Next Article in Journal
Assessment of Prevalence and Diversity of Antimicrobial Resistant Escherichia coli from Retail Meats in Southern California
Next Article in Special Issue
Clinical Equivalence between Generic Versus Branded Antibiotics: Systematic Review and Meta-Analysis
Previous Article in Journal
Antimicrobial Activity of Apis mellifera Bee Venom Collected in Northern Peru
Previous Article in Special Issue
Usefulness of Routine Laboratory Tests for Follow up of Patients Receiving Outpatient Parenteral Antimicrobial Therapy Run by Infectious Diseases Fellows
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Review

Clinical Diagnostics of Bacterial Infections and Their Resistance to Antibiotics—Current State and Whole Genome Sequencing Implementation Perspectives

1
Department of Biotechnology, Inland Norway University of Applied Sciences, Holsetgata, 222317 Hamar, Norway
2
Institute of Clinical Medicine, Faculty of Health Science, UiT The Arctic University of Norway, Hansine Hansens veg, 189019 Tromsø, Norway
*
Author to whom correspondence should be addressed.
These authors contributed equally to this work.
Antibiotics 2023, 12(4), 781; https://doi.org/10.3390/antibiotics12040781
Submission received: 9 February 2023 / Revised: 19 March 2023 / Accepted: 13 April 2023 / Published: 19 April 2023
(This article belongs to the Special Issue Diagnostics, Treatment and Prevention of Bacterial Infections)

Abstract

:
Antimicrobial resistance (AMR), defined as the ability of microorganisms to withstand antimicrobial treatment, is responsible for millions of deaths annually. The rapid spread of AMR across continents warrants systematic changes in healthcare routines and protocols. One of the fundamental issues with AMR spread is the lack of rapid diagnostic tools for pathogen identification and AMR detection. Resistance profile identification often depends on pathogen culturing and thus may last up to several days. This contributes to the misuse of antibiotics for viral infection, the use of inappropriate antibiotics, the overuse of broad-spectrum antibiotics, or delayed infection treatment. Current DNA sequencing technologies offer the potential to develop rapid infection and AMR diagnostic tools that can provide information in a few hours rather than days. However, these techniques commonly require advanced bioinformatics knowledge and, at present, are not suited for routine lab use. In this review, we give an overview of the AMR burden on healthcare, describe current pathogen identification and AMR screening methods, and provide perspectives on how DNA sequencing may be used for rapid diagnostics. Additionally, we discuss the common steps used for DNA data analysis, currently available pipelines, and tools for analysis. Direct, culture-independent sequencing has the potential to complement current culture-based methods in routine clinical settings. However, there is a need for a minimum set of standards in terms of evaluating the results generated. Additionally, we discuss the use of machine learning algorithms regarding pathogen phenotype detection (resistance/susceptibility to an antibiotic).

1. Introduction

Antibiotics are natural or synthetic chemical compounds that inhibit bacterial growth and are the first-line drugs against bacterial infection. Bacteria have rapidly developed and spread antibiotic resistance across different bacterial classes in response to unnecessary antibiotic use, flaws in proper diagnostic tools, poor hygiene conditions, and suboptimal healthcare practices. During resistance development, antibiotics gradually become less effective, and bacteria can adapt and continue to grow in their presence [1].
Antibiotic resistance arises mainly via four different mechanisms: reducing efflux transport, target modification, limiting drug uptake, and enzyme-catalyzed inactivation [2,3]. Efflux pumps are a large family of protein pumps in bacteria that emit antibiotics from inside the cell to the outside [4]. Bacteria also develop resistance toward a specific class of antibiotics upon a series of DNA mutations or producing specific enzymes, resulting in modification of the targets of that class of antibiotic [5]. Alternatively, some proteins can bind to antibiotics or their targets, reducing antibiotic uptake [6]. Bacteria also inactivate antibiotics by producing enzymes that recognize and destroy antibiotics’ structure [7]. It has been shown that bacteria develop resistance via post-translational mechanisms as well [8]. These resistance mechanisms can be categorized as intrinsic or expected resistance (found across all strains/bacteria) or acquired (first appearing only in a few strains and then spreading to microorganisms of distant taxonomical relatedness). Therefore, acquired resistance poses a greater risk to human and animal health.
Antibiotic resistance, although having existed before antibiotics were discovered [9], is strongly driven via antibiotic use in human and veterinary medicine. Since the 1960s, antibiotic resistance has dramatically increased, fast becoming a global public health concern [10,11]. A recent meta-analysis of resistant bacteria burden on human health and well-being revealed that in 2019 alone, a striking 1.27 M deaths were caused directly by antibiotic-resistant bacteria (ARBs), and 4.95 M deaths were associated with ARBs. This number has surpassed the human immunodeficiency virus (HIV) and malaria [11]. Given the elevated use of antibiotics in 2020 due to the coronavirus disease (COVID-19) pandemic and low AMR resilience in many countries, this number has potentially increased in recent years. Recently published data indicate that the mortality rate increased in COVID patients who received antibiotics [12,13]. A recent study based on metagenomic sequencing of 757 sewage samples from 243 cities in 101 countries suggested that certain geographies are more prone to transmission events and should receive additional attention [14]. An estimated death rate of more than 10 million people annually by 2050 may be realistic due to antibiotic resistance [15,16,17]. From a healthcare perspective, antibiotic resistance causes more extended hospitalization of patients suffering from an infection. This eventually imposes a tremendous additional cost for the hospitals and could cost an extra USD 300 billion to more than USD 1 trillion annually by 2050 globally [18].
Furthermore, antibiotic resistance can indirectly affect the success rate of other procedures, such as surgeries and chemotherapy [19,20]. From an economic perspective, antibiotic resistance has multiple effects on the global economy. For instance, it has been reported that antibiotic resistance has led to a shortage in protein resources (sick animals not responding to antibiotic treatment and/or antibiotic residues in meat/milk precluding their consumption) [21], a pronounced increase in poverty and inequity [22], deficiency in productivity due to sickness [23], and significantly decreased global trade [24].

2. Current Methods for the Clinical Diagnosis of Bacterial Infection

2.1. Pathogen Identification

2.1.1. Culture-Dependent Techniques

In these methods, culturable bacteria are enriched using a non-specific or selective medium and then characterized based on their morphology and metabolic traits. [25] Biochemical tests are based on qualitative biochemical reactions, e.g., visible changes in media because of bacterial metabolic activity. In colony morphology, microorganisms are identified based on the morphological properties of a specific strain [25]. The conventional approaches benefit from simplicity, reproducibility, and low cost. However, one should remember that the enrichment of microorganisms is a labor-intensive and time-consuming procedure, and colonies often require 1–3 days and sometimes longer to grow [26]. Besides that, correct pathogen identification based on culture often requires vast expertise, which might lead to inaccurate diagnosis [27]. To minimize such inaccuracy in biochemical methods, analytical profile index (API) tests [28] and automated systems such as VITEK (bioMerieux, Inc., Marcy-l’Étoile, France) have been developed [29].
The principle of immunological techniques is based on the interaction between diagnostic antibodies and certain antigenic elements of bacteria. Among different immunological assays, the enzyme-linked immunosorbent assay (ELISA) is the most frequently used technique in microbiology [30]. The ELISA technique became popular for its ability to identify multiple pathogens simultaneously, high-throughput capacity, low cost, and the opportunity to gauge the invading pathogens. Nonetheless, immunological techniques suffer from their dependency on the number of antibodies against pathogens epitopes (in the case of indirect ELISA). Furthermore, as some antigens are shared between bacterial species, the antigen–antibody binding is not specific. Therefore, immunological techniques lack selectivity and sensitivity.

2.1.2. Mass Spectroscopy

Several mass spectroscopy (MS) techniques have been developed in recent years. Matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-ToF MS) is the most common MS technique to identify bacterial species and even strains [31]. In this technique, the prepared bacteria are exposed to short laser pulses, and consequently, their proteins become ionized. The ions travel through a vacuum tube and are registered by the sensor, thus creating a spectrum. The resulting spectrum is then compared with a database of known bacteria spectra. A more in-depth methodology of MALDI-ToF is described by Singhal et al. [32]. Although microbial identification using MALDI-ToF MS is fast and accurate, the protein spectra for closely related species, e.g., Escherichia coli and Shigella, are highly similar [33], which hampers correct identification.
Furthermore, initial and maintenance costs are the major concerns [32]. It has been estimated that a MALDI-ToF MS system requires EUR 200,000 per year for maintenance, limiting its usage to large hospitals capable of covering these costs [34]. Furthermore, the limited application of the MALDI-ToF MS system to the early detection of antibiotic resistance [35] makes this technique less attractive for the AMR field. However, a recent study demonstrated improved prediction of antibiotic resistance based on MALDI-ToF data by implementing a machine learning algorithm [36].

2.1.3. Nucleic-Acid-Based Techniques

Nucleic-acid-based techniques, such as the polymerase chain reaction (PCR), reverse transcriptase PCR (RT-PCR), and transcription-mediated amplification (TMA), are used in microbial diagnostic laboratories. The main feature of such techniques is their ability to amplify specific regions of the pathogen genome (conserved genes or 16 S ribosomal RNA genes), commonly used as taxonomic markers [37,38]. Although nucleic-acid-based techniques are highly sensitive and accurate, bacterial infections are, in some cases, caused by multiple bacteria, which might cause species–species or strain–strain interactions in this type of assay [39]. The PCR also requires primer design, which limits flexibility as one needs to have prior knowledge of the bacterial species present in the clinical samples. Generally, PCR techniques are culture-independent; however, a low-grade infection may require pathogen enrichment before analysis [40]. To overcome such limitations, some syndromic panels (taking benefits from the multiplexed real-time PCR) have recently been developed and certified for diagnostics, which allows users to quickly identify the most common pathogens [41,42]. The main disadvantages of these panels are the limited number of pathogens and AMR resistance profiles covered by the currently available kits [41,42].

