence data. However, comprehensive in-depth analysis of the resulting data remains an arduous and time-consuming task. In order to keep pace with these promising but challenging developments and to transform raw data into valuable information, standardized analyses and scalable software tools are needed. Here, we introduce ASA3P, a fully automatic, locally executable and scalable assembly, annotation and analysis pipeline for bacterial genomes. The pipeline automatically executes necessary data processing steps, i.e. quality clipping and assembly of raw sequencing reads, scaffolding of contigs and annotation of the resulting genome sequences. Furthermore, ASA3P conducts comprehensive genome characterizations and analyses, e.g. taxonomic classification, detection of antibiotic resistance genes and identification of virulence factors. All results are presented via an HTML5 user interface providing aggregated information, interactive visualizations and access to intermediate results in standard bioinformatics file formats. We distribute ASA3P in two versions: a locally executable Docker container for small-to-medium-scale projects and an OpenStack based cloud computing version able to automatically create and manage self-scaling compute clusters. Thus, automatic and standardized analysis of hundreds of bacterial genomes becomes feasible within hours. The software and further information is available at: asap.computational.bio.

Acinetobacter stercoris sp. nov. isolated from output source of a mesophilic german biogas plant with anaerobic operating conditions.

The Gram-stain-negative, oxidase negative, catalase positive strain KPC-SM-21T, isolated from a digestate of a storage tank of a mesophilic German biogas plant, was investigated by a polyphasic taxonomic approach. Phylogenetic identification based on the nearly full-length 16S rRNA gene revealed highest gene sequence similarity to Acinetobacter baumannii ATCC 19606T (97.0%). Phylogenetic trees calculated based on partial rpoB and gyrB gene sequences showed a distinct clustering of strain KPC-SM-21T with Acinetobacter gerneri DSM 14967T = CIP 107464T and not with A. baumannii, which was also supported in the five housekeeping genes multilocus sequence analysis based phylogeny. Average nucleotide identity values between whole genome sequences of strain KPC-SM-21T and next related type strains supported the novel species status. The DNA G + C content of strain KPC-SM-21T was 37.7 mol%. Whole-cell MALDI-TOF MS analysis supported the distinctness of the strain to type strains of next related Acinetobacter species. Predominant fatty acids were C18:1 ω9c (44.2%), C16:0 (21.7%) and a summed feature comprising C16:1 ω7c and/or iso-C15:0 2-OH (15.3%). Based on the obtained genotypic, phenotypic and chemotaxonomic data we concluded that strain KPC-SM-21T represents a novel species of the genus Acinetobacter, for which the name Acinetobacter stercoris sp. nov. is proposed. The type strain is KPC-SM-21T (= DSM 102168T = LMG 29413T).

ReadXplorer 2-detailed read mapping analysis and visualization from one single source.

MOTIVATION: The vast amount of already available and currently generated read mapping data requires comprehensive visualization, and should benefit from bioinformatics tools offering a wide spectrum of analysis functionality from just one source. Appropriate handling of multiple mapped reads during mapping analyses remains an issue that demands improvement. RESULTS: The capabilities of the read mapping analysis and visualization tool ReadXplorer were vastly enhanced. Here, we present an even finer granulated read mapping classification, improving the level of detail for analyses and visualizations. The spectrum of automatic analysis functions has been broadened to include genome rearrangement detection as well as correlation analysis between two mapping data sets. Existing functions were refined and enhanced, namely the computation of differentially expressed genes, the read count and normalization analysis and the transcription start site detection. Additionally, ReadXplorer 2 features a highly improved support for large eukaryotic data sets and a command line version, enabling its integration into workflows. Finally, the new version is now able to display any kind of tabular results from other bioinformatics tools. AVAILABILITY AND IMPLEMENTATION: http://www.readxplorer.org CONTACT: readxplorer@computational.bio.uni-giessen.deSupplementary information: Supplementary data are available at Bioinformatics online.

