Human Microbiome Journal 9 (2018) 1–6

Contents lists available at ScienceDirect

Human Microbiome Journal

journal homepage: www.elsevier.com/locate/humic

Metagenomic and clinical microbiology T

⁎
Sophie Amrane, Jean-Christophe Lagier
Aix Marseille Univ, IRD, AP-HM, MEPHI, IHU-Méditerranée Infection, Marseille, France

A R T I C LE I N FO A B S T R A C T

Keywords: As a result of Next Generation Sequencing methods, metagenomic studies have become increasingly widespread.
Metagenomic After being first applied to microbiome description, metagenomics is currently proposed as a diagnostic tool in
Clinical microbiology clinical microbiology, although this application remains confined to the field of research. In this review, we will
Next Generation Sequencing discuss the application of metagenomics to the detection of bacterial pathogens and demonstrate that the in-
Diagnosis
terpretation of the metagenomic results may fluctuate depending on the type of sample analyzed. However, we
Limitation
propose a view of metagenomic application to the evaluation of antimicrobial resistance, epidemic investigations
and forensic medicine. Secondly, we present the many limits of metagenomic interpretation and application in
routine clinical microbiology. From our perspective, metagenomics is not yet reliable enough for general use in
clinical microbiology.

1. Introduction In this review, we proposed exploring the current applications of

Clinical microbiology has dramatically evolved since the end of the
20th century with the appearance of OMICS technologies. The multi-
plication of molecular tools has enabled rapid diagnosis without any
culture step and has changed the way that some infections (such as
infectious sexual diseases caused by Chlamydia trachomatis) are de-
tected [1]. Simultaneously, the identification of cultured microorgan-
isms has been facilitated as a result of MALDI-TOF mass spectrometry
and its application to clinical microbiology [2]. Currently, microbial
antibiotic sensibility testing remains the lengthiest part of the diagnosis
process [1]. Classical clinical microbiology methods rely on the Koch
postulate, which highlights the importance of a pure culture of a mi-
croorganism to prove its pathogenicity. However, since the advent of
molecular tools and metagenomic analyses, culture sometimes appears
to have been dropped [3]. Nevertheless, attempts by some authors to
predict that molecular tools would replace pure culture were rapidly
abandoned [3].
Metagenomics, the principle of which relies on the genomic analysis
of a sample from a complex environment containing more than one
microorganism, provides a view of the composition of this sample.
Metagenomic studies became increasingly accessible with the advent of
Next Generation Sequencing (NGS) [4]. For metagenomic analysis, NGS
can be used with two different approaches: targeted metagenomics on a
specific chosen amplified region (such as the 16S region) or shotgun
metagenomics, which relies on the amplification of all the sequences in
a sample without hypothesizing about its content [4].
metagenomics in clinical microbiology, focusing on the risks of these
methods and difficulties with their interpretation. We propose de-
termining whether metagenomics could be currently used as a diag-
nostic tool.
2. Current applications of metagenomics to clinical microbiology
2.1. Detection of pathogens
Metagenomics has been used for the detection of pathogens in
clinical samples. We will describe its application to bacterial micro-
biology. Most studies report the evaluation of metagenomics compared
to traditional culture method. However, in some cases, metagenomic
analysis has made diagnosis, which was not previously performed by
culture or standard molecular tools (Table 1).
Few studies have reported on the use of metagenomics as a tool for
prospective diagnosis. Wilson et al. first reported a case of encephalitis
which was solved by shotgun metagenomics. Despite extensive ex-
plorations including 16S rRNA amplification and sequencing on cere-
brospinal fluid, no etiology had been found. Leptospirosis was identified
in the cerebrospinal fluid (CSF) by metagenomics and was not detected
in a control sample. The diagnosis was further confirmed by specific
molecular tools and serology [5]. Confirmation of the diagnosis by both
Leptospira specific PCR and serology could lead us to think that these
examinations could have be performed prior to NGS, however, the very
atypical clinical manifestations demonstrate the contribution which

⁎
Corresponding author at: Aix Marseille Université, IRD, AP-HM, MEPHI, IHU – Méditerranée Infection, 19-21 Boulevard Jean Moulin, 13005 Marseille, France.
E-mail address: jclagier@yahoo.fr (J.-C. Lagier).

https://doi.org/10.1016/j.humic.2018.06.001
Received 18 January 2018; Received in revised form 1 June 2018; Accepted 15 June 2018
Available online 14 July 2018
2452-2317/ © 2018 The Author(s). Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).
 S. Amrane, J.-C. Lagier

Table 1
Diagnosis by metagenomic analysis.
Studies Sampling Cases Metagenomic analysis Diagnosis