2.2. Antibiotic Susceptibility Profiling

2.2.1. Phenotypic Techniques

The broth dilution test is one of the initial methods in antibiotic resistance screening. Here, a series of antibiotic dilutions within a liquid culture medium are prepared, and the bacteria of interest are inoculated. Then, the tubes are incubated overnight at 37 °C, and bacterial growth is detected. The lowest concentration of antibiotics that inhibit growth is considered the minimal inhibitory concentration (MIC) [43]. The method is highly reproducible, and the MIC value is highly informative. However, users might face time limitations and a lack of flexibility regarding antibiotic choice [44].
With the antimicrobial gradient method, bacteria are cultured on an agar medium and co-incubated with a few plastic test strips, each containing a gradient concentration of a specific antibiotic. The point where the lower segment of the ellipse-shaped growth inhibition zone touches the test strip would be considered the minimum inhibitory concentration (MIC) value [45]. The method is simple and easy to run and requires no specific expertise; however, like the broth dilution test, this method is limited to the number of antibiotics and could be costly if one tends to test multiple antibiotics [46,47].
Disc diffusion tests have been commonly used for years in many diagnostic laboratories. The single concentration of antibiotics impregnated into small round paper is placed on the agar surface with cultured bacteria. After 16–24 h incubation at 37 °C, the zone diameter with no bacterial growth close to each antibiotic disc is measured and compared with the reference values [43]. This method is simple and cheap to run, while to date, no automated system has been developed for fast screening of the no-growth zone on the plates.
The main drawback of phenotypic methods is that they take up to a few days due to culturing. However, the new culture-based technology, Accelerate PhenoSystem (APS, Accelerate Diagnostics, Inc., Tucson, AZ, USA), allows pathogen ID and MIC values to be obtained within 2 and 7 h, respectively. The system uses the fluorescence in situ hybridization (FISH) technique for identifying common pathogens in bloodstream infections and morphokinetic cellular analysis for AST [43,45,48]. Another recent technology that allows same-day pathogen ID and AST is based on acoustic-enhanced flow cytometry [49]. When used for infection identification in peritoneal-dialysis-associated peritonitis, this technology provided an infection confirmation within 1 h and AST within 3–6 h after sample reception.

2.2.2. Molecular Techniques

The PCR and the qPCR (in some cases) are the most valuable techniques for quantifying and detecting resistance genes in bacteria in clinical laboratories [50]. New PCR variants such as digital droplet PCR (ddPCR) offered even higher sensitivity than qPCR for SARS-CoV-2 detection [51]. Another study suggested a hybrid combination of the PCR and the isothermal PCR as a highly sensitive method for SARS-CoV-2 diagnostics [52]. The advantage of the qPCR over the PCR is the quantitative measurement of the targeted gene(s). The main weakness of these methods is that one needs to decide about the targeted gene(s) in advance. Therefore, both assays have a limited capacity to detect unexpected/unanticipated genes.
Furthermore, the variants of resistance genes might be difficult to detect. Another critical point in the field of antibiotic resistance is the discrepancy between the nucleic acid amplification test and AST results. Culture-dependent AST results are considered the gold standard for resistance detection. The Clinical and Laboratory Standards Institute (CLSI) has provided specific guidelines for resolving such disagreements in Gram-positive organisms and ESBL and carbapenemase in Gram-negative organisms [53,54].

3. The Potential of Whole Genome Sequencing (WGS) for Use in Clinical Routine

3.1. DNA Sequencing Technologies

Sequencing technology has seen tremendous development during the last few decades. The human genome project was one of the main reasons for the rapid establishment of high-throughput WGS as a valuable tool for studying an organism’s genome. The first generation of sequencing was based on chain-terminating inhibitors proposed by Sanger et al. [55]. In Sanger sequencing, DNA fragments are PCR amplified, and then denaturized PCR products are further used as templates. The new DNA strand, complementary to the denaturized PCR product, is synthesized using chain-terminating fluorescence-labeled di deoxynucleotides triphosphates (ddNTPs). Whenever DNA polymerase incorporates a ddNTP, the extension terminates. Then, these newly synthesized strands are loaded into capillary electrophoresis. The time of their passage through the capillary depends on the size of the fragments—the shorter the fragment, the faster it passes. Upon exit from the capillary, a laser excites fluorescent tags of the ddNTPs in each fragment, recording the emitted light. This method was previously widely used for microbial species identification [56]. Sanger sequencing does not allow high throughput and is time- and resource-demanding.
The limitations of Sanger sequencing led to the development of the next (second)-generation sequencing (NGS) methods at the beginning of 2005 [57]. The main difference between Sanger sequencing and NGS is that NGS allows the sequencing of millions of fragments simultaneously. One of the most popular NGS technologies is Illumina sequencing (Illumina Ltd., San Diego, CA, USA). The high-throughput nature of NGS decreased the cost and time of pathogen identification significantly and has the potential to be considered as a routine diagnostic method [58]. For instance, NGS contributed significantly to diagnosing more than 150 genetic diseases [59]. Although NGS was an evolutionary step in sequencing technology, it also suffers from specific limitations, such as high initial costs and short read lengths (maximum 300 bp). Furthermore, complex bioinformatic tools and techniques are required to analyze data after sequencing.
Third-generation sequencing (TGS) emerged in early 2011 with the introduction of the PacBio system [57]. TGS can produce long reads with an average length of 10 kb and a maximum length of 80 kb [60,61], covering the repeating regions in the genome, which makes the assembly and characterization of large structural changes in the genome less challenging [57]. However, the application of the PacBio system in clinical microbiology is limited by its initial costs, portability, and the complexity of data analyses and software setup [62]. Another commonly used example of TGS was introduced by Oxford Nanopore Technology (ONT). Nanopore sequencing is becoming a popular sequencing method for several reasons.
First, considering the portability of some of its variants (MinION and Flongle), real-time sequencing, and data analyses, this technology greatly reduces the time required from sample preparation to results. Moreover, nanopore sequencing allows direct DNA sequencing to omit the PCR amplification step (depending on the library preparation kit); therefore, the sequencing data have a more consistent genome coverage [60]. Nanopore technology comprises nano-sensors and specific “channel“ types through which single-stranded DNA penetrates during the sequencing run. Passaging the DNA through the pore results in electric current changes sensed by reader proteins. The current change is base specific, and all alterations in the electrical signal can be translated into nucleotide bases. Despite nanopore sequencing’s significant usefulness, one of its main drawbacks is higher sequencing error compared to Illumina, which is now gradually improved by new chemistry, kits, and more accurate base calling algorithms [63,64]. In addition, the bioinformatic toolkits for nanopore sequencing data analysis need to be better developed compared to tools designed for NGS data analyses [65].

3.2. WGS Application in Clinical Settings

The currently implemented methods for pathogen identification and antibiotic resistance detection either require a long time (culturing-based techniques) or can only detect predefined targets (PCR-based techniques and phenotypic assays). Regarding culture-based methods, there is a risk of overlooking fastidious pathogens [66]. Moreover, testing AMR for antibiotic susceptibility testing (AST) can further take 18–24 h [67]. All these factors can potentially be eliminated using untargeted whole genome sequencing (WGS), which has become more affordable (although still expensive compared to conventional methods) and, importantly, portable. Although WGS cannot be widely implemented in routine clinical labs at the time of publication, there is a growing body of research on the potential of WGS in clinical routine infection profiling, especially in severe cases (f.ex. sepsis) where rapid and proper antibiotic treatment initiation is a matter of life or death. In addition, direct sequencing methods can also detect dead bacteria when antibiotic therapy has been administered before sampling [68].
WGS commonly refers to the sequencing genome of a single organism. Biological samples are rarely mono-colonized, however. When whole genomes of several species are sequenced simultaneously, it is referred to as metagenome sequencing (meta-WGS). Although not routine at the time of publication, meta-WGS paves its way to being widely used in several clinical applications. Illumina sequencing has a 99–99.9% average accuracy rate and is thus often regarded as the gold standard of sequencing [69]. ONT platforms, on the other hand, despite having higher error rates, are more suitable for clinical application since they output sequencing information in real-time. Real-time ONT metagenomic sequencing (mONS), for example, has been successfully implemented in lower respiratory tract infections (sputum), urinary tract infections (urine), central nervous system (cerebral spinal fluid), sepsis (blood), surgical site infections, and orthopedic devices with up to 100% sensitivity and specificity in pathogen detection (Table 1). These studies demonstrate that both bacterial identification and antibiotic resistance genes detection were reached within hours of sample acquisition. Considering that conventional bacterial identification and antibiotic susceptibility testing may take up to a week, mONS clearly offers a significant leap toward real-time diagnostics in the future.
To date, the use of meta-WGS in routine clinical diagnostics is limited by several factors, including complex workflows, still slow turnaround times, and relatively high costs. Moreover, there is the need for extensive data analysis, which requires both computational infrastructure and bioinformatics expertise. With sequencing prices constantly decreasing, data mining becomes a bottleneck for using meta-WGS in clinical routine.

3.3. WGS Data Analysis

Several steps should be taken to obtain reliable information on pathogen identity and genotypes from raw sequencing data (Figure 1A). Although the analysis flow is relatively independent of the platform used to generate the data, the software used differs. Here, we will discuss these steps regarding Illumina and ONT data analysis.

3.3.1. Base-Calling and Demultiplexing

Both base-calling and demultiplexing are commonly performed automatically right after sequencing. Many different base callers and demultiplexing algorithms have been developed for the Illumina platform [83,84]. Compared with Illumina, fewer algorithms have been developed for the ONT platform. The MinKNOW software used to drive ONT sequencing has an integrated deep neural network base caller known as Guppy [85]. Using ONT, the user might choose between base calling on the flow after the sequencing is finished or saving raw data and base calling later using more powerful computational resources. It is also possible to choose between three options of base-calling in MinKNOW: fast but least accurate (85–92% median read accuracy), highly accurate, or super accurate base calling with a median read accuracy of 90–96% [85]. Demultiplexing is also commonly performed automatically, and the user only needs to provide a list of barcodes or barcode IDs used during the library preparation.

3.3.2. Adapter Removal and Quality Filtering

Adapter removal and quality filtering are essential to achieve a high-quality genome assembly. Adapters can be removed automatically during the base calling and demultiplexing step. However, it is advisable to perform both removal of the remaining adapters and a quality check before proceeding with the downstream analysis. FastQC for short reads [86] or LongQC [87] and pycoQC [88] for long ONT reads scan raw sequencing data, report per base, per sequence quality statistics, and by searching for overrepresented sequences (i.e., adapters). Commonly used tools for Illumina adapter removal and quality filtering are Trimmomatic [89] or TrimGalore [90], which wraps cutadapt [90] for adapter removal, and FastQC [86] for quality filtering. For ONT sequencing, adapter removal can be performed using Guppy or porechop [91], although the latter has been discontinued and is no longer developed for new library prep versions.

3.3.3. Filtering of Human Reads

This step is essential for routine clinic lab use to protect patients’ identities and privacy. When pathogens are initially isolated from the clinical sample before sequencing, this step can be omitted since no patient-identifiable data will be sequenced along with the pathogen. If, however, a clinical specimen is sequenced as is (for example, the sample directly or from the culture of pus/blood/urine, etc.), removing all human DNA from the dataset is of utmost importance before the analysis starts. Many algorithms can be used for that matter (bbmap, bowtie2, and hisat2), and all of them are based on mapping query data to a reference human genome and removing those reads that map to the reference.