Bacterial genome sequences of uncharacterized Chitinophaga species isolated from the International Space Station.

We report four Chitinophaga sp. strains isolated from wastewater collected onboard the International Space Station. Here, we present three finished and one draft genome. Taxonomic ranks established by genome-based analysis indicate that these Chitinophaga sp. strains represent candidates for a new species.

Draft genome sequence of Galbibacter sp. PAP.153, isolated from a marine sponge in Papua, Indonesia.

Galbibacter sp. PAP.153 was isolated from a marine sponge. Here, we report its 4.12 Mbp draft genome sequence and rate its specialized metabolite production capacity with specific focus on the chemotaxonomic marker flexirubin.

Draft genome sequences of 21 Pedobacter strains isolated from amphibian specimens.

The genomes of 21 Pedobacter strains isolated from the European salamander Salamandra salamandra and different Madagascan frog species were sequenced using Illumina sequencing. Here, we report their draft genome sequences (~4.7-7.2 Mbp in size) to allow comparative genomics and taxonomic assignment of these strains.

Split k-mer analysis compared to cgMLST and SNP-based core genome analysis for detecting transmission of vancomycin-resistant enterococci: results from routine outbreak analyses across different hospitals and hospitals networks in Berlin, Germany.

The increase of Vancomycin-resistant Enterococcus faecium (VREfm) in recent years has been partially attributed to the rise of specific clonal lineages, which have been identified throughout Germany. To date, there is no gold standard for the interpretation of genomic data for outbreak analyses. New genomic approaches such as split k-mer analysis (SKA) could support cluster attribution for routine outbreak investigation. The aim of this project was to investigate frequent clonal lineages of VREfm identified during suspected outbreaks across different hospitals, and to compare genomic approaches including SKA in routine outbreak investigation. We used routine outbreak laboratory data from seven hospitals and three different hospital networks in Berlin, Germany. Short-read libraries were sequenced on the Illumina MiSeq system. We determined clusters using the published Enterococcus faecium-cgMLST scheme (threshold ≤20 alleles), and assigned sequence and complex types (ST, CT), using the Ridom SeqSphere+ software. For each cluster as determined by cgMLST, we used pairwise core-genome SNP-analysis and SKA at thresholds of ten and seven SNPs, respectively, to further distinguish cgMLST clusters. In order to investigate clinical relevance, we analysed to what extent epidemiological linkage backed the clusters determined with different genomic approaches. Between 2014 and 2021, we sequenced 693 VREfm strains, and 644 (93 %) were associated within cgMLST clusters. More than 74 % (n=475) of the strains belonged to the six largest cgMLST clusters, comprising ST117, ST78 and ST80. All six clusters were detected across several years and hospitals without apparent epidemiological links. Core SNP analysis identified 44 clusters with a median cluster size of three isolates (IQR 2-7, min-max 2-63), as well as 197 singletons (41.4 % of 475 isolates). SKA identified 67 clusters with a median cluster size of two isolates (IQR 2-4, min-max 2-19), and 261 singletons (54.9 % of 475 isolates). Of the isolate pairs attributed to clusters, 7 % (n=3064/45 596) of pairs in clusters determined by standard cgMLST, 15 % (n=1222/8500) of pairs in core SNP-clusters and 51 % (n=942/1880) of pairs in SKA-clusters showed epidemiological linkage. The proportion of epidemiological linkage differed between sequence types. For VREfm, the discriminative ability of the widely used cgMLST based approach at ≤20 alleles difference was insufficient to rule out hospital outbreaks without further analytical methods. Cluster assignment guided by core genome SNP analysis and the reference free SKA was more discriminative and correlated better with obvious epidemiological linkage, at least recently published thresholds (ten and seven SNPs, respectively) and for frequent STs. Besides higher overall discriminative power, the whole-genome approach implemented in SKA is also easier and faster to conduct and requires less computational resources.

Identification of staphyloxanthin and derivates in yellow-pigmented Staphylococcus capitis subsp. capitis.