Nakamura, 2008 [50] Stool 34 y.o. man, diarrhea and fever after eating chicken. Retrospective analysis of stool sample with Clostridium jejuni
Standard culture negative, PCR for Norovirus negative. shotgun pyrosequencing (Roche 454) Confirmation by Campylobacter specific PCR and culture on
Negative control: not mentioned specific media
Wilson, 2014 [5] CSF 14 y.o. man, SCID post HSCT, meningoencephalitis and status epilepticus. Prospective analysis of CSF by shotgun Leptospira
Diagnostic workup including brain biopsy and 16S PCR on CSF(x2) was sequencing(Illumina). Confirmation by Leptospira specific PCR on CSF and serology
unrevealing. Negative control: serum from unrelated
patient.
Mongkolrattanothai, 2017 CSF 11 y.o. girl, fever, headache, nausea, shivering for four weeks Prospective analysis of CSF by shotgun Brucella
[51] Standard culture and mycobacterial culture negative, no 16S PCR sequencing (Illumina) Confirmation by Brucella specific PCR on CSF and serology
performed on CSF. Neurotuberculosis suspected Negative control: elution buffer
Salzberg, 2016 [17] Brain biopsy 67 y.o. women, osteomyelitis, lung disease, multifocal brain and spinal Prospective analysis of brain biopsy by M. tuberculosis Confirmation by anatomopathology (granuloma),
lesion. shotgun sequencing (Illumina) brain cultures stay negative.
Standard analysis didn’t led to a diagnosis before brain biopsy. Quantiferon Negative control: not mentioned
test showed indeterminate results (×4)
Yao, 2016 [52] CSF 3 cases of suspected encephalitis caused by L. monocytogenes Prospective analysis of CSF by shotgun L. monocytogenes
CSF Cultures were negative. sequencing (Ion Torrent) Confirmation by L. monocytogenes specific PCR on CSF.Cases 1
Negative control: 3 CSFs from patients with and 3: blood culture positive for L. monocytogenes, case 2: blood

2
autoimmune encephalitis. culture positive for gram-positive bacteria.
Thoendel, 2017 [8] Articular fluid 52 y.o. man, hypogammaglobulinemia post B cell lymphoma, late left TKA Retrospective analysis of sonicate fluid Mycoplasma salivarium
infection with no documentation (bacterial, fungal, mycobacterial culture from arthroplasty extraction by shotgun Failure of specific PCR on anatomopathology slides.
of articular fluid and per operative samples negative, Mycoplasma hominis sequencing (Illumina) Failure of sonicate fluid culture on specific media
specific PCR on articular fluid negative, no 16S PCR conducted). Negative control: not mentioned
Fukui, 2015 [6] Cardiac valve 35 y.o. man, Infective endocarditis requiring valve replacement. Prospective analysis of the valve by Abiotrophia defective
Hemoculture and valve culture remain negative (16S PCR on the valve not shotgun sequencing (Illumina) Confirmation: not mentioned
mentioned) Negative control: not mentioned
Abril, 2016 [16] Plasma 60 y.o. man with septic shock and acute respiratory distress. Prospective analysis of plasma by shotgun Capnocytophaga canimorsus (result on day 5)
Gram smear: gram negative bacilli sequencing (Illumina) Confirmation by culture and MADI-TOF identification (result on
Negative control: used day 5)
Failure of two 16S amplification by PCR on the bacterial road,
success on the third attempt at day 55.
Ye, 2015 [53] Blood 2 y.o. boy, alloHSCT for leukemia, fever and rash Prospective analysis of blood by shotgun Propionibacterium acnes
No diagnosis, no 16S PCR performed, failure of probabilistic treatments sequencing (Ion) Confirmation by P. acnes specific PCR on blood: no statistical
Negative control: blood from another difference between case and control.
patient of the same ward (although positive
for P. acnes)
Pendleton, 2017 [15] BAL 41 y.o. woman with connective tissue and interstitial lung diseases with Prospective analysis of BAL by shotgun Pseudomonas aeruginosa (results at H9)
hypoxic respiratory failure sequencing (Ion) Confirmed by culture (results at H23)
59 y.o. man with abdominal sepsis and hypoxic respiratory failure Negative control: not mentioned Staphylococcus aureus
Confirmed by culture the next day
Human Microbiome Journal 9 (2018) 1–6
S. Amrane, J.-C. Lagier Human Microbiome Journal 9 (2018) 1–6