3.3.4. Pathogen Identification

Pathogen identification can be performed both before and after the genome assembly. Since genome assembly is a time-consuming step, we will shortly review the methods to perform assembly-free taxonomy identification. Since Illumina reads are short, only tools that split reads into small ‘words’ (kmers) and compare their mapping to prebuilt index databases can be used. Kraken [92], or its newer version Kraken2 [93], and Centrifuge [94] are commonly used to classify rapidly reads as viral, bacterial, and archaeal domains using the kmer-based approach. On the other hand, ONT reads are long, and in addition to Kraken2 or Centrifuge, a common BLAST-based search against NCBI reference databases (RefSeq/RefProk) can also be used for pathogen ID. We have shown that BLAST provides more accurate results than other tools, with negligible difference in the time taken for the prediction [67].

3.3.5. Genome Assembly and Assembly Quality Control

For Illumina data, the SPAdes [95] assembler is widely used. It contains additional options of metagenome assembly, plasmid assembly, or hybrid assembly, given that long reads are also available for the pathogen. In addition, SPAdes also include an option for additional read error correction before assembly. For ONT data, common assemblers include Canu [96], Minimap/Miniasm [97], Flye [98], and Raven [99]. The Trycycler tool enables a consensus based on results from different assemblers, thus improving the accuracy of the assembly [100]. Our recent publication showed that hybrid assembly constructed using both short and long reads considerably improved plasmid assembly, antimicrobial resistance genes, and identification of virulence factors in clinical isolates [101]. QUAST [102] is commonly used to assess the quality of the assembly by the length of the assembly, the number of contigs, and assembly contiguity. Completeness of the genome assembly can be evaluated using BUSCO [103] or CheckM [104]. GenomeQC comprises both genome assembly statistics and genome completeness assessment [105].

3.3.6. Resistome Identification and Phenotype Inference

Several databases and algorithms are available, and their choice often depends on users’ preferences. Commonly used tools for resistome search include ABRicate [106], AMRFinderPlus [107], ResFinder [108], RGI [109], and Ariba [110]. Ariba and ResFinder can be used on unassembled Illumina data, whereas RGI (CARD), ABRicate, and AMRFinderPlus require assembled contigs/long reads. ONT unassembled reads allow users to identify whether a given antibiotic gene class (such as SHV or CTX-M) is present. However, subtyping the gene is not feasible until genome assembly is performed [67,77]. ResFinder and RGI report antibiotic resistance genes (ARGs) and single mutations associated with antimicrobial resistance and predict a phenotype based on the findings. Choosing the appropriate database is challenging, and our previous publication showed that resistome analysis results could be database dependent [111].
Based on the descriptions of the steps above, the whole sequencing data analysis is challenging for lab personnel unfamiliar with bioinformatic tools. Moreover, most of these tools are command-line based, which also constrain the inclusion of WGS into routine clinical lab use. Therefore, there is an unmet need for automatic bioinformatic pipelines tailored for lab personnel with no bioinformatics and/or command-line use background. Both Illumina and ONT offer in-built user-friendly solutions for taxonomy identification and antibiotic gene search: specific workflows in EPI2ME (ONT) or BaseSpace (Illumina, San Diego, CA, USA) were designed for these tasks. Several open-source pipelines for pathogen identification and AMR surveillance are available such as Bactopia [112], TORMES [113], Nullarbor [114], and CRuMPIT [81] (Figure 1B). However, these workflows are more suited for clinical epidemiology rather than diagnostics. They can hardly be streamlined for routine use in clinics due to a lack of important steps specific to clinics. Moreover, most of them are tailored for Illumina sequence data which is impossible to use for rapid pathogen typing in clinics. To our knowledge, no ONT sequence data analysis automated pipeline, which would include both taxonomy and ARG identification, has been developed to date.

4. Machine Learning in Data Analysis for Tackling Antibiotic Resistance

WGS data were mainly used to discern isolates’ genotypes and the presence/absence of antibiotic-resistance genes [115]. In some cases, where antibiotic resistance is highly correlated with ARGs only, this approach showed high concordance between genotype and phenotype. For example, the detection of resistance genes in Campylobacter jejuni isolates had a 98% correlation to their phenotype against ciprofloxacin, nalidixic acid, erythromycin, gentamycin, streptomycin, and tetracycline [116]. However, such a high correlation is rare, and in many cases, the detection of ARGs does not associate with the resistant phenotype [117]. Bacteria may become antibiotic-resistant by changing expression levels of porins, efflux pumps, or enhancing biofilm formation [118,119,120]. These changes are impossible to account for using only acquired ARG screening. Machine learning has gained more interest in clinical applications since the rise and availability of next-generation sequencing technologies coupled with advanced and computationally powerful computers.
Machine learning (ML) is a broad set of algorithms, and statistical models used to find connections and correlations between complex data [121]. Its main characteristic that distinguishes ML from other statistical approaches is that the computer learns from the data itself, searching for the peculiarities, i.e., features, in the dataset without being explicitly programmed which features and in which combinations to look for. This approach enables researchers and programmers to elucidate connections in large datasets that are sometimes impossible to find using other conventional statistical models. ML comprises various clustering methods, regressions, decision trees, neural networks, and vector machines, etc. [122].
ML can also aid empirical antibiotic therapy when no AST data for the isolate are available. Feretzakis et al. applied several ML algorithms, including random forest classification, instant-based learning, sequential minimal optimization, and others, to make a weighted empirical decision on which antibiotic should be used based solely on the infectious agent’s Gram stain, the site of infection, and the patient demographics [123]. Although their clinical data were limited to only these three variables, they demonstrated a receiver operating characteristic (ROC) area of up to 0.72 for predicting resistant infections and potentially aiding clinicians to make a more informed decision on antibiotic prescription before time-consuming AST results are in place.
In addition, ML algorithms have been combined with biochemical screening to identify antibiotic targets to enhance drug efficacy [124]. This approach enabled the authors to detect that purine biosynthesis and adenine limitation might increase antibiotic lethality-novel mechanisms that remained undiscovered using only a conventional bioinformatics approach. In 2021, Truong et al. implemented machine learning to enable real-time industrial monitoring of resistance changes in bacterial growth. They measured near-infrared spectroscopy spectra from E. coli isolates cultured in 96-well microtiter plates with varying concentrations of tetracycline. They then used partial least squares discriminant analysis (PLS-DA) for discriminating resistant and susceptible strains.
WGS and ML technologies have also been broadly used for improved pathogen identification [125], healthcare outbreak detection [126], elucidating drug effectiveness with regards to microbiome interactions [127], the search for antibiotic mechanisms of action [124], and AST phenotype prediction [26,125,128,129,130]. There are several publications where machine learning enables the prediction of the resistance phenotype based on the genotype [130,131,132]. Some of these models require genome assemblies [128], whereas others use raw sequencing data and thus save the time needed for assembly [130]. Most models yield categorical predictions, i.e., whether the pathogen is susceptible or resistant to the given drug. The Nguyen et al., 2018 [128] models are one of the few that predict the MIC value. However, all phenotype prediction models are heavily skewed towards the dataset used for training. Typically, when these models are used to predict the phenotype of pathogens with a distinct resistance profile, they perform poorly. Due to the relatively low number of isolates with WGS and phenotypic data spanning all possible resistance profiles, the development of the ‘universal’ model seems far-fetched at the time of publication. However, there is a possibility of combining already developed models into so-called ensembles, where data from each isolate are analyzed by all models included, and an isolate is deemed resistant if any of the models predict this class. We have previously used this approach by combining Nguyen et al., 2018 models with AMR-Diag neural networks. We demonstrated that this approach increased prediction accuracy from 25% to 96% for combined data from two datasets [130]. However, most genotype-to-phenotype models have focused on strains isolated from monoculture infections (E. coli, Klebsiella pneumoniae, and Acinetobacter baumannii), and there is a need for the development of an ML-based method for AST prediction of several other antibiotic-bacteria combinations.

5. Future Perspectives

The European Committee on Antimicrobial Susceptibility Testing (EUCAST) and the European Centre for Disease Prevention and Control have indicated that WGS-based detection represents the future of AMR surveillance [133,134]. Rapid pathogen identification, AMR detection, and antibiotic susceptibility testing (AST) are in great demand due to the rise of drug-resistant bacteria. The current paradigm of culture-based AST methods suffers from a long turnaround time (4–5 days) (Figure 2) [67,135]. Genome sequencing seems promising, especially for critically ill patients and in cases where time is critical. In a clinical setting, a combination of direct sequencing from biological material (high bacterial load infections, e.g., a urinary tract infection, mastitis, or meningitis) or bacterial colonies in positive cultures (low bacterial load infection, e.g., bloodstream infection) has the potential to become a routine tool for clinical microbiology (Figure 2). Recent publications show that the use of WGS for pathogen identification and resistance detection has been more widely applied (Table 1). Evidence of their use has increased for some biological samples and bacterial classes. Current sequencing methods involve complex workflows with slow turnaround times and relatively high costs. However, in the recent past, sequencing has become more affordable and more widely applied. However, there is a need for different bioinformatics tools to analyze WGS data, which should adhere to a set of minimum standards and be standardized to be equivalent in terms of the results generated [133]. Such standardization must be implemented at different levels, including the amount of data generated specifically when it comes to third-generation sequencing using different chemistry kits, data quality assessment criteria, cut-off values for trimming and filtering in different software, BLAST cut-off values, taxonomy criteria, and coverage and identity cutoff values for AMR and plasmid detection. In addition, there is a need for reference AMR database curation. Moreover, the bioinformatic tools available today should be assembled into pipelines tailored for use with sensitive data and a user-friendly interface. This allows existing clinical microbiology lab personnel to effectively analyze sequencing data without extensive training in command-line programming.
Interestingly, recent research has demonstrated the potential of the photonics technologies (Raman spectroscopy and phase microscopy) assay as a rapid and first-stage tool for species-, resistance-, and strain-level classification, which WGS can follow up for confirmation [136,137]. This will limit the search space for the sequence alignments so that computations can be performed within seconds. Such a workflow can lead to potential future use in clinical microbiology. Furthermore, recent studies applied Raman spectroscopy to study bacterial composition in dental plaque biofilms [138] and in dry surface biofilms in a hospital environment [139].
The available evidence for using WGS as a tool to infer AST accurately is either poor or non-existent [133]. However, as highlighted by the World Health Organization (WHO) and the EUCAST, there are several major bottlenecks, such as data collection, resistance database establishment, bioinformatic analyses, and establishing a reproducible analytical pipeline [133,140]. Moreover, assessing genotypic data against phenotypic data (i.e., clinical breakpoints) represents a more formidable challenge. This will be necessary if WGS-based testing is to guide clinical decision-making. However, there is an urgent need for agreement on the appropriate and effective principles and quality control metrics to facilitate standardization and harmonization of analytical approaches and interpretative criteria for WGS-based predictive AST. More focused studies and resources are needed as a priority to improve knowledge. The ML AST tools should also be developed considering the requirements for artificial intelligence robustness set forth by the European Commission [141,142]. The developers of these tools should thoroughly assess whether the ML system meets all the standards, not only for robustness and accuracy but also vulnerability against tampering, biases caused by uneven sampling, and compliance with data confidentiality.