Introduction: Staphylococcus capitis naturally colonizes the human skin but as an opportunistic pathogen, it can also cause biofilm-associated infections and bloodstream infections in newborns. Previously, we found that two strains from the subspecies S. capitis subsp. capitis produce yellow carotenoids despite the initial species description, reporting this subspecies as non-pigmented. In Staphylococcus aureus, the golden pigment staphyloxanthin is an important virulence factor, protecting cells against reactive oxygen species and modulating membrane fluidity. Methods: In this study, we used two pigmented (DSM 111179 and DSM 113836) and two non-pigmented S. capitis subsp. capitis strains (DSM 20326T and DSM 31028) to identify the pigment, determine conditions under which pigment-production occurs and investigate whether pigmented strains show increased resistance to ROS and temperature stress. Results: We found that the non-pigmented strains remained colorless regardless of the type of medium, whereas intensity of pigmentation in the two pigmented strains increased under low nutrient conditions and with longer incubation times. We were able to detect and identify staphyloxanthin and its derivates in the two pigmented strains but found that methanol cell extracts from all four strains showed ROS scavenging activity regardless of staphyloxanthin production. Increased survival to cold temperatures (-20°C) was detected in the two pigmented strains only after long-term storage compared to the non-pigmented strains. Conclusion: The identification of staphyloxanthin in S. capitis is of clinical relevance and could be used, in the same way as in S. aureus, as a possible target for anti-virulence drug design.

Draft Genome Sequence of Sinomicrobium sp. Strain PAP.21, Isolated from a Coast Sample of Papua, Indonesia.

Sinomicrobium sp. strain PAP.21 (EXT111902) was isolated from the coast of Cenderawasih Bay National Park in West Papua, Indonesia. Its genome was assembled into 151 contigs with a total size of 5.439 Mbp, enabling the prediction of its specialized metabolite production capacity.

Draft Genome Sequences of Algoriphagus sp. Strain PAP.12 and Roseivirga sp. Strain PAP.19, Isolated from Marine Samples from Papua, Indonesia.

Algoriphagus sp. strain PAP.12 (EXT111900) and Roseivirga sp. strain PAP.19 (EXT111901) were isolated from marine samples. Here, we report their draft genome sequences, 5.032 Mbp and 4.583 Mbp in size, respectively, and rate their specialized metabolite production capacity. Taxonomic ranks established by genome-based analysis indicate that Algoriphagus sp. strain PAP.12 represents a candidate new species.

Hospital sanitary facilities on wards with high antibiotic exposure play an important role in maintaining a reservoir of resistant pathogens, even over many years.

BACKGROUND: Hospitals with their high antimicrobial selection pressure represent the presumably most important reservoir of multidrug-resistant human pathogens. Antibiotics administered in the course of treatment are excreted and discharged into the wastewater system. Not only in patients, but also in the sewers, antimicrobial substances exert selection pressure on existing bacteria and promote the emergence and dissemination of multidrug-resistant clones. In previous studies, two main clusters were identified in all sections of the hospital wastewater network that was investigated, one K. pneumoniae ST147 cluster encoding NDM- and OXA-48 carbapenemases and one VIM-encoding P. aeruginosa ST823 cluster. In the current study, we investigated if NDM- and OXA-48-encoding K. pneumoniae and VIM-encoding P. aeruginosa isolates recovered between 2014 and 2021 from oncological patients belonged to those same clusters. METHODS: The 32 isolates were re-cultured, whole-genome sequenced, phenotypically tested for their antimicrobial susceptibility, and analyzed for clonality and resistance genes in silico. RESULTS: Among these strains, 25 belonged to the two clusters that had been predominant in the wastewater, while two others belonged to a sequence-type less prominently detected in the drains of the patient rooms. CONCLUSION: Patients constantly exposed to antibiotics can, in interaction with their persistently antibiotic-exposed sanitary facilities, form a niche that might be supportive for the emergence, the development, the dissemination, and the maintenance of certain nosocomial pathogen populations in the hospital, due to antibiotic-induced selection pressure. Technical and infection control solutions might help preventing transmission of microorganisms from the wastewater system to the patient and vice versa, particularly concerning the shower and toilet drainage. However, a major driving force might also be antibiotic induced selection pressure and parallel antimicrobial stewardship efforts could be essential.