metagenomics made to this clinical case. Fukui et al. reported a case of confirm the result).
culture negative endocarditis (blood culture and valve culture remained
negative but no 16S PCR was performed on blood or the cardiac valve)
2.2. Resistome evaluation
where shotgun metagenomics revealed the presence of Abiotrophia de-
fectiva [6]. Nevertheless, the use of broad range PCR on the blood or
Since most of the bacteria composing the gut microbiota are un-
valve would probably have resulted in diagnosis, as previously de-
cultivable bacteria [18], the use of NGS is interesting for the detection
scribed [7]. In three other studies (Table 1) metagenomics detected a
of antibiotic resistance genes (ARG). In a metagenomic study on gut
pathogen missed by previous explorations and actually oriented the
microbiota with a focus on the detection of methicillin resistant Sta-
therapeutic approach. In all these cases, the studied sample was pre-
phylococcus aureus, vancomycin resistant Enterococcus and multi drug
sumed to be sterile. Other authors used metagenomics as a diagnostic
resistant (MDR) Enterobacteriae in three groups (high risk inpatients,
tool for the retrospective evaluation of unsolved cases (Table 1).
low risk outpatients and controls), Andersen et al. demonstrated that
Thoendel et al. used metagenomics to analyze a sonicate fluid sample
ARGs were significantly higher among patients versus controls. More-
from an arthroplasty and detected Mycoplasma salivarium which had
over, they demonstrated that ARGs could be detected by metagenomics
never been detected in bone and joint infections [8]. Nonetheless, in
and overlooked by culture [19]. Gyarmati et al. used metagenomics to
this case, following total knee arthroplasty (TKA) removal and re-
analyze blood from neutropenic patients at different points in their
placement by a spacer, the patient underwent six weeks of empiric
infection. More neutropenic was the patient; more microorganisms
antibiotic treatment with cefepime and vancomycin, and experienced
were detected in blood stream. However, they did highlight resistance
good evolution and TKA reimplantation. These two antibiotics are not
to tetracycline for two patients and resistance to macrolide for three
active against Mycoplasma. Only gentamycin in the spacer could have
patients [20]. In their study, Zhou et al. identified 27 ARGs with at least
treated the detected microorganism [8]. Indeed, the role of M. sali-
one resistance gene in half of the samples. 25.9% of the samples con-
varium in this infection remains questionable.
tained ARGs against cephalosporin and 25.9% of the samples contained
Other studies reported the efficiency of metagenomics for pathogen
ARGs against tetracycline [11]. In a model of clinical urine samples and
screening in comparison to traditional methods. Hasman et al. analyzed
healthy urine samples spiked with MDR Escherichia coli, Schmidt et al.
urine samples from patients with suspected urine infections, using
used direct metagenomics to detect 92% of the ARGs detected by direct
culture and metagenomic and demonstrating good concordance be-
sequencing of the bacterial strains [21]. Metagenomics can be an in-
tween the two methods [9]. However, metagenomic analysis frequently
teresting tool for simultaneous detection of a pathogen and analysis of
revealed more than one pathogen and did not discriminate responsible
its ARG.
pathogen. Furthermore, the authors did not mention the use of control
samples during sequencing. Hilton et al. used culture and metage-
nomics (16S targeted and shotgun) to analyze clinical samples from 2.3. Epidemic investigation
patients with ventilator-associated pneumonia [10]. Metagenomics
correctly identified the responsible pathogen, but the results were In an investigation into the outbreak of Shiga toxigenic Escherichia
polymicrobial for each sample and the responsible pathogen was not coli O104:H4, the authors retrospectively analyzed 40 culture positive
always the most widely detected. Zhou et al. evaluated the concordance samples. Using shotgun metagenomics, they reassembled genes from
between the metagenomic focus of 16S or shotgun and PCR for the the responsible pathogens and detected this sequences in 67% of the
detection of Clostridium difficile [11]. They detected the bacterium in samples [22]. Nonetheless, the determinant gene was not detected in all
90% of the samples. Kirstahler et al. compared metagenomic with two the samples, despite a high sequencing depth. Huang et al. investigated
DNA extraction methods to standard bacterial culture for diagnosis of two simultaneous foodborne outbreaks caused by Salmonella enterica.
endophtalmitis [12]. For the 12 culture positive vitreous samples, Using metagenomics, they proved that the two outbreaks were in-
concordance with metagenomic results was found in 9 cases (75%). dependent of one another and were caused by different isolates. They
Most interestingly, the authors used as control saline solution, vitreous also found a probable bacterial coinfection [23].
samples supposed not infected and blank DNA extraction solution; they Nevertheless, no metagenomic analysis has been prospectively ap-
highlighted frequent environmental contaminants [12]. All these stu- plied to an outbreak.
dies were conducted on samples that were presumed not to be sterile.
For interpretation purposes, it is essential to have the results of the
2.4. Forensic medicine
sequencing in a control group, as previously demonstrated [13]. In-
deed, pathogenic bacteria can be found in control samples without
An alternative use of metagenomics in clinical microbiology is its
clinical symptoms [14]. All the difficulties with interpretation result
application to forensic medicine. Microbiota can be used as personal
from the choice of a responsible pathogen from more than one pathogen
identity card. As an example, salivary microbiota can be used to dif-
detected by the method.
ferentiate between two individuals [24]. Different microbiota are spe-
On the other hand, metagenomics has also been used for rapid di-
cific to different parts of the body or different kinds of fluid. On an
agnosis before standard culture results. Pseudomonas aeruginosa was
object which has been touched, much more bacteria are deposed than
detected in a bronchoalveolar lavage by Ion technology sequencing
human DNA and metagenomic analysis can help to identify or exclude
seventeen hours before the results of the standard culture were avail-
suspects in criminal cases or to identify the nature of a sample [25].
able [15]. Capnocytophaga canimorsus was detected in the blood of a
septic patient before MALDI-TOF mass spectrometry results on the
bacterial strain were complete [16]. 3. Current limits to the interpretation of metagenomic data in
In conclusion, metagenomics could become a diagnostic tool in rare clinical microbiology
clinical situations, but its interpretation differs depending on the
sample types. A single bacterium detected in a cerebrospinal fluid In a recent study, Abayasekara et al. evaluated metagenomic sen-
sample is easiest to interpret in comparison with polymicrobial results sitivity at 52.7% and specificity at 91.8% for pathogen detection
from a sample known unsterile such as a sputum sample. Of the fifteen compare to standard culture for various clinical samples [26]. This low
studies cited in this part, eleven proposed a diagnosis as a result of sensitivity highlighted the need of a very cautious interpretation of
metagenomics but only eight used metagenomics as a diagnosis tool in metagenomic results. In this part we will describe the limits and bias
a prospective way. Moreover, only two of these studies [6,17] used that could interfere with result’s interpretation in each step of the
metagenomic as the only diagnostic tool (no other molecular tool could metagenomic process.