Author Contributions

R.A. designed the study. All authors (R.A., E.A. and A.K.) participated in collecting and reviewing the literature, discussing, and writing the manuscript. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Acknowledgments

The authors would like to thank Robert Wilson for proofreading the manuscript.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Dadgostar, P. Antimicrobial Resistance: Implications and Costs. Infect. Drug Resist. 2019, 12, 3903–3910. [Google Scholar] [CrossRef]
  2. Reygaert, W.C. An Overview of the Antimicrobial Resistance Mechanisms of Bacteria. AIMS Microbiol. 2018, 4, 482–501. [Google Scholar] [CrossRef] [PubMed]
  3. Munita, J.M.; Arias, C.A. Mechanisms of Antibiotic Resistance. Microbiol. Spectr. 2016, 4, 481–511. [Google Scholar] [CrossRef] [PubMed]
  4. Amaral, L.; Martins, A.; Spengler, G.; Molnar, J. Efflux Pumps of Gram-Negative Bacteria: What They Do, How They Do It, with What and How to Deal with Them. Front. Pharmacol. 2014, 4, 168. [Google Scholar] [CrossRef] [PubMed]
  5. Lambert, P.A. Bacterial Resistance to Antibiotics: Modified Target Sites. Adv. Drug Deliv. Rev. 2005, 57, 1471–1485. [Google Scholar] [CrossRef]
  6. Peterson, E.; Kaur, P. Antibiotic Resistance Mechanisms in Bacteria: Relationships between Resistance Determinants of Antibiotic Producers, Environmental Bacteria, and Clinical Pathogens. Front. Microbiol. 2018, 9, 2928. [Google Scholar] [CrossRef] [PubMed]
  7. Egorov, A.M.; Ulyashova, M.M.; Rubtsova, M.Y. Bacterial Enzymes and Antibiotic Resistance. Acta Nat. 2018, 10, 33–48. [Google Scholar] [CrossRef]
  8. Arora, G.; Bothra, A.; Prosser, G.; Arora, K.; Sajid, A. Role of Post-Translational Modifications in the Acquisition of Drug Resistance in Mycobacterium tuberculosis. FEBS J. 2021, 288, 3375–3393. [Google Scholar] [CrossRef] [PubMed]
  9. Allen, H.K.; Donato, J.; Wang, H.H.; Cloud-Hansen, K.A.; Davies, J.; Handelsman, J. Call of the Wild: Antibiotic Resistance Genes in Natural Environments. Nat. Rev. Microbiol. 2010, 8, 251–259. [Google Scholar] [CrossRef]
  10. Kadri, S.S. Key Takeaways from the U.S. Cdc’s 2019 Antibiotic Resistance Threats Report for Frontline Providers. Crit. Care Med. 2020, 48, 939–945. [Google Scholar] [CrossRef]
  11. Murray, C.J.L.; Ikuta, K.S.; Sharara, F.; Swetschinski, L.; Aguilar, G.R.; Gray, A.; Han, C.; Bisignano, C.; Rao, P.; Wool, E.; et al. Global Burden of Bacterial Antimicrobial Resistance in 2019: A Systematic Analysis. Lancet 2022, 399, 629–655. [Google Scholar] [CrossRef] [PubMed]
  12. Rosca, A.; Balcaen, T.; Lanoix, J.P.; Michaud, A.; Moyet, J.; Marcq, I.; Schmit, J.L.; Bloch, F.; Deschasse, G. Mortality Risk and Antibiotic Use for Covid-19 in Hospitalized Patients over 80. Biomed. Pharmacother. 2022, 146, 112481. [Google Scholar] [CrossRef] [PubMed]
  13. Sokolović, D.; Drakul, D.; Joksimović, B.; Lalović, N.; Avram, N.; Milić, M.; Nogo-Živanović, D.; Mijović, B. Consumption of Antibiotics in Primary Care Setting before and During Covid-19 Pandemic in Republic of Srpska, Bosnia and Herzegovina. Antibiotics 2022, 11, 1319. [Google Scholar] [CrossRef]
  14. Munk, P.; Brinch, C.; Moller, F.D.; Petersen, T.N.; Hendriksen, R.S.; Seyfarth, A.M.; Kjeldgaard, J.S.; Svendsen, C.A.; van Bunnik, B.; Berglund, F.; et al. Genomic Analysis of Sewage from 101 Countries Reveals Global Landscape of Antimicrobial Resistance. Nat. Commun. 2022, 13, 7251. [Google Scholar] [CrossRef]
  15. O’Neill, J. Tackling Drug-Resistant Infections Globally: Final Report and Recommendations; Government of the United Kingdom: London, UK, 2016.
  16. Pereira, B.; Kulkarni, S. Antibiotic Misuse and Improper Practices in India: Identifying the Scope to Improve through a Narrative Review. Int. J. Risk Saf. Med. 2021, 33, 357–364. [Google Scholar] [CrossRef] [PubMed]
  17. Saha, M.; Sarkar, A. Review on Multiple Facets of Drug Resistance: A Rising Challenge in the 21st Century. J. Xenobiot 2021, 11, 197–214. [Google Scholar] [CrossRef]
  18. Chokshi, A.; Sifri, Z.; Cennimo, D.; Horng, H. Global Contributors to Antibiotic Resistance. J. Glob. Infect. Dis. 2019, 11, 36–42. [Google Scholar]
  19. Anderson, M.; Clift, C.; Schulze, K.; Sagan, A.; Nahrgang, S.; Ouakrim, D.A.; Mossialos, E. European Observatory Policy Briefs. In Averting the Amr Crisis: What Are the Avenues for Policy Action for Countries in Europe? European Observatory on Health Systems and Policies: Copenhagen, Denmark, 2019. [Google Scholar]
  20. Santoro-Lopes, G.; de Gouvêa, E.F. Multidrug-Resistant Bacterial Infections after Liver Transplantation: An Ever-Growing Challenge. World J. Gastroenterol. 2014, 20, 6201–6210. [Google Scholar] [CrossRef]
  21. Van Boeckel, T.P.; Brower, C.; Gilbert, M.; Grenfell, B.T.; Levin, S.A.; Robinson, T.P.; Teillant, A.; Laxminarayan, R. Global Trends in Antimicrobial Use in Food Animals. Proc. Natl. Acad. Sci. USA 2015, 112, 5649–5654. [Google Scholar] [CrossRef] [PubMed]
  22. Jonas, O.B.; Irwin, A.; Berthe; Jean, F.C.; Gall, L.; Francois, G.; Marquez, P.V. Drug-Resistant Infections a Threat to Our Economic Future; World Bank Group: Washington, DC, USA, 2017. [Google Scholar]
  23. Naylor, N.R.; Atun, R.; Zhu, N.; Kulasabanathan, K.; Silva, S.; Chatterjee, A.; Knight, G.M.; Robotham, J.V. Estimating the Burden of Antimicrobial Resistance: A Systematic Literature Review. Antimicrob. Resist. Infect. Control. 2018, 7, 58. [Google Scholar] [CrossRef]
  24. Lekagul, A.; Tangcharoensathien, V.; Yeung, S. Patterns of Antibiotic Use in Global Pig Production: A Systematic Review. Vet. Anim. Sci. 2019, 7, 100058. [Google Scholar] [CrossRef] [PubMed]
  25. van Belkum, A.; Durand, G.; Peyret, M.; Chatellier, S.; Zambardi, G.; Schrenzel, J.; Shortridge, D.; Engelhardt, A.; Dunne, W.M., Jr. Rapid Clinical Bacteriology and Its Future Impact. Ann. Lab. Med. 2013, 33, 14–27. [Google Scholar] [CrossRef] [PubMed]
  26. Bradley, P.; Gordon, N.C.; Walker, T.M.; Dunn, L.; Heys, S.; Huang, B.; Earle, S.; Pankhurst, L.J.; Anson, L.; De Cesare, M.; et al. Rapid Antibiotic-Resistance Predictions from Genome Sequence Data for Staphylococcus Aureus and Mycobacterium Tuberculosis. Nat. Commun. 2015, 6, 10063. [Google Scholar] [CrossRef]
  27. Lievens, B.; Thomma, B.P. Recent Developments in Pathogen Detection Arrays: Implications for Fungal Plant Pathogens and Use in Practice. Phytopathology 2005, 95, 1374–1380. [Google Scholar] [CrossRef]
  28. Mohamed Salleh, N.A.B.; Tanaka, Y.; Sutarlie, L.; Su, X. Detecting Bacterial Infections in Wounds: A Review of Biosensors and Wearable Sensors in Comparison with Conventional Laboratory Methods. Analyst 2022, 147, 1756–1776. [Google Scholar] [CrossRef]
  29. Książczyk, M.; Kuczkowski, M.; Dudek, B.; Korzekwa, K.; Tobiasz, A.; Korzeniowska-Kowal, A.; Paluch, E.; Wieliczko, A.; Bugla-Płoskońska, G. Application of Routine Diagnostic Procedure, Vitek 2 Compact, Maldi-Tof Ms, and Pcr Assays in Identification Procedure of Bacterial Strain with Ambiguous Phenotype. Curr. Microbiol. 2016, 72, 570–582. [Google Scholar] [CrossRef] [PubMed]
  30. Verma, J.; Saxena, S.; Babu, S.G. Elisa-Based Identification and Detection of Microbes. In Analyzing Microbes: Manual of Molecular Biology Techniques; Springer: Berlin, Heidelberg, 2013. [Google Scholar]
  31. Torres-Sangiao, E.; Rodriguez, C.L.; García-Riestra, C. Application and Perspectives of Maldi–Tof Mass Spectrometry in Clinical Microbiology Laboratories. Microorganisms 2021, 9, 1539. [Google Scholar] [CrossRef]
  32. Singhal, N.; Kumar, M.; Kanaujia, P.K.; Virdi, J.S. Maldi-Tof Mass Spectrometry: An Emerging Technology for Microbial Identification and Diagnosis. Front. Microbiol. 2015, 6, 791. [Google Scholar] [CrossRef]
  33. Rychert, J. Benefits and Limitations of Maldi-Tof Mass Spectrometry for the Identification of Microorganisms. J. Infect. Epidemiol. 2019, 2, 1–5. [Google Scholar] [CrossRef]
  34. Wieser, A.; Schneider, L.; Jung, J.; Schubert, S. Maldi-Tof Ms in Microbiological Diagnostics-Identification of Microorganisms and Beyond (Mini Review). Appl. Microbiol. Biotechnol. 2012, 93, 965–974. [Google Scholar] [CrossRef]
  35. Florio, W.; Baldeschi, L.; Rizzato, C.; Tavanti, A.; Ghelardi, E.; Lupetti, A. Detection of Antibiotic-Resistance by Maldi-Tof Mass Spectrometry: An Expanding Area. Front. Cell. Infect. Microbiol. 2020, 10, 572909. [Google Scholar] [CrossRef] [PubMed]