Deep Transfer Learning Enables Robust Prediction of Antimicrobial Resistance for Novel Antibiotics.

Antimicrobial resistance (AMR) has become one of the serious global health problems, threatening the effective treatment of a growing number of infections. Machine learning and deep learning show great potential in rapid and accurate AMR predictions. However, a large number of samples for the training of these models is essential. In particular, for novel antibiotics, limited training samples and data imbalance hinder the models' generalization performance and overall accuracy. We propose a deep transfer learning model that can improve model performance for AMR prediction on small, imbalanced datasets. As our approach relies on transfer learning and secondary mutations, it is also applicable to novel antibiotics and emerging resistances in the future and enables quick diagnostics and personalized treatments.

Multi-label classification for multi-drug resistance prediction of Escherichia coli.

Antimicrobial resistance (AMR) is a global health and development threat. In particular, multi-drug resistance (MDR) is increasingly common in pathogenic bacteria. It has become a serious problem to public health, as MDR can lead to the failure of treatment of patients. MDR is typically the result of mutations and the accumulation of multiple resistance genes within a single cell. Machine learning methods have a wide range of applications for AMR prediction. However, these approaches typically focus on single drug resistance prediction and do not incorporate information on accumulating antimicrobial resistance traits over time. Thus, identifying multi-drug resistance simultaneously and rapidly remains an open challenge. In our study, we could demonstrate that multi-label classification (MLC) methods can be used to model multi-drug resistance in pathogens. Importantly, we found the ensemble of classifier chains (ECC) model achieves accurate MDR prediction and outperforms other MLC methods. Thus, our study extends the available tools for MDR prediction and paves the way for improving diagnostics of infections in patients. Furthermore, the MLC methods we introduced here would contribute to reducing the threat of antimicrobial resistance and related deaths in the future by improving the speed and accuracy of the identification of pathogens and resistance.

Phenotypic and genomic assessment of the potential threat of human spaceflight-relevant Staphylococcus capitis isolates under stress conditions.

Previous studies have reported that spaceflight specific conditions such as microgravity lead to changes in bacterial physiology and resistance behavior including increased expression of virulence factors, enhanced biofilm formation and decreased susceptibility to antibiotics. To assess if spaceflight induced physiological changes can manifest in human-associated bacteria, we compared three spaceflight relevant Staphylococcus capitis isolates (DSM 111179, ISS; DSM 31028, clean room; DSM 113836; artificial gravity bedrest study) with the type strain (DSM 20326T). We tested the three strains regarding growth, colony morphology, metabolism, fatty acid and polar lipid pattern, biofilm formation, susceptibility to antibiotics and survival in different stress conditions such as treatment with hydrogen peroxide, exposure to desiccation, and irradiation with X-rays and UV-C. Moreover, we sequenced, assembled, and analyzed the genomes of all four strains. Potential genetic determinants for phenotypic differences were investigated by comparative genomics. We found that all four strains show similar metabolic patterns and the same susceptibility to antibiotics. All four strains were considered resistant to fosfomycin. Physiological differences were mainly observed compared to the type strain and minor differences among the other three strains. The ISS isolate and the bedrest study isolate exhibit a strong delayed yellow pigmentation, which is absent in the other two strains. Pigments were extracted and analyzed by UV/Vis spectroscopy showing characteristic carotenoid spectra. The ISS isolate showed the highest growth rate as well as weighted average melting temperature (WAMT) of fatty acids (41.8°C) of all strains. The clean room isolate showed strongest biofilm formation and a high tolerance to desiccation. In general, all strains survived desiccation better in absence of oxygen. There were no differences among the strains regarding radiation tolerance. Phenotypic and genomic differences among the strains observed in this study are not inevitably indicating an increased virulence of the spaceflight isolate. However, the increased growth rate, higher WAMT and colony pigmentation of the spaceflight isolate are relevant phenotypes that require further research within the human spaceflight context. We conclude that combining genetic analysis with classical microbiological methods allows the detailed assessment of the potential threat of bacteria in highly regulated and extreme environments such as spaceflight environments.