3
S. Amrane, J.-C. Lagier
3.1. Sample collection and sample preparation
In clinical microbiology, the collection and preparation of samples
must be organized. The time between collecting a sample from the
patient and its analysis may cause a deterioration in the DNA and lead
to false interpretations. This has been studied on infants’ stool samples,
where a storage for more than two hours at room temperature led to
modifications in the relative abundance of some microbiota genus, in-
cluding an increase in the relative abundance of Streptococcus up to 5.2
fold [27]. The authors noticed that 75% of the genus with an increase of
more than two-fold relative abundance after one hour at room tem-
perature were aero tolerant, and they suggest that oxygen exposure is
an important factor to consider. Another difficulty is the quantity of
extracted DNA, depending of the sample tested: a minimum amount of
DNA is required to obtain metagenomic sequencing. In their study,
Ruppé et al. planned to sequence 179 samples from bone and joint
infections, but due to the proportion of extracted DNA being too low,
they could only sequence 24/179 (13.4%) of their samples [28].
3.2. Extraction bias
It has been well described that DNA extraction methods influence
the results. Indeed, the final results can be very different from the ex-
pected sample composition: for example, pyrosequencing failed to de-
tect Gram negative bacteria [29]. With a comparative analysis of two
DNA extraction methods used during the Human Microbiome Project, it
has been proven that the methodology influences results with differ-
ences to the phyla level [30]. Brooks et al. used a model with 80 mock
bacteria to demonstrate that the DNA extraction kit influenced results,
proving that extraction and amplification are the two steps generating
most of the bias in metagenomic analysis [31]. Angelakis et al. tested
ten DNA extraction methods on the same stool sample. They obtained
consistent species diversity (Shannon index from 1.37 to 4.4) and
richness (Chao index from 227 to 2,714) variations between the dif-
ferent tested protocols [32].
3.3. Human DNA and environmental microorganisms in the sample
Human DNA represents the majority of most clinical samples, and
could account for 99% of the reads produced [17]. Microbial DNA
enrichment can be suggested in samples with low bacterial abundance.
In a comparative study between two enrichment kits, Thoendel et al.
were able to amplify DNA with an enrichment factor from 13 to 9580
[33]. In another hand, human DNA can although be sequenced in the
aim to analyze host response [34]. In targeted metagenomics, the de-
sired bacteria can be directly targeted with probes before amplification
and sequencing. It has been proposed for the detection of Myco-
bacterium tuberculosis from the sputum [35].
Environmental contaminants could skew the analysis, especially in
samples with a low biomass [36]. Contaminants can, however, be
created in silico [37]. A negative control can be sequenced in parallel
with the sample with a view to detecting these contaminants [12,37].
Control samples can be spike with calibrate DNA or entire micro-
organism [38]. “Swipe test” with swabbing of the lab environment can
also be proposed for detection of this contaminants [34].
3.4. Depth bias
Metagenomics is only able to distinguish bacteria with concentra-
tions greater than 10^6 bacteria per gram of feces [39]. Due to the
majority of human DNA in each sample, a very high sequencing depth
in shotgun metagenomic is required to be able to sequence the expected
pathogen [40]. However, some enteric pathogens can be aggressive in a
very low concentrations, such as Salmonella Typhi and Shigella [39] and
would be missed by metagenomic analysis. To overcome this limitation,
protocols of pathogens DNA enrichment and human DNA removal are
4
Human Microbiome Journal 9 (2018) 1–6
available [38]. Controls spiked with a constant concentration of bac-
teria might be used [38].
3.5. Bioinformatic pipelines and standardization
Once sequencing is obtained, the raw data must be transformed into
a taxonomic list of bacteria ordered by relative abundance. During this
process, various steps must be respected: sequence adaptors must be
removed from NGS, human DNA sequences must be removed, reads
must be assembled into contigs, each sequence must be cross-referenced
to a database [41]. The choice of database and the parameters defining
a match with the genus or species level must be chosen according to the
desired speed and accuracy of analysis [41]. An incomplete reference
database would be a confounding factor for this step of the analyze
[38]. Currently, each laboratory constructs its own pipeline and there is
no unique platform to perform all these steps. In clinical microbiology,
pipelines are available for pathogen screening (SURPI for example)
[42]. In addition to the lack of standardization, limitations persist such
as allocating sequences for AMR genes to the detected pathogen or to
the background microbiota [43]. All informatics tools require specific
expertise and bioinformaticians need solid biological training to guide
their research according to the question being asked [40]. Moreover, all
the data generated represent a huge volume of information, and pro-
visions for their storage must be anticipated.
3.6. Interpretation of results
Once the microbial composition of the sample is available, the role
of each bacterium must be determined.
First, it must be determined whether the sequence detected belongs
to an active bacterium. Metagenomics only detects DNA, there is no
way of knowing whether the pathogens detected are dead, quiescent
(for example after first line of antibiotics) or alive. Metatranscriptomics,
i.e. RNA amplification, can be proposed for active bacteria detection
[4]. This is a key point to be resolved in a near future.
In a case of polymicrobial results, responsibility for each bacterium
detected must then be assessed. In a comparison study between meta-
taxonomics, shotgun metagenomics and culture for the diagnosis of
pneumonia, Hilton et al. demonstrated that the pathogen with the most
relative abundance is not always the pathogen responsible for the dis-
ease [10]. Moreover, metagenomic results always show more than one
microorganism, and there are no clues to determine which bacterium is
responsible for the disease. Stämmlet et al. proposed calibrating the
metagenomic results by adding a controlled amount of bacteria to a
clinical sample and determining the microbial load detected by the
reads [44].
Finally, unidentified operational taxonomic units (OTUs) should not
be overlooked. A large proportion (80%) of bacteria detected by me-
tagenomic studies corresponds to uncultivated bacteria or bacteria
which are not yet cultivable [39]. If metagenomics is used as a diag-
nostic tool, there is currently no clue whether the sequence with the
highest relative abundance corresponds to unidentified OTUs. Another
OMICs approach known as culturomics is currently proposed. By cul-
turing the same sample with various culture conditions, new bacteria
can be discovered, which helps to increase the number of known bac-
teria [39]. Genomes of these new bacteria are then available for me-
tagenomics projects.
4. Current technical limitations of the generalization of
metagenomics to clinical microbiology
4.1. Time and cost limitations of results
With technological advances, the time between sampling and the
final results is decreasing. As metagenomics is more widely used, its
cost is becoming more affordable for laboratories. Most NGS
S. Amrane, J.-C. Lagier
Fig. 1. Number of publications per year on metagenomics and metagenomics in clinical microbiology. Pubmed research 12/30/2017.
technologies used in research laboratories are complete within 24–72 h
(excluding analysis time) and third generation sequencing technologies
can even take less than one hour [45]. Some authors, such as Af-
shinnekoo et al., have proposed protocols for the metagenomic analysis
of clinical samples with specific deadlines. A first track process known
as STAT makes analysis possible in less than 24 h for bacterial patho-
gens, antimicrobial resistance identification and virulence factor. Two
other tracks enable a deeper analysis with RNA enriched methods or
targeted PCR amplification before sequencing and last between five and
seven days [46]. In clinical bacteriology, the time between sampling
and results has been significantly reduced since the use of MALDI-TOF
MS for bacterial identification [40]. Nevertheless, the timescale for
performing an antibiogram cannot yet be shortened. Because metage-
nomics can detect ARGs, information on microbial resistance in a
sample can be obtained quickly [46]. It seems reasonable to expect that,
within a few years, NGS methods will enable faster diagnosis than
conventional bacteriology [37].
The cost for whole genome sequencing reduced from 100,000 dol-
lars to close to 1000 dollars [47] in 2011. However, the real cost of the
entire sequencing process is higher. As sequencing costs decrease,
bioinformatic costs increased [47], with an estimation for sampling,
sequencing, data management, and downstream analysis standing at
6500 dollars. Moreover, sequencing projects require qualified per-
sonnel and hiring in the bioinformatics sector has increased since the
advent of NGS [48].
4.2. Interpretations for physicians
At the end of this process, results may be available for physicians.
Because the final decision to prescribe antibiotics is made by the phy-
sician, they need to know the methods that have been used and the
reliability of the results [40]. A real discussion between physicians and
biologists is essential, especially if the final metagenomic results are
uncertain (with regards to pathogens or contaminants for instance)
[49]. It is interesting to notice that, compared to the global metage-
nomic literature; metagenomics applied to clinical microbiology re-
presents only a small proportion of the available literature (Fig. 1),
indicating that this application is still restricted to research.
5. Conclusion
The use of metagenomics in clinical microbiology remains rare.
Despite around fifteen studies using this method to detect bacterial
pathogens in clinical samples, in only two cases metagenomics seems to
have been the only tool to reach the diagnosis. The application of
5
Human Microbiome Journal 9 (2018) 1–6
metagenomics to ARG detection appears interesting but has not yet
been tested as a routine tool leading to modifications in antibiotic
treatment. Other applications such as epidemic investigation or forensic
medicine remain anecdotal.
The democratization of this technic currently appears unlikely, due
to the multiple biases in its interpretation, the lack of standardization
and its availability in only a few big laboratories.
In our evaluation, in most cases diagnosis was possible as the result
of traditional culture methods or simpler molecular tools such as real
time PCR. In conclusion, we consider that metagenomics is currently
not reliable enough for a general use in clinical microbiology.
Acknowledgements
This work has benefited from the French State support, managed by
the ‘Agence Nationale pour la Recherche’ including the "Programme
d’Investissement d’avenir" under the reference Méditerranée Infection
10-IAHU-03.
References
[1] van Belkum A, Durand G, Peyret M, Chatellier S, Zambardi G, Schrenzel J, et al.
Rapid clinical bacteriology and its future impact. Ann Lab Med 2013;33(1):14.
[2] Seng P, Drancourt M, Gouriet F, La Scola B, Fournier P, Rolain JM, et al. Ongoing
revolution in bacteriology: routine identification of bacteria by matrix-assisted laser
desorption ionization time-of-flight mass spectrometry. Clin Infect Dis Aug.
2009;49(4):543–51.
[3] Fredricks DN, Relman DA. Sequence-based identification of microbial pathogens: a
reconsideration of Koch’s postulates. Clin Microbiol Rev 1996;9(1):18–33.
[4] Lavezzo E, Barzon L, Toppo S, Palù G. Third generation sequencing technologies
applied to diagnostic microbiology: benefits and challenges in applications and data
analysis. Expert Rev Mol Diagn 2016;16(9):1011–23.
[5] Wilson MR, Naccache SN, Samayoa E, Biagtan M, Bashir H, Yu G, et al. Actionable
diagnosis of neuroleptospirosis by next-generation sequencing. N Engl J Med
2014;370(25):2408–17.
[6] Fukui Y, Aoki K, Okuma S, Sato T, Ishii Y, Tateda K. Metagenomic analysis for
detecting pathogens in culture-negative infective endocarditis. J Infect Chemother
2015;21(12):882–4.
[7] Fournier P, Thuny F, Richet H, Lepidi H, Casalta J, Arzouni J, et al. Comprehensive
diagnostic strategy for blood culture-negative endocarditis: a prospective study of
819 new cases. Clin Infect Dis 2010;51(2):131–40.
[8] Thoendel M, Jeraldo P, Greenwood-Quaintance KE, Chia N, Abdel MP, Steckelberg
JM, et al. A novel prosthetic joint infection pathogen, mycoplasma salivarium,
identified by metagenomic shotgun sequencing. Clin Infect Dis 2017;65(2):332–5.
[9] Hasman H, Saputra D, Sicheritz-Ponten T, Lund O, Svendsen CA, Frimodt-Møller N,
et al. Rapid whole-genome sequencing for detection and characterization of mi-
croorganisms directly from clinical samples. J Clin Microbiol 2014;52(1):139–46.
[10] Hilton SK, Castro-Nallar E, Pérez-Losada M, Toma I, McCaffrey TA, Hoffman EP,
et al. Metataxonomic and metagenomic approaches vs. culture-based techniques for
clinical pathology. Front Microbiol 2016;7.
[11] Zhou Y, Wylie KM, El Feghaly RE, Mihindukulasuriya KA, Elward A, Haslam DB,
et al. Metagenomic approach for identification of the pathogens associated with
S. Amrane, J.-C. Lagier
diarrhea in stool specimens. J Clin Microbiol 2016;54(2):368–75.
[12] Kirstahler P, Bjerrum SS, Friis-Møller A, la Cour M, Aarestrup FM, Westh H, et al.
Genomics-based identification of microorganisms in human ocular body fluid. Sci.
Rep. 2018;8(1):4126.
[13] Dubourg G, Fenollar F. Epidemiologic studies need asymptomatic controls. Clin.
Microbiol. Infect. Aug. 2015;21(8):e51–2.
[14] Bousbia S, Papazian L, Saux P, Forel JM, Auffray J-P, Martin C, et al. Repertoire of
intensive care unit pneumonia microbiota. PLoS One 2012;7(2):e32486.
[15] Pendleton KM, Erb-Downward JR, Bao Y, Branton WR, Falkowski NR, Newton DW,
Huffnagle GB, Dickson RP. Rapid pathogen identification in bacterial pneumonia
using real-time metagenomics. Am J Respir Crit Care Med 2017. p. rccm.201703-
0537LE.
[16] Abril MK, Barnett AS, Wegermann K, Fountain E, Strand A, Heyman BM, et al.
Diagnosis of capnocytophaga canimorsus sepsis by whole-genome next-generation
sequencing,. Open Forum Infect Dis 2016;3(3). p. ofw144.
[17] Salzberg SL, Breitwieser FP, Kumar A, Hao H, Burger P, Rodriguez FJ, et al. Next-
generation sequencing in neuropathologic diagnosis of infections of the nervous
system. Neurol – Neuroimmunol Neuroinflammation Aug. 2016;3(4):e251.
[18] Eckburg PB, Bik EM, Bernstein CN, Purdom E, Dethlefsen L, Sargent M, Gill SR,
Nelson KE, Relman DA. Diversity of the human intestinal microbial flora. Science
(80-) 2005;308(5728):1635–8.
[19] Andersen H, Connolly N, Bangar H, Staat M, Mortensen J, Deburger B, et al. Use of
shotgun metagenome sequencing to detect fecal colonization with multidrug-re-
sistant bacteria in children. J Clin Microbiol 2016;54(7):1804–13.
[20] Gyarmati P, Kjellander C, Aust C, Song Y, Öhrmalm L, Giske CG. Metagenomic
analysis of bloodstream infections in patients with acute leukemia and therapy-
induced neutropenia. Sci Rep 2016;6(1):23532.
[21] Schmidt K, Mwaigwisya S, Crossman LC, Doumith M, Munroe D, Pires C, et al.
Identification of bacterial pathogens and antimicrobial resistance directly from
clinical urines by nanopore-based metagenomic sequencing. J Antimicrob
Chemother 2017;72(1):104–14.
[22] Loman NJ, Constantinidou C, Christner M, Rohde H, Chan JZ-M, Quick J, et al. A
culture-independent sequence-based metagenomics approach to the investigation of
an outbreak of shiga-toxigenic Escherichia coli O104:H4. JAMA 2013;309(14):1502.
[23] Huang AD, Luo C, Pena-Gonzalez A, Weigand MR, Tarr CL, Konstantinidis KT.
Metagenomics of two severe foodborne outbreaks provides diagnostic signatures
and signs of coinfection not attainable by traditional methods. Appl Environ
Microbiol 2017;83(3):e02577–2616.
[24] Leake SL, Pagni M, Falquet L, Taroni F, Greub G. The salivary microbiome for
differentiating individuals: proof of principle. Microbes Infect 2016;18(6):399–405.
[25] Schmedes SE, Sajantila A, Budowle B. Expansion of microbial forensics. J Clin
Microbiol Aug. 2016;54(8):1964–74.
[26] Abayasekara LM, Perera J, Chandrasekharan V, Gnanam VS, Udunuwara NA,
Liyanage DS, Bulathsinhala NE, Adikary S, Aluthmuhandiram JVS, Thanaseelan CS,
Tharmakulasingam DP, Karunakaran T, Ilango J. Detection of bacterial pathogens
from clinical specimens using conventional microbial culture and 16S metage-
nomics: a comparative study. BMC Infect Dis 2017;17(1):631.
[27] Guo Y, Li S-H, Kuang Y-S, He J-R, Lu J-H, Luo B-J, et al. Effect of short-term room
temperature storage on the microbial community in infant fecal samples. Sci Rep
2016;6(1):26648.
[28] Ruppé E, Lazarevic V, Girard M, Mouton W, Ferry T, Laurent F, et al. Clinical
metagenomics of bone and joint infections: a proof of concept study. Sci Rep
2017;7(1):7718.
[29] Hugon P, Lagier J-C, Robert C, Lepolard C, Papazian L, Musso D, et al. Molecular
studies neglect apparently gram-negative populations in the human gut microbiota.
J Clin Microbiol 2013;51(10):3286–93.
[30] Wesolowska-Andersen A, Bahl M, Carvalho V, Kristiansen K, Sicheritz-Pontén T,
Gupta R, et al. Choice of bacterial DNA extraction method from fecal material in-
fluences community structure as evaluated by metagenomic analysis. Microbiome
2014;2(1):19.
[31] Brooks JP, Edwards DJ, Harwich MD, Rivera MC, Fettweis JM, Serrano MG, et al.
The truth about metagenomics: quantifying and counteracting bias in 16S rRNA
studies. BMC Microbiol 2015;15(1):66.
Human Microbiome Journal 9 (2018) 1–6
[32] Angelakis E, Bachar D, Henrissat B, Armougom F, Audoly G, Lagier J-C, et al.
Glycans affect DNA extraction and induce substantial differences in gut metage-
nomic studies. Sci Rep May 2016;6:26276.
[33] Thoendel M, Jeraldo PR, Greenwood-Quaintance KE, Yao JZ, Chia N, Hanssen AD,
et al. Comparison of microbial DNA enrichment tools for metagenomic whole
genome sequencing. J Microbiol Methods Aug. 2016;127:141–5.
[34] Ruppé E, Schrenzel J. Messages from the second International Conference on
Clinical Metagenomics (ICCMg2). Microbes Infect 2018;20(4):222–7.
[35] Brown AC, Bryant JM, Einer-Jensen K, Holdstock J, Houniet DT, Chan JZM, et al.
Rapid whole-genome sequencing of mycobacterium tuberculosis isolates directly
from clinical samples. J Clin Microbiol 2015;53(7):2230–7.
[36] Salter SJ, Cox MJ, Turek EM, Calus ST, Cookson WO, Moffatt MF, et al. Reagent and
laboratory contamination can critically impact sequence-based microbiome ana-
lyses. BMC Biol 2014;12(1):87.
[37] Ruppé E, Greub G, Schrenzel J. Messages from the first International Conference on
Clinical Metagenomics (ICCMg). Microbes Infect 2017.
[38] Schlaberg R, Chiu CY, Miller S, Procop GW, Weinstock G. Validation of metage-
nomic next-generation sequencing tests for universal pathogen detection. Arch
Pathol Lab Med 2017. p. arpa.2016-0539-RA.
[39] Lagier J-C, Million M, Hugon P, Armougom F, Raoult D. Human gut microbiota:
repertoire and variations. Front Cell Infect Microbiol 2012;2:136.
[40] Simner PJ, Miller S, Carroll KC. Understanding the promises and hurdles of meta-
genomic next-generation sequencing as a diagnostic tool for infectious diseases.
Clin Infect Dis 2017.
[41] Miller RR, Montoya V, Gardy JL, Patrick DM, Tang P. Metagenomics for pathogen
detection in public health. Genome Med 2013;5(9):81.
[42] Naccache SN, Federman S, Veeraraghavan N, Zaharia M, Lee D, Samayoa E, et al. A
cloud-compatible bioinformatics pipeline for ultrarapid pathogen identification
from next-generation sequencing of clinical samples. Genome Res
2014;24(7):1180–92.
[43] Forbes JD, Knox NC, Ronholm J, Pagotto F, Reimer A. Metagenomics: the next
culture-independent game changer. Front Microbiol 2017;8(7):1325–38.
[44] Stämmler F, Gläsner J, Hiergeist A, Holler E, Weber D, Oefner PJ, et al. Adjusting
microbiome profiles for differences in microbial load by spike-in bacteria.
Microbiome 2016;4(1):28.
[45] Deurenberg RH, Bathoorn E, Chlebowicz MA, Couto N, Ferdous M, García-Cobos S,
et al. Application of next generation sequencing in clinical microbiology and in-
fection prevention. J Biotechnol 2017;243:16–24.
[46] Afshinnekoo E, Chou C, Alexander N, Ahsanuddin S, Schuetz AN, Mason CE.
Precision metagenomics: rapid metagenomic analyses for infectious disease diag-
nostics and public health surveillance. J Biomol Tech 2017;28(1):40–5.
[47] Sboner A, Mu X, Greenbaum D, Auerbach RK, Gerstein MB. The real cost of se-
quencing: higher than you think!. Genome Biol May 2011;12(8):125.
[48] Muir P, Li S, Lou S, Wang D, Spakowicz DJ, Salichos L, et al. The real cost of
sequencing: scaling computation to keep pace with data generation. Genome Biol
2016;17(1):53.
[49] Goldberg B, Sichtig H, Geyer C, Ledeboer N, Weinstock GM. Making the leap from
research laboratory to clinic: challenges and opportunities for next-generation se-
quencing in infectious disease diagnostics. MBio 2015;6(6):e01888–1915.
[50] Nakamura S, Maeda N, Miron IM, Yoh M, Izutsu K, Kataoka C, et al. Metagenomic
diagnosis of bacterial infections. Emerg Infect Dis 2008;14(11):1784–6.
[51] Mongkolrattanothai K, Naccache SN, Bender JM, Samayoa E, Pham E, Yu G, Dien
Bard J, Miller S, Aldrovandi G, Chiu CY. Neurobrucellosis: unexpected answer from
metagenomic next-generation sequencing. J Pediatric Infect Dis Soc 2017. p.
piw066.
[52] Yao M, Zhou J, Zhu Y, Zhang Y, Lv X, Sun R, et al. Detection of listeria mono-
cytogenes in CSF from three patients with meningoencephalitis by next-generation
sequencing. J Clin Neurol 2016;12(4):446.
[53] Ye M, Wei W, Yang Z, Li Y, Cheng S, Wang K, et al. Rapid diagnosis of
Propionibacterium acnes infection in patient with hyperpyrexia after hematopoietic
stem cell transplantation by next-generation sequencing: a case report. BMC Infect
Dis 2015;16(1):5.