  36. Weis, C.; Rieck, B.; Balzer, S.; Cuénod, A.; Egli, A.; Borgwardt, K. Improved Maldi-Tof Ms Based Antimicrobial Resistance Prediction through Hierarchical Stratification. bioRxiv 2022. [Google Scholar] [CrossRef]
  37. Johnson, J.S.; Spakowicz, D.J.; Hong, B.-Y.; Petersen, L.M.; Demkowicz, P.; Chen, L.; Leopold, S.R.; Hanson, B.M.; Agresta, H.O.; Gerstein, M.; et al. Evaluation of 16s Rrna Gene Sequencing for Species and Strain-Level Microbiome Analysis. Nat. Commun. 2019, 10, 5029. [Google Scholar] [CrossRef] [PubMed]
  38. Srinivasan, R.; Karaoz, U.; Volegova, M.; MacKichan, J.; Kato-Maeda, M.; Miller, S.; Nadarajan, R.; Brodie, E.L.; Lynch, S.V. Use of 16s Rrna Gene for Identification of a Broad Range of Clinically Relevant Bacterial Pathogens. PLoS ONE 2015, 10, e0117617. [Google Scholar] [CrossRef]
  39. Rogers, G.B.; Carroll, M.P.; Bruce, K.D. Studying Bacterial Infections through Culture-Independent Approaches. J. Med. Microbiol. 2009, 58, 1401–1418. [Google Scholar] [CrossRef]
  40. Pilecky, M.; Schildberger, A.; Orth-Höller, D.; Weber, V. Pathogen Enrichment from Human Whole Blood for the Diagnosis of Bloodstream Infection: Prospects and Limitations. Diagn. Microbiol. Infect. Dis. 2019, 94, 7–14. [Google Scholar] [CrossRef] [PubMed]
  41. Ramanan, P.; Bryson, A.L.; Binnicker, M.J.; Pritt, B.S.; Patel, R. Syndromic Panel-Based Testing in Clinical Microbiology. Clin. Microbiol. Rev. 2018, 31, e00024-17. [Google Scholar] [CrossRef]
  42. Bard, J.D.; McElvania, E. Panels and Syndromic Testing in Clinical Microbiology. Clin. Lab. Med. 2020, 40, 393–420. [Google Scholar] [CrossRef]
  43. Wiegand, I.; Hilpert, K.; Hancock, R.E. Agar and Broth Dilution Methods to Determine the Minimal Inhibitory Concentration (Mic) of Antimicrobial Substances. Nat. Protoc. 2008, 3, 163–175. [Google Scholar] [CrossRef]
  44. Marroki, A.; Bousmaha-Marroki, L. Antibiotic Resistance Diagnostic Methods for Pathogenic Bacteria. In Encyclopedia of Infection and Immunity; Rezaei, N., Ed.; Elsevier: Oxford, UK, 2022; pp. 320–341. [Google Scholar]
  45. Balouiri, M.; Sadiki, M.; Ibnsouda, S.K. Methods for in Vitro Evaluating Antimicrobial Activity: A Review. J. Pharm. Anal. 2016, 6, 71–79. [Google Scholar] [CrossRef]
  46. Gajic, I.; Kabic, J.; Kekic, D.; Jovicevic, M.; Milenkovic, M.; Culafic, D.M.; Trudic, A.; Ranin, L.; Opavski, N. Antimicrobial Susceptibility Testing: A Comprehensive Review of Currently Used Methods. Antibiotics 2022, 11, 427. [Google Scholar] [CrossRef] [PubMed]
  47. Reller, L.B.; Weinstein, M.; Jorgensen, J.H.; Ferraro, M.J. Antimicrobial Susceptibility Testing: A Review of General Principles and Contemporary Practices. Clinical. Infect. Dis. 2009, 49, 1749–1755. [Google Scholar]
  48. Chapot, V.; Effenberg, L.; Dohmen-Ruetten, J.; Buer, J.; Kehrmann, J. Evaluation of the Accelerate Pheno System for Rapid Identification and Antimicrobial Susceptibility Testing of Positive Blood Culture Bottles Inoculated with Primary Sterile Specimens from Patients with Suspected Severe Infections. J. Clin. Microbiol. 2021, 59, e02637-20. [Google Scholar] [CrossRef] [PubMed]
  49. Inglis, T.J.J.; Paton, T.F.; Kopczyk, M.K.; Mulroney, K.T.; Carson, C.F. Same-day antimicrobial susceptibility test using acoustic-enhanced flow cytometry visualized with supervised machine learning. J. Med. Microbiol. 2020, 69, 657–669. [Google Scholar] [CrossRef]
  50. Weile, J.; Knabbe, C. Current Applications and Future Trends of Molecular Diagnostics in Clinical Bacteriology. Anal. Bioanal. Chem. 2009, 394, 731–742. [Google Scholar] [CrossRef]
  51. Vasudevan, H.N.; Xu, P.; Servellita, V.; Miller, S.; Liu, L.; Gopez, A.; Chiu, C.Y.; Abate, A.R. Digital Droplet Pcr Accurately Quantifies Sars-Cov-2 Viral Load from Crude Lysate without Nucleic Acid Purification. Sci. Rep. 2021, 11, 780. [Google Scholar] [CrossRef]
  52. Varlamov, D.A.; Blagodatskikh, K.A.; Smirnova, E.V.; Kramarov, V.M.; Ignatov, K.B. Combinations of Pcr and Isothermal Amplification Techniques Are Suitable for Fast and Sensitive Detection of Sars-Cov-2 Viral Rna. Front. Bioeng. Biotechnol. 2020, 8, 604793. [Google Scholar] [CrossRef]
  53. Yee, R.; Bard, J.D.; Simner, P.J. The Genotype-to-Phenotype Dilemma: How Should Laboratories Approach Discordant Susceptibility Results? J. Clin. Microbiol. 2021, 59, e00138-20. [Google Scholar] [CrossRef]
  54. CLSI. Performance Standards for Antimicrobial Susceptibility Testing, M100. In Appendix H. Using Molecular Assays for Resistance Detection; CLSI: Wayne, PA, USA, 2020. [Google Scholar]
  55. Sanger, F.; Nicklen, S.; Coulson, A.R. DNA Sequencing with Chain-Terminating Inhibitors. Proc. Natl. Acad. Sci. USA 1977, 74, 5463–5467. [Google Scholar] [CrossRef]
  56. Ludwig, W.; Schleifer, K.H. Bacterial Phylogeny Based on 16s and 23s Rrna Sequence Analysis. FEMS Microbiol. Rev. 1994, 15, 155–173. [Google Scholar] [CrossRef]
  57. Athanasopoulou, K.; Boti, M.A.; Adamopoulos, P.G.; Skourou, P.C.; Scorilas, A. Third-Generation Sequencing: The Spearhead Towards the Radical Transformation of Modern Genomics. Life 2021, 12, 30. [Google Scholar] [CrossRef] [PubMed]
  58. Besser, J.; Carleton, H.A.; Gerner-Smidt, P.; Lindsey, R.L.; Trees, E. Next-Generation Sequencing Technologies and Their Application to the Study and Control of Bacterial Infections. Clin. Microbiol. Infect. 2018, 24, 335–341. [Google Scholar] [CrossRef] [PubMed]
  59. Rabbani, B.; Tekin, M.; Mahdieh, N. The Promise of Whole-Exome Sequencing in Medical Genetics. J. Hum. Genet. 2014, 59, 5–15. [Google Scholar] [CrossRef]
  60. Goodwin, S.; McPherson, J.D.; McCombie, W.R. Coming of Age: Ten Years of Next-Generation Sequencing Technologies. Nat. Rev. Genet. 2016, 17, 333–351. [Google Scholar] [CrossRef]
  61. van Dijk, E.L.; Jaszczyszyn, Y.; Naquin, D.; Thermes, C. The Third Revolution in Sequencing Technology. Trends Genet. 2018, 34, 666–681. [Google Scholar] [CrossRef]
  62. Hufnagel, D.E.; Hufford, M.B.; Seetharam, A.S. Sequeltools: A Suite of Tools for Working with Pacbio Sequel Raw Sequence Data. BMC Bioinform. 2020, 21, 429. [Google Scholar] [CrossRef]
  63. Wang, Y.; Zhao, Y.; Bollas, A.; Wang, Y.; Au, K.F. Nanopore Sequencing Technology, Bioinformatics and Applications. Nat. Biotechnol. 2021, 39, 1348–1365. [Google Scholar] [CrossRef]
  64. Rang, F.J.; Kloosterman, W.P.; de Ridder, J. From Squiggle to Basepair: Computational Approaches for Improving Nanopore Sequencing Read Accuracy. Genome. Biol. 2018, 19, 90. [Google Scholar] [CrossRef]
  65. Petersen, L.M.; Martin, I.W.; Moschetti, W.E.; Kershaw, C.M.; Tsongalis, G.J. Third-Generation Sequencing in the Clinical Laboratory: Exploring the Advantages and Challenges of Nanopore Sequencing. J. Clin. Microbiol. 2019, 58, e01315-19. [Google Scholar] [CrossRef]
  66. Vaca, D.J.; Dobler, G.; Fischer, S.F.; Keller, C.; Konrad, M.; von Loewenich, F.D.; Orenga, S.; Sapre, S.U.; van Belkum, A.; Kempf, V.A.J. Contemporary Diagnostics for Medically Relevant Fastidious Microorganisms Belonging to the Genera Anaplasma, Bartonella, Coxiella, Orientia and Rickettsia. FEMS Microbiol. Rev. 2022, 46, fuac013. [Google Scholar] [CrossRef]
  67. Taxt, A.M.; Avershina, E.; Frye, S.A.; Naseer, U.; Ahmad, R. Rapid Identification of Pathogens, Antibiotic Resistance Genes and Plasmids in Blood Cultures by Nanopore Sequencing. Sci. Rep. 2020, 10, 7622. [Google Scholar] [CrossRef]
  68. Bourke, T.W.; McKenna, J.P.; Coyle, P.V.; Shields, M.D.; Fairley, D.J. Diagnostic Accuracy of Loop-Mediated Isothermal Amplification as a near-Patient Test for Meningococcal Disease in Children: An Observational Cohort Study. Lancet Infect. Dis. 2015, 15, 552–558. [Google Scholar] [CrossRef]
  69. Stoler, N.; Nekrutenko, A. Sequencing Error Profiles of Illumina Sequencing Instruments. NAR Genom. Bioinform. 2021, 3, lqab019. [Google Scholar] [CrossRef]
  70. Charalampous, T.; Kay, G.L.; Richardson, H.; Aydin, A.; Baldan, R.; Jeanes, C.; Rae, D.; Grundy, S.; Turner, D.J.; Wain, J.; et al. Nanopore Metagenomics Enables Rapid Clinical Diagnosis of Bacterial Lower Respiratory Infection. Nat. Biotechnol. 2019, 37, 783–792. [Google Scholar] [CrossRef]
  71. Janes, V.A.; Matamoros, S.; Munk, P.; Clausen, P.T.L.C.; Koekkoek, S.M.; Koster, L.A.M.; Jakobs, M.E.; de Wever, B.; Visser, C.E.; Aarestrup, F.M.; et al. Metagenomic DNA Sequencing for Semi-Quantitative Pathogen Detection from Urine: A Prospective, Laboratory-Based, Proof-of-Concept Study. Lancet Microbe 2022, 3, e588–e597. [Google Scholar] [CrossRef]