Bakta: rapid and standardized annotation of bacterial genomes via alignment-free sequence identification.

Command-line annotation software tools have continuously gained popularity compared to centralized online services due to the worldwide increase of sequenced bacterial genomes. However, results of existing command-line software pipelines heavily depend on taxon-specific databases or sufficiently well annotated reference genomes. Here, we introduce Bakta, a new command-line software tool for the robust, taxon-independent, thorough and, nonetheless, fast annotation of bacterial genomes. Bakta conducts a comprehensive annotation workflow including the detection of small proteins taking into account replicon metadata. The annotation of coding sequences is accelerated via an alignment-free sequence identification approach that in addition facilitates the precise assignment of public database cross-references. Annotation results are exported in GFF3 and International Nucleotide Sequence Database Collaboration (INSDC)-compliant flat files, as well as comprehensive JSON files, facilitating automated downstream analysis. We compared Bakta to other rapid contemporary command-line annotation software tools in both targeted and taxonomically broad benchmarks including isolates and metagenomic-assembled genomes. We demonstrated that Bakta outperforms other tools in terms of functional annotations, the assignment of functional categories and database cross-references, whilst providing comparable wall-clock runtimes. Bakta is implemented in Python 3 and runs on MacOS and Linux systems. It is freely available under a GPLv3 license at https://github.com/oschwengers/bakta. An accompanying web version is available at https://bakta.computational.bio.

Platon: identification and characterization of bacterial plasmid contigs in short-read draft assemblies exploiting protein sequence-based replicon distribution scores.

Plasmids are extrachromosomal genetic elements that replicate independently of the chromosome and play a vital role in the environmental adaptation of bacteria. Due to potential mobilization or conjugation capabilities, plasmids are important genetic vehicles for antimicrobial resistance genes and virulence factors with huge and increasing clinical implications. They are therefore subject to large genomic studies within the scientific community worldwide. As a result of rapidly improving next-generation sequencing methods, the quantity of sequenced bacterial genomes is constantly increasing, in turn raising the need for specialized tools to (i) extract plasmid sequences from draft assemblies, (ii) derive their origin and distribution, and (iii) further investigate their genetic repertoire. Recently, several bioinformatic methods and tools have emerged to tackle this issue; however, a combination of high sensitivity and specificity in plasmid sequence identification is rarely achieved in a taxon-independent manner. In addition, many software tools are not appropriate for large high-throughput analyses or cannot be included in existing software pipelines due to their technical design or software implementation. In this study, we investigated differences in the replicon distributions of protein-coding genes on a large scale as a new approach to distinguish plasmid-borne from chromosome-borne contigs. We defined and computed statistical discrimination thresholds for a new metric: the replicon distribution score (RDS), which achieved an accuracy of 96.6 %. The final performance was further improved by the combination of the RDS metric with heuristics exploiting several plasmid-specific higher-level contig characterizations. We implemented this workflow in a new high-throughput taxon-independent bioinformatics software tool called Platon for the recruitment and characterization of plasmid-borne contigs from short-read draft assemblies. Compared to PlasFlow, Platon achieved a higher accuracy (97.5 %) and more balanced predictions (F1=82.6 %) tested on a broad range of bacterial taxa and better or equal performance against the targeted tools PlasmidFinder and PlaScope on sequenced Escherichia coli isolates. Platon is available at: http://platon.computational.bio/.