  72. Zhang, L.; Huang, W.; Zhang, S.; Li, Q.; Wang, Y.; Chen, T.; Jiang, H.; Kong, D.; Lv, Q.; Zheng, Y.; et al. Rapid Detection of Bacterial Pathogens and Antimicrobial Resistance Genes in Clinical Urine Samples with Urinary Tract Infection by Metagenomic Nanopore Sequencing. Front. Microbiol. 2022, 13, 858777. [Google Scholar] [CrossRef]
  73. Morsli, M.; Bechah, Y.; Coulibaly, O.; Toro, A.; Fournier, P.E.; Houhamdi, L.; Drancourt, M. Direct Diagnosis of Pasteurella Multocida Meningitis Using Next-Generation Sequencing. Lancet Microbe 2022, 3, e6. [Google Scholar] [CrossRef]
  74. Morsli, M.; Kerharo, Q.; Amrane, S.; Parola, P.; Fournier, P.E.; Drancourt, M. Real-Time Whole Genome Sequencing Direct Diagnosis of Streptococcus pneumoniae Meningitis: A Case Report. J. Infect. 2021, 83, 709–737. [Google Scholar] [CrossRef]
  75. Morsli, M.; Kerharo, Q.; Delerce, J.; Roche, P.-H.; Troude, L.; Drancourt, M. Haemophilus Influenzae Meningitis Direct Diagnosis by Metagenomic Next-Generation Sequencing: A Case Report. Pathogens 2021, 10, 461. [Google Scholar] [CrossRef]
  76. Jang, Y.; Kim, S.; Kim, N.; Son, H.; Ha, E.J.; Koh, E.J.; Phi, J.H.; Park, C.-K.; Kim, J.E.; Kim, S.-K.; et al. Nanopore 16s Sequencing Enhances the Detection of Bacterial Meningitis after Neurosurgery. Ann. Clin. Transl. Neurol. 2022, 9, 312–325. [Google Scholar] [CrossRef]
  77. Avershina, E.; Frye, S.A.; Ali, J.; Taxt, A.M.; Ahmad, R. Ultrafast and Cost-Effective Pathogen Identification and Resistance Gene Detection in a Clinical Setting Using Nanopore Flongle Sequencing. Front. Microbiol. 2022, 13, 822402. [Google Scholar] [CrossRef] [PubMed]
  78. Whittle, E.; Yonkus, J.A.; Jeraldo, P.; Alva-Ruiz, R.; Nelson, H.; Kendrick, M.L.; Grys, T.E.; Patel, R.; Truty, M.J.; Chia, N. Optimizing Nanopore Sequencing for Rapid Detection of Microbial Species and Antimicrobial Resistance in Patients at Risk of Surgical Site Infections. Msphere 2022, 7, e00964-21. [Google Scholar] [CrossRef] [PubMed]
  79. Yonkus, J.A.; Whittle, E.; Alva-Ruiz, R.; Abdelrahman, A.M.; Horsman, S.E.; Suh, G.A.; Cunningham, S.A.; Nelson, H.; Grotz, T.E.; Smoot, R.L.; et al. Answers in Hours: A Prospective Clinical Study Using Nanopore Sequencing for Bile Duct Cultures. Surgery 2022, 171, 693–702. [Google Scholar] [CrossRef]
  80. Ahmadi, A.; Khezri, A.; Nørstebø, H.; Ahmad, R. A Culture-, Amplification-Independent, and Rapid Method for Identification of Pathogens and Antibiotic Resistance Profile in Bovine Mastitis Milk. Front. Microbiol. 2022, 13, 1104701. [Google Scholar] [CrossRef] [PubMed]
  81. Sanderson, N.D.; Street, T.L.; Foster, D.; Swann, J.; Atkins, B.L.; Brent, A.J.; McNally, M.A.; Oakley, S.; Taylor, A.; Peto, T.E.A.; et al. Real-Time Analysis of Nanopore-Based Metagenomic Sequencing from Infected Orthopaedic Devices. BMC Genom. 2018, 19, 714. [Google Scholar] [CrossRef] [PubMed]
  82. Wang, C.-X.; Huang, Z.; Fang, W.; Zhang, Z.; Fang, X.; Li, W.; Yang, B.; Chen, L.; Liang, X.; Hu, H.; et al. Preliminary Assessment of Nanopore-Based Metagenomic Sequencing for the Diagnosis of Prosthetic Joint Infection. Int. J. Infect. Dis. 2020, 97, 54–59. [Google Scholar] [CrossRef] [PubMed]
  83. Cacho, A.; Smirnova, E.; Huzurbazar, S.; Cui, X. A Comparison of Base-Calling Algorithms for Illumina Sequencing Technology. Brief. Bioinform. 2015, 17, 786–795. [Google Scholar] [CrossRef]
  84. Wilkins, O.G.; Capitanchik, C.; Luscombe, N.M.; Ule, J. Ultraplex: A Rapid, Flexible, All-in-One Fastq Demultiplexer. Wellcome Open Res. 2021, 6, 141. [Google Scholar] [CrossRef]
  85. Perešíni, P.; Boža, V.; Brejová, B.; Vinař, T. Nanopore Base Calling on the Edge. Bioinformatics 2021, 37, 4661–4667. [Google Scholar] [CrossRef]
  86. Andrews, S. Fastqc: A Quality Control Tool for High Throughput Sequence Data. 2015. Available online: http://www.bioinformatics.babraham.ac.uk/projects/fastqc/ (accessed on 9 February 2023).
  87. Fukasawa, Y.; Ermini, L.; Wang, H.; Carty, K.; Cheung, M.S. Longqc: A Quality Control Tool for Third Generation Sequencing Long Read Data. G3 Genes Genom. Genet. 2020, 10, 1193–1196. [Google Scholar] [CrossRef]
  88. Leger, A.; Leonardi, T. Pycoqc, Interactive Quality Control for Oxford Nanopore Sequencing. J. Open Source Softw. 2019, 4, 1236. [Google Scholar] [CrossRef]
  89. Bolger, A.M.; Lohse, M.; Usadel, B. Trimmomatic: A Flexible Trimmer for Illumina Sequence Data. Bioinformatics 2014, 30, 2114–2120. [Google Scholar] [CrossRef] [PubMed]
  90. Martin, M. Cutadapt Removes Adapter Sequences from High-Throughput Sequencing Reads. EMBnet J. 2011, 17, 10–12. [Google Scholar] [CrossRef]
  91. Wick, R.R.; Judd, L.M.; Gorrie, C.L.; Holt, K.E. Completing Bacterial Genome Assemblies with Multiplex Minion Sequencing. Microbial. Genom. 2017, 3, e000132. [Google Scholar] [CrossRef]
  92. Wood, D.E.; Salzberg, S.L. Kraken: Ultrafast Metagenomic Sequence Classification Using Exact Alignments. Genome Biol. 2014, 15, R46. [Google Scholar] [CrossRef] [PubMed]
  93. Wood, D.E.; Lu, J.; Langmead, B. Improved Metagenomic Analysis with Kraken 2. Genome Biol. 2019, 20, 257. [Google Scholar] [CrossRef]
  94. Kim, D.; Song, L.; Breitwieser, F.P.; Salzberg, S.L. Centrifuge: Rapid and Sensitive Classification of Metagenomic Sequences. Genome Res. 2016, 26, 1721–1729. [Google Scholar] [CrossRef]
  95. Bankevich, A.; Nurk, S.; Antipov, D.; Gurevich, A.A.; Dvorkin, M.; Kulikov, A.S.; Lesin, V.M.; Nikolenko, S.I.; Pham, S.; Prjibelski, A.D.; et al. Spades: A New Genome Assembly Algorithm and Its Applications to Single-Cell Sequencing. J. Comput. Biol. 2012, 19, 455–477. [Google Scholar] [CrossRef]
  96. Koren, S.; Walenz, B.P.; Berlin, K.; Miller, J.R.; Bergman, N.H.; Phillippy, A.M. Canu: Scalable and Accurate Long-Read Assembly Via Adaptive K-Mer Weighting and Repeat Separation. Genome Res. 2017, 27, 722–736. [Google Scholar] [CrossRef]
  97. Li, H. Minimap and Miniasm: Fast Mapping and De Novo Assembly for Noisy Long Sequences. Bioinformatics 2016, 32, 2103–2110. [Google Scholar] [CrossRef]
  98. Kolmogorov, M.; Yuan, J.; Lin, Y.; Pevzner, P.A. Assembly of Long, Error-Prone Reads Using Repeat Graphs. Nat. Biotechnol. 2019, 37, 540–546. [Google Scholar] [CrossRef] [PubMed]
  99. Vaser, R.; Šikić, M. Time- and Memory-Efficient Genome Assembly with Raven. Nat. Comput. Sci. 2021, 1, 332–336. [Google Scholar] [CrossRef]
  100. Wick, R.R.; Judd, L.M.; Cerdeira, L.T.; Hawkey, J.; Méric, G.; Vezina, B.; Wyres, K.L.; Holt, K.E. Trycycler: Consensus Long-Read Assemblies for Bacterial Genomes. Genome Biol. 2021, 22, 266. [Google Scholar] [CrossRef] [PubMed]
  101. Khezri, A.; Avershina, E.; Ahmad, R. Hybrid Assembly Provides Improved Resolution of Plasmids, Antimicrobial Resistance Genes, and Virulence Factors in Escherichia Coli and Klebsiella Pneumoniae Clinical Isolates. Microorganisms 2021, 9, 2560. [Google Scholar] [CrossRef]
  102. Gurevich, A.; Saveliev, V.; Vyahhi, N.; Tesler, G. Quast: Quality Assessment Tool for Genome Assemblies. Bioinformatics 2013, 29, 1072–1075. [Google Scholar] [CrossRef]
  103. Simão, F.A.; Waterhouse, R.M.; Ioannidis, P.; Kriventseva, E.V.; Zdobnov, E.M. Busco: Assessing Genome Assembly and Annotation Completeness with Single-Copy Orthologs. Bioinformatics 2015, 31, 3210–3212. [Google Scholar] [CrossRef] [PubMed]
  104. Parks, D.H.; Imelfort, M.; Skennerton, C.T.; Hugenholtz, P.; Tyson, G.W. Checkm: Assessing the Quality of Microbial Genomes Recovered from Isolates, Single Cells, and Metagenomes. Genome Res. 2015, 25, 1043–1055. [Google Scholar] [CrossRef]
  105. Manchanda, N.; Portwood, J.L.; Woodhouse, M.R.; Seetharam, A.S.; Lawrence-Dill, C.J.; Andorf, C.M.; Hufford, M.B. Genomeqc: A Quality Assessment Tool for Genome Assemblies and Gene Structure Annotations. BMC Genom. 2020, 21, 193. [Google Scholar] [CrossRef]
  106. Seemann, T. ABRicate: Mass screening of contigs for antimicrobial resistance or virulence genes. Available online: https://github.com/tseemann/abricate (accessed on 9 February 2023).
  107. Feldgarden, M.; Brover, V.; Gonzalez-Escalona, N.; Frye, J.G.; Haendiges, J.; Haft, D.H.; Hoffmann, M.; Pettengill, J.B.; Prasad, A.B.; Tillman, G.E.; et al. Amrfinderplus and the Reference Gene Catalog Facilitate Examination of the Genomic Links among Antimicrobial Resistance, Stress Response, and Virulence. Sci. Rep. 2021, 11, 12728. [Google Scholar] [CrossRef]
  108. Bortolaia, V.; Kaas, R.S.; Ruppe, E.; Roberts, M.C.; Schwarz, S.; Cattoir, V.; Philippon, A.; Allesoe, R.L.; Rebelo, A.R.; Florensa, A.F.; et al. Resfinder 4.0 for Predictions of Phenotypes from Genotypes. J. Antimicrob. Chemother. 2020, 75, 3491–3500. [Google Scholar] [CrossRef]
  109. Alcock, B.P.; Raphenya, A.R.; Lau, T.T.Y.; Tsang, K.K.; Bouchard, M.; Edalatmand, A.; Huynh, W.; Nguyen, A.V.; Cheng, A.A.; Liu, S.; et al. Card 2020: Antibiotic Resistome Surveillance with the Comprehensive Antibiotic Resistance Database. Nucleic. Acids. Res. 2020, 48, D517–D525. [Google Scholar] [CrossRef] [PubMed]
  110. Hunt, M.; Mather, A.E.; Sánchez-Busó, L.; Page, A.J.; Parkhill, J.; Keane, J.A.; Harris, S.R. Ariba: Rapid Antimicrobial Resistance Genotyping Directly from Sequencing Reads. Microbial. Genom. 2017, 3, e000131. [Google Scholar] [CrossRef] [PubMed]
  111. Khezri, A.; Avershina, E.; Ahmad, R. Plasmid Identification and Plasmid-Mediated Antimicrobial Gene Detection in Norwegian Isolates. Microorganisms 2021, 9, 52. [Google Scholar] [CrossRef] [PubMed]
  112. Petit, R.A.; Read, T.D. Bactopia: A Flexible Pipeline for Complete Analysis of Bacterial Genomes. mSystems 2020, 5, e00190-20. [Google Scholar] [CrossRef]
  113. Quijada, N.M.; Rodríguez-Lázaro, D.; Eiros, J.M.; Hernández, M. Tormes: An Automated Pipeline for Whole Bacterial Genome Analysis. Bioinformatics 2019, 35, 4207–4212. [Google Scholar] [CrossRef]
  114. Seemann, T.; da Silva, A.G.; Bulach, D.M.; Schultz, M.B.; Kwong, J.C.; Howden, B.P. Nullarbor: Pipeline to generate complete public health microbiology reports from sequenced isolates. Available online: https://github.com/tseemann/nullarbor (accessed on 9 February 2023).
  115. Schürch, A.C.; van Schaik, W. Challenges and Opportunities for Whole-Genome Sequencing–Based Surveillance of Antibiotic Resistance. Ann. N. Y. Acad. Sci. 2017, 1388, 108–120. [Google Scholar] [CrossRef]
  116. Dahl, L.G.; Joensen, K.G.; Østerlund, M.T.; Kiil, K.; Nielsen, E.M. Prediction of Antimicrobial Resistance in Clinical Campylobacter Jejuni Isolates from Whole-Genome Sequencing Data. Eur. J. Clin. Microbiol. Infect. Dis. 2021, 40, 673–682. [Google Scholar] [CrossRef]
  117. Rebelo, A.R.; Bortolaia, V.; Leekitcharoenphon, P.; Hansen, D.S.; Nielsen, H.L.; Ellermann-Eriksen, S.; Kemp, M.; Røder, B.L.; Frimodt-Møller, N.; Søndergaard, T.S.; et al. One Day in Denmark: Comparison of Phenotypic and Genotypic Antimicrobial Susceptibility Testing in Bacterial Isolates from Clinical Settings. Front. Microbiol. 2022, 13, 804627. [Google Scholar] [CrossRef]
  118. Qi, L.; Li, H.; Zhang, C.; Liang, B.; Li, J.; Wang, L.; Du, X.; Liu, X.; Qiu, S.; Song, H. Relationship between Antibiotic Resistance, Biofilm Formation, and Biofilm-Specific Resistance in Acinetobacter Baumannii. Front. Microbiol. 2016, 7, 483. [Google Scholar] [CrossRef]
  119. Moran, R.A.; Anantham, S.; Holt, K.E.; Hall, R.M. Prediction of Antibiotic Resistance from Antibiotic Resistance Genes Detected in Antibiotic-Resistant Commensal Escherichia Coli Using Pcr or Wgs. J. Antimicrob. Chemother. 2016, 72, 700–704. [Google Scholar]
  120. Kumburu, H.H.; Sonda, T.; van Zwetselaar, M.; Leekitcharoenphon, P.; Lukjancenko, O.; Mmbaga, B.T.; Alifrangis, M.; Lund, O.; Aarestrup, F.M.; Kibiki, G.S. Using Wgs to Identify Antibiotic Resistance Genes and Predict Antimicrobial Resistance Phenotypes in Mdr Acinetobacter Baumannii in Tanzania. J. Antimicrob. Chemother. 2019, 74, 1484–1493. [Google Scholar] [CrossRef] [PubMed]
  121. Dey, A. Machine Learning Algorithms: A Review. Int. J. Comput. Sci. Inf. Technol. 2016, 7, 1174–1179. [Google Scholar]
  122. Sarker, I.H. Machine Learning: Algorithms, Real-World Applications and Research Directions. SN Comput. Sci. 2021, 2, 160. [Google Scholar] [CrossRef] [PubMed]
  123. Feretzakis, G.; Loupelis, E.; Sakagianni, A.; Kalles, D.; Martsoukou, M.; Lada, M.; Skarmoutsou, N.; Christopoulos, C.; Valakis, K.; Velentza, A.; et al. Using Machine Learning Techniques to Aid Empirical Antibiotic Therapy Decisions in the Intensive Care Unit of a General Hospital in Greece. Antibiotics 2020, 9, 50. [Google Scholar] [CrossRef] [PubMed]
  124. Yang, J.H.; Wright, S.N.; Hamblin, M.; McCloskey, D.; Alcantar, M.A.; Schrübbers, L.; Lopatkin, A.J.; Satish, S.; Nili, A.; Palsson, B.O.; et al. A White-Box Machine Learning Approach for Revealing Antibiotic Mechanisms of Action. Cell 2019, 177, 1649–1661.e9. [Google Scholar] [CrossRef]
  125. Ferreira, I.; Beisken, S.; Lueftinger, L.; Weinmaier, T.; Klein, M.; Bacher, J.; Patel, R.; von Haeseler, A.; Posch, A.E. Species Identification and Antibiotic Resistance Prediction by Analysis of Whole-Genome Sequence Data by Use of Aresdb: An Analysis of Isolates from the Unyvero Lower Respiratory Tract Infection Trial. J. Clin. Microbiol. 2020, 58, e00273-20. [Google Scholar] [CrossRef]
  126. Sundermann, A.J.; Chen, J.; Kumar, P.; Ayres, A.M.; Cho, S.-T.; Ezeonwuka, C.; Griffith, M.P.; Miller, J.K.; Mustapha, M.M.; Pasculle, A.W.; et al. Whole-Genome Sequencing Surveillance and Machine Learning of the Electronic Health Record for Enhanced Healthcare Outbreak Detection. Clin. Infect. Dis. 2022, 75, 476–482. [Google Scholar] [CrossRef]
  127. McCoubrey, L.E.; Gaisford, S.; Orlu, M.; Basit, A.W. Predicting Drug-Microbiome Interactions with Machine Learning. Biotechnol. Adv. 2022, 54, 107797. [Google Scholar] [CrossRef]
  128. Nguyen, M.; Brettin, T.; Long, S.W.; Musser, J.M.; Olsen, R.J.; Olson, R.; Shukla, M.; Stevens, R.L.; Xia, F.; Yoo, H.; et al. Developing an in Silico Minimum Inhibitory Concentration Panel Test for Klebsiella Pneumoniae. Sci. Rep. 2018, 8, 1–11. [Google Scholar] [CrossRef]
  129. Kim, J.; Greenberg, D.E.; Pifer, R.; Jiang, S.; Xiao, G.; Shelburne, S.A.; Koh, A.; Xie, Y.; Zhan, X. Vampr: Variant Mapping and Prediction of Antibiotic Resistance Via Explainable Features and Machine Learning. PLOS Comput. Biol. 2020, 16, e1007511. [Google Scholar] [CrossRef]
  130. Avershina, E.; Sharma, P.; Taxt, A.M.; Singh, H.; Frye, S.A.; Paul, K.; Kapil, A.; Naseer, U.; Kaur, P.; Ahmad, R. Amr-Diag: Neural Network Based Genotype-to-Phenotype Prediction of Resistance Towards Β-Lactams in Escherichia Coli and Klebsiella Pneumoniae. Comput. Struct. Biotechnol. J. 2021, 19, 1896–1906. [Google Scholar] [CrossRef] [PubMed]
  131. Sunuwar, J.; Azad, R.K. A Machine Learning Framework to Predict Antibiotic Resistance Traits and yet Unknown Genes Underlying Resistance to Specific Antibiotics in Bacterial Strains. Brief Bioinform. 2021, 22, bbab179. [Google Scholar] [CrossRef] [PubMed]
  132. Ren, Y.; Chakraborty, T.; Doijad, S.; Falgenhauer, L.; Falgenhauer, J.; Goesmann, A.; Hauschild, A.C.; Schwengers, O.; Heider, D. Prediction of Antimicrobial Resistance Based on Whole-Genome Sequencing and Machine Learning. Bioinformatics 2022, 38, 325–334. [Google Scholar] [CrossRef] [PubMed]
  133. Ellington, M.J.; Ekelund, O.; Aarestrup, F.M.; Canton, R.; Doumith, M.; Giske, C.; Grundman, H.; Hasman, H.; Holden, M.T.G.; Hopkins, K.L.; et al. The Role of Whole Genome Sequencing in Antimicrobial Susceptibility Testing of Bacteria: Report from the Eucast Subcommittee. Clin. Microbiol. Infect. 2017, 23, 2–22. [Google Scholar] [CrossRef] [PubMed]
  134. WHO. Global Antimicrobial Resistance and Use Surveillance System (Glass). Available online: https://www.who.int/initiatives/glass (accessed on 24 October 2022).
  135. Li, Y.; Yang, X.; Zhao, W. Emerging Microtechnologies and Automated Systems for Rapid Bacterial Identification and Antibiotic Susceptibility Testing. SLAS Technol. 2017, 22, 585–608. [Google Scholar] [CrossRef]
  136. Ahmad, A.; Hettiarachchi, R.; Khezri, A.; Singh Ahluwalia, B.; Wadduwage, D.N.; Ahmad, R. Highly sensitive quantitative phase microscopy and deep learning aided with whole genome sequencing for rapid detection of infection and antimicrobial resistance. Front. Microbiol. 2023, 14, 1154620. [Google Scholar] [CrossRef]
  137. Lister, A.; Avershina, E.; Ali, J.; Devitt, G.; Hanrahan, N.; Highmore, C.; Webb, J.; Muller, F.; Mahajan, S.; Ahmad, R. Multi-Excitation Raman Spectroscopy Complements Whole Genome Sequencing for Rapid Detection of Bacterial Infection and Resistance in Who Priority Pathogens. bioRxiv 2022. [Google Scholar] [CrossRef]
  138. Gieroba, B.; Krysa, M.; Wojtowicz, K.; Wiater, A.; Pleszczyńska, M.; Tomczyk, M.; Sroka-Bartnicka, A. The Ft-Ir and Raman Spectroscopies as Tools for Biofilm Characterization Created by Cariogenic Streptococci. Int. J. Mol. Sci. 2020, 21, 3811. [Google Scholar] [CrossRef]
  139. Tewes, T.J.; Centeleghe, I.; Maillard, J.Y.; Platte, F.; Bockmühl, D.P. Raman Microscopic Analysis of Dry-Surface Biofilms on Clinically Relevant Materials. Microorganisms 2022, 10, 1369. [Google Scholar] [CrossRef]
  140. Afolayan, A.O.; Bernal, J.F.; Gayeta, J.M.; Masim, M.L.; Shamanna, V.; Abrudan, M.; Abudahab, K.; Argimón, S.; Carlos, C.C.; Sia, S.; et al. Overcoming Data Bottlenecks in Genomic Pathogen Surveillance. Clin. Infect. Dis. 2021, 73, S267–S274. [Google Scholar] [CrossRef]
  141. EU. Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions Building Trust in Human-Centric Artificial Intelligence; EU: Brussels, Belgium, 2019. [Google Scholar]
  142. EU. Draft Ethics Guidelines for Trustworthy AI; EU: Brussels, Belgium, 2021. [Google Scholar]
Figure 1. Pipelines for sequencing data analysis for clinical application. (A) Analysis steps need to be performed for pathogen identification and resistance markers search in clinical samples. (B) Existing pipelines and steps they include. Yellow—Illumina data only; green—ONT data only. Connected dots represent steps that are automatically performed in the workflow, whereas isolated dots represent independent analysis steps that need to be called by the user. The figure was created using draw.io.
Figure 1. Pipelines for sequencing data analysis for clinical application. (A) Analysis steps need to be performed for pathogen identification and resistance markers search in clinical samples. (B) Existing pipelines and steps they include. Yellow—Illumina data only; green—ONT data only. Connected dots represent steps that are automatically performed in the workflow, whereas isolated dots represent independent analysis steps that need to be called by the user. The figure was created using draw.io.
Antibiotics 12 00781 g001
Figure 2. An overview of the current paradigm and future scenario in clinical microbiology. The current paradigm for the diagnosis in the case of infection suffers from a high turnaround time for both pathogen identification and antibiotic susceptibility testing (4–5 days). The potential future scenario with the use of whole genome sequencing has the potential to remarkably reduce the turnaround time. In the future scenario, sequencing can be implemented following direct sequencing from biological material (<24 h) or from bacterial colonies in the culture (24–36 h). Sequencing needs to be followed by proper bioinformatic analyses and ideally combined with a machine learning approach. Created with BioRender.com.
Figure 2. An overview of the current paradigm and future scenario in clinical microbiology. The current paradigm for the diagnosis in the case of infection suffers from a high turnaround time for both pathogen identification and antibiotic susceptibility testing (4–5 days). The potential future scenario with the use of whole genome sequencing has the potential to remarkably reduce the turnaround time. In the future scenario, sequencing can be implemented following direct sequencing from biological material (<24 h) or from bacterial colonies in the culture (24–36 h). Sequencing needs to be followed by proper bioinformatic analyses and ideally combined with a machine learning approach. Created with BioRender.com.
Antibiotics 12 00781 g002
Table 1. The use of real-time metagenomic Oxford Nanopore sequencing (mONS) for infection identification and ARG detection. LRTI: lower respiratory tract infection, UTI: urinary tract infection, BSI: bloodstream infection, SSI: surgical site infection, PJI: prosthetic joint infections, IBS: intraoperative bile aspirates, JF: joint fluid, PT: prosthetic tissue. * NR: not reported, ** the number of analyzed samples <10, and sensitivity/specificity estimates should be treated with care.
Table 1. The use of real-time metagenomic Oxford Nanopore sequencing (mONS) for infection identification and ARG detection. LRTI: lower respiratory tract infection, UTI: urinary tract infection, BSI: bloodstream infection, SSI: surgical site infection, PJI: prosthetic joint infections, IBS: intraoperative bile aspirates, JF: joint fluid, PT: prosthetic tissue. * NR: not reported, ** the number of analyzed samples <10, and sensitivity/specificity estimates should be treated with care.
DiseaseSample TypeNumber of samplesCulture IndependentTurnaround Time (h)Sensitivity (%)Specificity (%)ARG DetectionMain OutcomeReference
LRTI Sputum81yes696.641.7yesmONS can rapidly and accurately characterize bacterial LRIs and might contribute to a reduction in broad-spectrum antibiotic use[70]
UTIUrine86yesNR *71–100-yesmONS pathogen and resistome detection for the diagnosis of UTIs warrants prospective clinical validation[71]
76yes686.796.8yesmONS is a promising clinical diagnostic tool for infectious diseases[72]
MeningitisCerebral spinal fluid1yes~5--yesmONS enabled the direct diagnosis of meningitis caused by Pasteurella multocida, which is not routinely detected by current point-of-care assays[73]
1yes2--yesmONS enabled the diagnosis of community-acquired Streptococcus pneumoniae meningitis, providing a less than two-hour workflow including only a 20 min sequencing time, detection, identification, typing, and an in silico antibiogram[74]
1yes6--yesmONS could be considered for the point-of-care diagnosis of infectious meningitis, by direct identification of pathogenic genomes and their genotypes/serotypes[75]
285yesNR10094.4no16S rDNA mONS was more effective than conventional culture in postoperative bacterial meningitis and may contribute to evidence-based decisions for antibiotic maintenance and discontinuation[76]
BSIBlood8no3.5–6100 **83.1 **yesIdentification of pathogens was possible after 10 min of sequencing and all predefined AMR-encoding genes and plasmids from monoculture experiments were detected within one hour using raw mONS data[67]
2no6.5100 **-yesFlongle data allowed for rapid bacterial ID and resistome detection based on the first 1000–3000 generated sequences (10 min to 3 h from the sequencing start)[77]
SSIIBS42yes6–14100-yesThe results generated using mONS were similar to the current methods of detection but were obtained in a significantly shorter amount of time. mONS could be used to tailor antibiotics in surgical patients and reduce the use of broad-spectrum antibiotics[78]
42yes8100100yesRapid microbial profiling with mONS is feasible with broader organism and resistance profiling compared to standard cultures. mONS has perfect negative predictive values and can potentially improve antibiotic stewardship[79]
MastitisMilk24yes910092.3yesA proof-of-concept study that established an effective method for host removal and direct MinION sequencing from mastitis milk[80]
PJISonication fluids7yesNR100 **100 **noA novel, scalable pipeline for the real-time analysis of MinION sequence data and the use of this pipeline to show initial the proof of concept that mONS can provide rapid and accurate diagnosis for prosthetic joint infections[81]
JF and PT9yes14–22100 **80 **yesA proof-of-concept study that established that mONS can function as a rapid and accurate tool in PJI diagnostic microbiology[82]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Avershina, E.; Khezri, A.; Ahmad, R. Clinical Diagnostics of Bacterial Infections and Their Resistance to Antibiotics—Current State and Whole Genome Sequencing Implementation Perspectives. Antibiotics 2023, 12, 781. https://doi.org/10.3390/antibiotics12040781

AMA Style

Avershina E, Khezri A, Ahmad R. Clinical Diagnostics of Bacterial Infections and Their Resistance to Antibiotics—Current State and Whole Genome Sequencing Implementation Perspectives. Antibiotics. 2023; 12(4):781. https://doi.org/10.3390/antibiotics12040781

Chicago/Turabian Style

Avershina, Ekaterina, Abdolrahman Khezri, and Rafi Ahmad. 2023. "Clinical Diagnostics of Bacterial Infections and Their Resistance to Antibiotics—Current State and Whole Genome Sequencing Implementation Perspectives" Antibiotics 12, no. 4: 781. https://doi.org/10.3390/antibiotics12040781

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop