Jiz 286

The Journal of Infectious Diseases
SUPPLEMENT ARTICLE
Next Generation Sequencing and Bioinformatics

Methodologies for Infectious Disease Research and Public
Health: Approaches, Applications, and Considerations for
Development of Laboratory Capacity
Irina Maljkovic Berry,1, Melanie C. Melendrez,2 Kimberly A. Bishop-Lilly,3 Wiriya Rutvisuttinunt,1 Simon Pollett,1,4 Eldin Talundzic,5 Lindsay Morton,6 and
Downloaded from https://academic.oup.com/jid/article/221/Supplement_3/S292/5586940 by guest on 04 April 2022

Richard G. Jarman1
1
Viral Diseases Branch, Walter Reed Army Institute of Research, Silver Spring, Maryland; 2Department of Biological Sciences, St Cloud State University, Minnesota; 3Genomics and Bioinformatics
Department, Biological Defense Research Directorate, Naval Medical Research Center-Frederick, Fort Detrick, Maryland; 4Department of Preventive Medicine and Biostatistics, Uniformed Services
University of the Health Sciences, Bethesda, Maryland; 5Division of Parasitic Diseases and Malaria, Center for Global Health, Centers for Disease Control and Prevention, Atlanta, Georgia; and
6
Global Emerging Infections Surveillance, Armed Forces Health Surveillance Branch, Silver Spring, Maryland
Next generation sequencing (NGS) combined with bioinformatics has successfully been used in a vast array of analyses for infectious
disease research of public health relevance. For instance, NGS and bioinformatics approaches have been used to identify outbreak
origins, track transmissions, investigate epidemic dynamics, determine etiological agents of a disease, and discover novel human
pathogens. However, implementation of high-quality NGS and bioinformatics in research and public health laboratories can be
challenging. These challenges mainly include the choice of the sequencing platform and the sequencing approach, the choice of bi-
oinformatics methodologies, access to the appropriate computation and information technology infrastructure, and recruiting and
retaining personnel with the specialized skills and experience in this field. In this review, we summarize the most common NGS and
bioinformatics workflows in the context of infectious disease genomic surveillance and pathogen discovery, and highlight the main
challenges and considerations for setting up an NGS and bioinformatics-focused infectious disease research public health labora-
tory. We describe the most commonly used sequencing platforms and review their strengths and weaknesses. We review sequencing
approaches that have been used for various pathogens and study questions, as well as the most common difficulties associated with
these approaches that should be considered when implementing in a public health or research setting. In addition, we provide a re-
view of some common bioinformatics tools and procedures used for pathogen discovery and genome assembly, along with the most
common challenges and solutions. Finally, we summarize the bioinformatics of advanced viral, bacterial, and parasite pathogen
characterization, including types of study questions that can be answered when utilizing NGS and bioinformatics.
Keywords. bioinformatics; public health; infectious disease; capacity building; pathogen discovery; genome assembly;
metagenomics; advanced characterization; next generation sequencing; high-throughput sequencing.
Next-generation sequencing (NGS) technology, or high- origins of pathogen emergence [11–17]. This, coupled with
throughput sequencing, combined with bioinformatics has be- improvements in sequencing error rates and simpler labora-
come a powerful tool for detection, identification, and analyses tory approaches, and the decreasing costs of NGS and com-
of human pathogens. Its advantages over conventional methods putational requirements, has made NGS and bioinformatics a
are many, as sequences produced can be used for more accurate more achievable and increasingly desirable feature of research
detection and characterization of pathogens, screening for pres- and public health laboratories around the world. However,
ence of resistance mutations/genes, vaccine escape variants, re- NGS is powerful but complex and nuanced, requiring signifi-
combination or reassortment, and virulence and pathogenicity cant experience and expertise for production of accurate and
factors [1–10]. The assembly and analyses of pathogen genomes informative results. In addition, implementation of NGS and
can shed light on pathogen spread, contact tracing, dynamics bioinformatics methods as routine surveillance and tracking
of epidemics, and even possible sources, times, and geographic tools necessitates specialized information technology (IT) and
quality management systems that can meet the goals of public
health laboratories.
Correspondence: Irina Maljkovic Berry, PhD, Walter Reed Army Institute of Research, 503
Many challenges exist in setting up a high-quality NGS and
Robert Grant Ave, 20910, Silver Spring, MD ([email protected]). bioinformatics laboratory capacity, such as choosing the right
The Journal of Infectious Diseases® 2020;221(S3):S292–307 sequencing platform, wet lab sequencing method, bioinfor-
Published by Oxford University Press for the Infectious Diseases Society of America 2019.
This work is written by (a) US Government employee(s) and is in the public domain in the US.
matics analyses tools, personnel with the right kind of skills
DOI: 10.1093/infdis/jiz286 and experience, and computational and IT infrastructure to
S292 • JID 2020:221 (Suppl 3) • Maljkovic Berry et al

support the analyses of large amounts of data produced by NGS have high-throughput but also a high single-pass sequence error
and bioinformatics. While some standards and guidelines have rate of 14%, which can be decreased by conversion to circular
been developed, these may not be broadly applicable to all in- consensus sequence reads to 2% [19, 20]. Oxford Nanopore
fectious disease and public health laboratories [18]. Thus, the released its first flash drive-sized single molecule sequencer, the
small nuances in the nucleic acid extraction and sequencing MinION, in 2014. Marketing of this product inspired a large
approaches, combined with the different sequencing platform user following, as the company allowed the scientific com-
capabilities, become important factors in capacity building. munity to dictate what needed to be developed for the unit in
Many sequencing and wet lab approaches exist, and it is impor- terms of hardware and software. Software developments fo-
tant to recognize their benefits and weaknesses. Furthermore, cused on correcting for the higher error rates (eg, 13%–20%)
with the myriad of bioinformatics tools available today, and of this platform [20–22]. Oxford Nanopore also released the
with the rapid growth and constant change in this field, it high-throughput PromethION and GridION platforms, which
becomes difficult to standardize analyses across laboratories allowed for parallelization of sequencing by stacking of multiple

and teams. Thus, the choice of bioinformatics tools and analyses flow cells.
becomes important to consider in NGS and bioinformatics lab- Selection of a platform depends heavily on a laboratory’s
oratory capacity development. Additionally, bioinformatics research objectives (Table 1). Generally speaking, whole-
will require adequate computational and IT infrastructure, in- genome sequencing of bacteria or viruses has been successful
cluding networks and storage systems, as well as personnel with on smaller targeted platforms, such as the MiSeq, NextSeq,
specialized knowledge and experience with analysis pipelines, or Ion Torrent [11, 13, 14, 23]. Some genome sequencing
wet lab methods, sequencing platform characteristics, and ide- applications, such as highly repetitive bacterial genome
ally familiarity with the pathogens of interest. All these become structures or bacteria with modular plasmid structures, have
important to consider during NGS and bioinformatics capacity required platforms that are more robust and can provide ei-
building in a research or public health setting. ther longer sequence reads (PacBio) or a moderate read length
In this review we summarize important factors and and greater depth (HiSeq, NovaSeq) [24]. Minor variant and
considerations for setting up a high-quality NGS and bioinfor- single nucleotide polymorphism (SNP) detection studies
matics focused infectious disease research and public health involving larger genomes and/or highly diverse organisms
laboratory, in settings with both limited, as well as substan- have been better served by higher throughput platforms
tial, resources. We focus on the most common methods in (HiSeq, NovaSeq) [16]. In addition to research objectives, the
sequencing and bioinformatics, and we describe some com- choice of platform depends on personnel experience and skill
monly faced challenges during this capability development. We levels. Ion Torrent is user friendly and simple in the labora-
provide recommendations that could enable a more streamlined tory, but the challenges of data analytics require personnel
process of NGS and bioinformatics laboratory implementation. with appropriate bioinformatics background. In comparison,
the MiSeq requires more training, but offers data storage and
NGS TECHNOLOGIES AND PLATFORMS platform bioinformatics support with a user-friendly graph-
Since the introduction of NGS technology in 2005, the number ical interface. A key attribute that needs to be considered is
of high-throughput sequencing platforms with different costs, the sequencing platform connectivity and training expertise
chemistries, capacities, and applications has increased dramat- and availability, which is a factor in many countries in Africa,
ically. Illumina alone offers many platforms, from sizes ame- South America, Central America, and Asia. These laboratories
nable to small laboratories/classrooms and clinical laboratories, not only have to consider the availability of skilled labora-
to large high-throughput sequencing centers. In addition tory and bioinformatics personnel, but also the availability
to its most versatile platform to date, the MiSeq, Illumina of reagents, ease of installing, running, and maintenance of a
has launched the GAIIx, MiSeqDx (the first Food and Drug sequencing platform, including the setup of IT infrastructure,
Administration-regulated, in vitro diagnostic testing plat- data storage, and power backups to support the instrument
form), NextSeq, NovaSeq, MiniSeq, and iSeq, to accommo- [25, 26]. Importantly, IT infrastructure and computational
date different cost levels and capacity needs. Meanwhile, the requirements have to be considered in the total cost of these
Ion Torrent/Ion S5 platform (acquired by Life Technologies), systems for all laboratories, but more so in developing nations
while having higher error rates as compared to the Illumina where availability is considerably scarcer.
systems, has continued to be utilized due to its affordability So far, the Illumina MiSeq system has proven to be the
and ease of use. In addition, 2 companies have pioneered the most commonly used platform for infectious disease research,
single-molecule sequencing market with platforms that offer pathogen surveillance, and pathogen discovery in research
ultralong reads. Pacific Biosciences (PacBio) was the first with and public health [3, 11, 27–29] (Table 2). The instrument is
the PacBioRS/RSII, and the newest platform, the Sequel, which compact enough to fit on a laboratory bench, has a fast run-
can obtain average read lengths of 10 kb. The PacBio platforms time as compared to other similar platforms, and has a strong
NGS and Bioinformatics Methodologies • JID 2020:221 (Suppl 3) • S293

Table 1. Examples of Currently Supported Sequencing Platforms and Their Advantages/Disadvantages
Sequencing Observed
Platform/Year Final Error Computational
Released Applications Rate, % Runtime Resources Advantages Disadvantages
Sanger ABI A ~0.1 20 min–48 None needed High quality, long reads, low Low throughput, high cost, substitution errors,
3730xl/ 2002 h cost for small studies sequenced material has to be pure to produce
good-quality sequence data
PacBio RSII/ V, M, E, HE, ~13 one 0.5–4 h Cluster needed Used in methylome research Indels, large lab footprint, expensive
2010 RT, CP, EP pass; <1
multipass
Ion Torrent/ A, V, M, E, ~1 4–7 h Powerful desktop Lower cost instrument, up- Higher error rate with homopolymer issues, more
PGM318/ HE, D, (chip) or cluster gradable, simple machine hands-on time, fewer overall reads, higher cost/
2010 PS MB, indel issues
ABI SOLiD A, V, M, E, ~5 one 6–10 Cluster needed Independent flow cell lanes, Longevity of platform, shorter reads, more gaps

5500xl/ HE, RT, pass; <0.1 days high accuracy, ability to rescue in assemblies, less even data distribution, high
Wildfire/ ML, SV, multipass failed sequencing cycles capital cost
2010 PS
Illumina A, V, M, E, ~0.1 4–55 h Cloud or Moderate cost/instrument Substitution errors, as the sequencing reaction
MiSeq/ 2011 HE, RT, server onsite and runs, low cost/MB, proceeds, the error rate increases
SV, D, PS (Basespace) fast run time, versatile
Oxford A, V, M, E, 4–20 1 Laptop or MinIT Longest individual reads, Lower throughput than other machines, low
Nanopore HE, RT, min– (i)a accessible user single-read pass accuracy, deletions
MinION/ SV, ME, 48 h community, portable
2014 EP, PS USB size
Illumina A, V, M, E, HE, ~0.1 12–30 h Included/cloud or High sequence yield Expensive, high concentrations of DNA, requires
NextSeq RT, ML, ME, SV, server onsite potential, easy to use, high indexing capabilities, issues with substitution
500/ 2015 C, MT, D, PS expandable errors
As the sequencing reaction proceeds, the error
rate increases
Illumina V, M, E, HE, ~0.1 13–44 h Server for analysis High sequence yield Expensive, high concentrations of DNA, requires
NovaSeq RT, ML, and storage potential, no application high indexing capabilities, issues with substitution
6000/ 2017 ME, SV, restrictions errors
C, MT As the sequencing reaction proceeds, the error
rate increases, higher frequency of duplicate
reads
PacBio Sequel/ V, M, E, HE, ~13 one 30 min– Cluster Fast, desktop sized Moderate throughput, expensive
2016 RT, CP, EP pass; <1 20 h recommended instrument, long reads
multipass
Oxford A, V, M, E, 4–20 one up to Cluster Higher output than MinION, Low single-read pass accuracy, issues with
Nanopore HE, RT, SV, pass; <1 64 h recommended longest individual reads, deletions
PromethION/ ME, EP, PS multipass accessible user community,
2018 scalable
Illumina A, V, M, ~0.1 9–17.5 h None needed Lower cost, faster sample Substitution errors, as the sequencing reaction
iSeq 100/ targeted-RT, PS, preparation, minimizes po- proceeds the error rate increases, prone to bar-
2018 D (planned) tential user error or need for code hopping, cannot be used at high altitudes
corrective maintenance, sin-
gle-use cartridges so upgrades
are in consumables only
Abbreviations: A, amplicon sequencing; C, ChIP-seq; CP, complex population sequencing; D, diagnostics; E, eukaryotic genome; EP, epigenetics; HE, human/exome genomics; M, microbial
genome; MB, mega base; ME, metagenomics; ML, methylation studies; MT, metatranscriptomics; PS, pathogen surveillance; RT, RNAseq/transcriptomics; SV, single nucleotide polymor-
phism/variation studies; V, viral genome.
a
https://nanoporetech.com/products/minit.
user support community. However, the field is increasingly software stabilization, the MinION may be an excellent ad-
demanding sequencing closer to the disease, and while the dition to the arsenal of current sequencing technologies for
MinION provides portability, the high error rates and the routine surveillance, especially in smaller laboratories with
continuous chemistry and software changes make this plat- limited resources. For instance, the MinION was successfully
form difficult to implement in routine public health surveil- used in the ZiBRA project for real-time Zika virus surveillance
lance laboratories. If used in a public health laboratory, the of mosquitoes and humans in Brazil, and in Guinea to perform
results may need to be validated with a different platform [15]. real-time surveillance during the ongoing Ebola outbreak [12,
However, with further improvements of this technology, like 36]. For the Ebola outbreak, results were obtained within 24
the most recent advances in laboratory-independent sample hours of receiving a positive sample, and sequencing on the in-
extraction and library preparation, portable computational strument took as little as 15 minutes, highlighting the potential
support (MinIT), and with additional error reduction and of the MinION for a rapid response to an ongoing outbreak.

Table 2. Sequencing Platforms, Sequencing and Bioinformatics Approaches, and Published Examples of their Applications
Sequencing Organism or
Platform Sample Type Goal Wet Lab Design Software or Pipeline Benefits Achieved
Illumina MiSeq Dengue virus Surveillance, Direct sample and ngs_mapper pipeline, PhyML, Rapid surveillance design, pre-
transmission viral isolate amplicon BEAST diction of burden of disease,
sequencing intercountry movement [11, 13]
Illumina MiSeq Enterovirus A71 Surveillance Viral isolate amplicon CLC Genomic Workbench, Circulation of genogroup C, new
and random/unbiased ClustalW, BLAST genogroup E, genetic exchanges,
sequencing emergence of pathogenic
lineages, recombination [30]
Ion Torrent, Zika virus Outbreak Viral isolate amplicon Mira, Geneious, MAFFT, Path-O- Clarification of cross-border viral
GS-FLX/ sequencing Gen, BEAST spread dynamics, hypothesis
GS-Junior testing for viral origin, gene var-
iant detection [14]

MinION Zika virus Outbreak Direct sample amplicon Metrichor, Nanonet, BWA MEM, Transmission reconstruction, con-
sequencing python scripts, zibraproject Zika tinental spread inference, variant
pipeline detection [15]
Illumina Zika virus Surveillance, Direct sample probe enrich- Trimmomatic, Novoalign, SAMtools, Reconstructing viral transmission
MiSeq viral introductions ment, amplicon sequencing Snakemake, Geneious, Cutaddapt, and introductions [29]
Prinseq-lite, Bowtie2, Picard tools,
custom scripts, PhyML, BEAST
Illumina CSF Diagnosis, Direct sample random/unbi- modified SURPI pipeline (SURPI+) Discovery of etiological agent,
HiSeq pathogen ased sequencing neurobrucellosis; resulted in
discovery CLIA-certified SOP validation [31]
Illumina Nasopharyngeal Pathogen Direct sample random/unbi- Taxonomer, Geneious Detection of respiratory viruses,
HiSeq swabs discovery ased sequencing strain typing, detected viruses
not found by the FDA-cleared
respiratory viral panel [32]
MinION, Fluid from lungs Pathogen dis- Bacterial isolate and direct Mothur, R Identification of pathogens in
Illumina MiSeq covery sample amplicon sequencing lungs of patients with pneumonia
and sepsis [33, 34]
MinION Enriched urine Diagnostics, Direct sample random/ Poretools, BLAST, CARD-LAST, Pathogen identification and
pathogen unbiased sequencing custom scripts, SAMtools, WIMP resistance gene identification in
discovery Metrichor application, Kraken, 4 h (sample to result; similar to
ARMA application PCR) [1]
MinION Ebola virus Outbreak, Direct sample amplicon MinKNOW, Metrichor CLI, Deployment of MinION se-
surveillance sequencing nanopolish, MarginAlign, RaxML quencer and analysis in the field;
low cost [12]
Illumina Ebola virus Surveillance Direct sample negative FastQC, Trimmomatic, BMTagger, Observation of rapid inter/
HiSeq PacBio enrichment, PRINSEQ, MetaVelvet, BLASTn, intrahost variant accumulation,
RS random/unbiased MEGAN, Picard, Lastal, Trinity, characterization of viral trans-
sequencing custom pipeline, novoalign, GATK, mission patterns, no evidence of
Geneious, MAFFT, MUSCLE, RDP3, additional zoonotic sources, SNP
snpEff, RaxML, BEAST, V-Phaser2 identification for monitoring [16]
Ion Torrent Legionella Outbreak Bacterial isolate random/ CLC Genomic Workbench, Real-time Legionella outbreak ge-
PGM pneumophila unbiased sequencing BioNumerics nomic surveillance, SNP analysis
and MLST profiling; identification
of links between environmental
and patient isolates [23]
Illumina MiSeq Vancomycin Outbreak Bacterial isolate random/ Newbler, PanSeq, Gegenees, Outbreak reconstruction, cor-
resistant investigation unbiased sequencing Geneious, ResFinder, Bowtie2, relation between antibiotic
Enterococcus SAMtools, bedtools susceptibilities and gene content,
faecium SNP analysis [35]
Illumina HiSeq, Bas-Congo virus Outbreak Direct sample, random/ PRICE de novo assembler, Novel virus discovery, outbreak
GS-FLX 454 investigation, unbiased sequencing Geneious, SOAP, BLAT, BLAST, surveillance, taxonomy [8]
pathogen MAFFT, Mr.Bayes, BEAST
discovery
Abbreviations: CLIA, Clinical Laboratory Improvement Amendments; CSF, cerebrospinal fluid; FDA, Food and Drug Administration; MLST, multilocus sequence typing; PCR, polymerase chain
reaction; SNP, single nucleotide polymorphism; SOP, standard operating procedure.
LABORATORY APPROACHES AND SEQUENCING etiologic agents, or for sequencing of bacterial isolates [23, 35].
METHODS In cases of suspected low pathogen abundance or detection of
In addition to the variety of sequencing platforms on the market, pathogens in samples containing high host nucleic acid content,
there are a variety of applications within NGS to consider. pathogen enrichment or host depletion procedures should be
For instance, metagenomics and unbiased sequencing may considered. Specific pathogen genomic amplification may be
be useful to broaden pathogen detection, elucidate unknown applied for samples where the agent is known, as is common

in outbreaks and epidemics. Kit selection and error rates There are many challenges involved in metagenomic
during amplification are also important considerations [37]. sequencing for infectious disease, and there are many choices
Developing laboratory capability thus necessitates knowledge to consider, which will usually depend on the study ques-
of the various applications, in order to inform decisions such tion. Sequencing of total nucleic acid (RNA and DNA) is the
as in which approaches to initially invest and how to maximize only approach that allows detection of all domains of life.
the sequencing success. A workflow of the most common labo- However, direct environmental samples contain nucleic acids
ratory approaches is illustrated in Figure 1. from many different sources (host, normal microbial flora,
fragmented DNA from organisms not necessarily present
Metagenomics and Pathogen Discovery: When and How to Look for the in the sample, and other potential pathogens), their amount
Signal in the Noise depending on the sample origin. For instance, nasopharyn-
Metagenomics is the study of an entire community of organisms geal and stool samples can be expected to have a great amount
via analysis of sequenced genomes and/or transcripts from an commensals and/or opportunistic pathogens, while samples

environmental sample (ie, soil, human, animal, water) and typ- like CSF and blood are expected to be cleaner (Figure 1). The
ically results in detection of organisms from all domains of amount of pathogen nucleic acid in more noisy samples is
life. In surveillance and diagnostics, this approach is usually usually overwhelmed by DNA/RNA from these background
undertaken when other more directed assays such as poly- organisms, and the signal of the pathogen may be too low for
merase chain reaction (PCR) fail. These assays may fail because assembly of a useful length of its genome, or even to detect the
of emergence of a novel pathogen, genetic evolution of an ex- pathogen. Thus, although using total nucleic acid extraction is
isting pathogen, or poor assay design. In pathogen discovery, most comprehensive, selecting DNA or RNA extraction only
the most commonly used samples for metagenomic sequencing may reduce the amount of the background noise, and may
have been blood, stool, cerebrospinal fluid (CSF), urine, or naso- benefit in higher yield of the pathogen genome in question.
pharyngeal swabs, where investigators have attempted to iden- In instances where there is a strong suspicion of the pathogen
tify the etiological agent responsible for an infection or other genome composition, a more specific approach (RNA or DNA
clinical syndrome [1, 8, 28, 32, 33]. For instance, following a extraction only) would be preferred. Working with clinical
deadly 2009 outbreak of acute hemorrhagic fever in Democratic and public health officials may provide additional sample in-
Republic of Congo, Grard et al [8] used 454 Roche sequencing formation that can aid in selection of an appropriate extrac-
to assemble and characterize the genome of a novel rhabdovirus tion and sequencing method.
(Bas-Congo virus, or BASV) in one of the patient’s acute serum Other choices that may affect the success of metagenomic
samples. Pathogen discovery can also be performed on samples sequencing include library preparation methods, nuances of li-
collected from environments (eg, vectors and animals) that brary quantitation, quality control and normalization, whether
have previously been associated with spillover of pathogens to and how much PhiX (sequencing control) to spike into an
humans, causing outbreaks, epidemics, and pandemics. With a Illumina run, calculation of desired coverage level, and use of
metagenomic approach, these environments have been screened adequate controls. The controls may include a positive control,
for presence of known or yet undiscovered pathogens, although an additional internal control (eg, spiked DNA or other known
the effectiveness of such an approach has been questioned [6, pathogen), and a negative control (water sample), and are es-
38–41]. Generally speaking, metagenomic sequencing is most pecially imperative in pathogen sequencing. These can be used
useful and cost efficient for pathogen discovery when at least to identify contamination and cross-contamination, and for
1 of the following criteria are met: (1) the identification of the downstream background noise removal. In fact, contamination
organism is not sufficient (one desires to go beyond discovery is a very common problem in metagenomic sequencing and it
to produce data for genomic characterization), (2) a coinfection occurs both in skilled and less experienced laboratories [37, 42].
is suspected, (3) other simpler assays are ineffective or will take
an inordinate amount of time, and/or (4) the goal is to screen Enrichment and Targeted Sequencing: Increasing the Odds of Success
environmental samples for previously undescribed or diver- Because of high levels of background noise in metagenomic
gent pathogens. For instance, when it is strongly suspected that sequencing, several target enrichment procedures have been
the etiologic agent is an existing pathogen with a divergent ge- developed that aim to increase the probability of capturing
nome sequence that evades a nucleic acid-based assay, and the pathogen-derived transcripts and/or genomes [43, 44]. Prior
assay would be redesigned if the divergent sequence could be knowledge of the pathogen genomic background can be used
obtained, then metagenomic sequencing is a suitable choice. to choose an appropriate enrichment technique and amplify the
Conversely, if simpler, more directed assays will suffice, or if sequence of interest. In general, there are 2 main approaches
identification of the specific agent will not result in any further that can be used to increase the amount of pathogen signal in a
actions being taken, such as downstream genome analyses, then sample: negative selection and positive enrichment.
the cost and effort involved in metagenomic sequencing and Negative selection (background depletion or subtraction)
analysis are likely not warranted. targets and eliminates the host and microbiome genomic

Sample Acquisition
Blood Urine Stool
Less background noise More background noise Sample
Serum CSF Swab Types
RNA/DNA Extraction
RNeasy PowerSoil Total RNeasy Minikit MagMAX Viral Viral RNA Minikit
RNA kit (Qiagen) Isolation kit (Qiagen) Commercially
(Qiagen) (ABI) available kits
Optimization strategies
Positive
Targeted DNAseq Library Negative selection

enrichment
Amplification: semi- prep with minimal- (ie, nucleases, RNA
(specific or pan-
random or specific bias fragmentation depletion, CRISPR)
viral probes)
QC: Library quality & quantity
Controls: spiked DNA, control pathogen,

High-Throughput Sequencing
+ PhiX, negative control (water)
Sub ources
res
es
d
stan
Limite
Wet Lab Workflow Considerations:

resourc
• Study design/hypothesis
tial
• Study goal, detection or genome assembly

• Number of samples
• Types of controls needed
Illumina NextSeq, • Sequencing depth required
Illumina MiSeq, iSeq
Novaseq • Sample matrix for extraction
Ion Torrent PGM
PacBio Sequel • QC approach
Oxford Nanopore
Oxford Nanopore • Laboratory and informatics resources
MinION
PromethION
Figure 1. Schematic diagram of common sequencing laboratory workflows and approaches. Abbreviations: CRISPR, clustered regularly interspaced short palindromic
repeats; CSF, cerebrospinal fluid; QC, quality control.
background, while aiming to preserve the nucleic acid derived removing background noise. This is commonly done through
from the pathogens of interest. Degradation of genomic back- hybridization-based target capture by probes, which are used to
ground can be performed through broad-spectrum digestion with pull out nucleic acid of interest for downstream amplification and
nucleases, such as DNase I for DNA background, or by removing sequencing. Probe-based enrichment has been used to allow for
abundant RNA species (rRNA, mtRNA, globin mRNA) using detection of viral genomes in Ebola virus outbreaks, Zika virus
sequence-specific RNA depletion kits [16, 45, 46]. Gu et al [47] epidemics and respiratory virus surveillance [29, 45, 49, 50]. Pan-
used clustered regularly interspaced short palindromic repeats viral probes have been shown to successfully identify diverse types
(CRISPR) approach to target and deplete human mitochondrial of pathogens in different clinical fluid and respiratory samples,
rRNA in clinical CSF samples, resulting in improved read cov- and have been used for sequencing and characterization of novel
erage of meningitis and encephalitis-associated pathogens in viruses [51–54]. In a precision public health surveillance approach,
those samples. Generally, however, subtraction approaches lead Cummings et al [52] used pan-viral probe capture to enrich
to a certain degree of loss of the targeted pathogen genome, as pathogens in samples from patients with influenza-negative se-
poor recovery may occur during the cleanup and additional vere acute respiratory infections (SARI). This approach resulted in
enrichment steps [48]. These approaches may thus not ini- identification of an unrecognized outbreak of measles-associated
tially be suitable for less experienced laboratories, or should be SARI, as well as detection of SARI associated with a novel
accompanied by an additional alternative approach. picobirnavirus. Pan-viral probes can also be used for preemptive
A simpler approach resulting in less loss of target is positive en- screening of environmental samples (of vector and animal origins)
richment, which is used to increase pathogen signal rather than for existence of emerging and even novel pathogen threats, thereby

complementing the conventional metagenomics approaches [7]. pathogen detection. These packages range from web-based or
However, the probe approach includes extra hybridization and commercially available software that is easy to use but the result
cleanup steps, requiring higher sample input, increasing the risk of accuracy relies on default parameters chosen by the software
losing the target, and increasing the cost and hands-on time. designer, to command line tools that allow for customization
Circumventing material loss that occurs in positive enrichment and potentially more-targeted results, assuming that bioinfor-
or negative selection can be achieved by the use of semirandom or matic expertise is available. While there have been efforts to
specific primers for direct pathogen genome amplification. PCR standardize workflows and provide guidance on best analyses
amplification using pathogen-specific primers has been success- practices, there is no consensus for the development and imple-
fully used for sequencing of pathogens from known outbreaks mentation of any specific bioinformatics workflows. Therefore,
and epidemics [3, 12, 13, 15, 29]. For example, pathogen-specific these are often developed in-house and customized based on
primers were utilized for amplicon sequencing of dengue viruses the needs of laboratories, making further standardization more
in acute febrile patients from Ecuador and Thailand, Plasmodium challenging. For laboratories with more-limited infrastruc-

in patients from Gabon, and influenza virus in gulls from Iceland ture and personnel expertise, the most user-friendly options
[11, 13, 55, 56]. It is efficient and low cost, removing almost all for pathogen discovery may be the Taxonomer, EDGE, or
sample background and amplifying only the genomic regions of Pathosphere pipelines, preferably installed on a server [58–60].
interest, thereby providing higher coverage for downstream path- However, if the expertise and larger computational support can
ogen analyses. However, a few mismatches between primers and be developed, other more-specific pipelines can be considered,
the targeted genome can result in failure to generate amplicons so that the processes can be tailored to the pathogen and/or a
of interest and affect the assembly of a full genome. In cases of specific problem that needs to be addressed (Table 2).
divergent pathogen genomes, laboratories have utilized random In general, regardless of the choice of software or pipeline for
sequencing amplification to assess the genomic composition of detection and identification of pathogens in a sample, there are
the pathogen and then generate pathogen-specific primers to im- common steps that can be followed for a successful analysis.
prove the quality and accuracy of the final genome assembly. The The raw data from a sequencing platform is usually cleaned,
use of DNASeq library preparation, with a minimal-bias fragmen- trimmed, and filtered to remove low-quality and duplicate
tation step and regularly updated primer sets, has been observed reads [24]. Removal of the host genome/transcriptome reads is
to consistently produce excellent sequence data across many dif- performed to decrease background noise (eg, host and environ-
ferent sample types [3, 11, 14, 30, 57]. mental reads) and increase the frequency of pathogen reads [8].
This step will also decrease downstream analysis time. Further
BIOINFORMATICS ANALYSES background noise removal is achieved by mapping of sample
Whereas the sequencing itself has been made widely accessible reads to the reads from the negative control to ensure elimi-
and more user friendly, the data analysis and interpretation that nation of any contaminating reads, such as those associated
follows still requires specialized bioinformatics expertise and ap- with the reagents or sampling storage medium. The remaining
propriate computational and IT resources. Routine and efficient reads are usually assembled de novo (described below), to pro-
processing and storage of gigabases of sequence data, which can duce long stretches of sequences (contigs) [24, 61]. Specifically
be produced by even a benchtop sequencer, will require an invest- for sequencing platforms that produce short reads, this step
ment in software and expensive hardware for networking, storage, ensures reliability of results and increased accuracy of down-
and data analyses. These costs can be minimized by using cloud stream pathogen identification. Taxonomic identification of
solutions and services, where possible. Computational infrastruc- the resulting contigs is performed by matching them to the
ture should be tailored to the specific laboratory needs, as should genomes and sequences in nucleotide or protein databases; for
advanced bioinformatics training of personnel. Importantly, only this, various versions of BLAST are most commonly used [1,
standardized and validated bioinformatics analysis tools should 8, 32, 61, 62]. Often, these databases are downloaded locally to
be incorporated in routine analysis workflows that are part of a improve processing time.
quality management and assurance system. Where appropriate, One of the more challenging steps of metagenomics and
automation of these analysis processes should be considered in pathogen discovery analyses is the interpretation of results.
an effort to improve overall turn-around time of results and re- Less-experienced laboratories may run into difficulties in
duce overall errors and costs. A workflow of the most common identifying false-negative and false-positive calls, including dis-
bioinformatics approaches and tools is summarized in Figure 2. crimination between background, contamination, commensals,
and true-positive pathogens. This is especially challenging
Pathogen Detection and Taxonomic Identification: What is Real and What when pathogen content in the sample is low, such as from
Is Not? samples taken in the beginning or the end of an acute infection.
Numerous software packages and workflows have been de- Thus, often times, accuracy of pathogen discovery becomes
veloped to facilitate metagenomic analysis and specifically a critical balance between the time of sampling, sample type,

High-Throughput Sequencing + Controls: spiked DNA, control pathogen,
PhiX, water sample
Quality Control and Preprocessing*
PhiX control Quality score and End trimming Read output Host Control Background negative
analysis duplicate read removal by quality statistics removal pathogen results control removal
Pathogen Discovery Genome Construction Advanced Analysis
2 De novo assembly, contig 1 2 Gene calling/Annotation, alignment,

identification, Genome assembly
antibiotic resistance gene identification
internal/external databases,
taxonomic classification 2 3
Manual Genome Curation Variant/SNP detection, phylogenetics, Analyses

3 MLST, population dynamics,
Results interpretation recombination
1 Taxonomer, EDGE, Pathosphere, 1 1

Geneious, CLC Genomics Geneious, MEGA, Nextstrain, RAST
SURPI+CLC Genomics Workbench Workbench, DNAStar, Mauve
3 Tools
2 3 BEAST, Mr. Bayes, PhyML, RaxML,
VirusSeeker, command line BLAST BWA, ngs_mapper, SOAPdenovo,
RDP4, ShortBRED
Bowtie2, MetaRay, Trinity, Bandage
*Computational requirements will depend on sequencing platform data output size.
1 Analyses and tools tailored to less experienced personnel or laboratories with more Bioinformatics workflow considerations :
limited computational resources • Available computational infrastructure
• Pipeline or analysis program requirements
2 Analyses and tools that may require specific additional training and/or custom standard-
• Pipeline output inspection/verification
operating procedures when used by less experienced personnel
• Reproducibility of computational outputs
3 Analyses and tools requiring more skilled/experienced personnel (command line, bash
scripting, interpretation) and/or higher computational infrastructure
Figure 2. Bioinformatics workflow and considerations for sequence analysis. Nondashed boxes describe analyses types and dashed boxes describe tools that can be used
for these analyses. Abbreviations: MLST, multilocus sequence typing; SNP, single nucleotide polymorphism.
depth and comprehensiveness of sequencing, technique effi- and long-read platforms and sequencing approaches. If read
ciency and analysis workflow robustness, data interpretation, quality and depth requirements are not met, the consensus se-
as well as clinical (symptoms, epidemiological, environmental quence should not be considered. Usually, such incomplete/
context) and pathogen biological insights. In a case report from gapped genomes are filled by additional sequencing [4, 15]. The
Mongkolrattanothai et al [31] an 11-year-old patient with head- quality of the assembly will also be affected by the assembly al-
ache, back pain, and nausea went through several diagnoses gorithm used. Many genome assembly and consensus calling
including Epstein-Barr virus, human herpesvirus 7, residual algorithms exist, and they vary greatly in their complexity, ac-
complications from a recent Salmonella infection, and puta- curacy, speed, and flexibility (Table 2). In general, 2 main ge-
tive tuberculosis disease. Finally, metagenomics sequencing re- nome assembly approaches exist, reference-based (mapping
vealed presence of Brucella, which was then further confirmed based) assembly and de novo assembly.
by both PCR and agglutinin test. The persistent symptoms, NGS Reference-based assembly is a very useful and accurate tool
and PCR testing showing Brucella, and positive confirmatory se- for assembly of known genomes, and can be especially bene-
rology allowed for a diagnosis of chronic neurobrucellosis [31]. ficial for laboratories with limited computational capacity or
Thus, the results of an NGS and bioinformatics metagenomic those with high sequencing throughput and/or when time is of
analysis, especially in diagnostic settings, should be confirmed the essence [11, 12, 29]. For instance, reference mapping was
with a different method, such as PCR. used for assembly of >500 dengue genomes from Thailand,
and combined with other data the results revealed that most
Genome Assembly: Putting the Pieces Together of dengue infections are obtained close to home [11]. During a
One of the main factors that plays a role in the accuracy and com- Legionella outbreak in a large Australian hospital, NGS and ge-
pleteness of a genome assembly is sequencing read quality and nome assembly through reference mapping was employed to, in
read depth of coverage. These aspects differ between short-read real time, distinguish the bacterial outbreak isolates [23]. In a

reference-based genome assembly, sample reads are mapped to genome construction. All sequencing platforms and assembly
a reference genome, and the reads are placed based on the best algorithms have limitations; errors are expected to be incorpo-
match and alignment to the reference [11, 13, 23]. Reference rated into the final results. Assembled genomes should always
mapping accuracy depends on the chosen reference and map- be checked to make sure they do not contain any unexpected
ping parameters. The reference genome must be closely related nucleotide insertions or deletions, as is often the case when
to the sequenced pathogen, in order for most of the reads to using the emPCR-based sequencing platforms (454 Roche
map accurately [12]. For some more variable organisms, such and Ion Torrent). These errors may be identified as unex-
as rapidly changing RNA viruses, references should remain pected reading frame shifts or stop codons. Artificial genome
within the context of the time and location of the sample col- substitutions may also occur due to sequencing or PCR error,
lected, to ensure genetic similarity. It may also be helpful to in- and due to use of degenerate primers during the amplification
itially map against a few representative reference strains to help process. When intrahost single nucleotide variants, or ambig-
identify the closest related one and improve overall mapping uous calls, are allowed in the consensus genome (as is often

of reads [24]. In addition, care should be taken with organisms the case for RNA viruses due to their existence as a population
that may have large insertions relative to the reference, as of variants within a single host), an aberrantly high number
these may go undetected if only reference mapping is used. of ambiguous positions may indicate problems during the ge-
BWA MEM and Bowtie2 remain some of the most widely used nome assembly process. Some of the most common parameters
programs for genome mapping [63, 64]. While the programs affecting variant calling and consensus validation process have
themselves are command-line based, they are also accessible been described in Jia et al [73].
in several commercial off-the-shelf tools such as Geneious and
CLC Genomics Workbench, open-source graphical or web- ADVANCED CHARACTERIZATION OF HUMAN
PATHOGENS
based tools like Galaxy Platform, Pathosphere, EDGE, and
other command-line based pipelines like ngs_mapper, GATK Once a high-quality genome is assembled, additional types of
and iMetAMOS [65, 66]. analysis can be performed. Some examples of the type of anal-
De novo genome assembly could be likened to putting to- ysis include reconstruction of pathogen transmission chains
gether a jigsaw puzzle without looking at the picture on the and outbreaks, and tracking of the selection and spread of re-
box; it relies upon connecting the sample reads to each other sistance, which can aid in epidemiological investigations [17,
using sequence match overlaps to generate longer sequences 74–76]. These types of analyses usually utilize more advanced
referred to as contigs. This process can thus be less accurate genomic and phylogenetic analysis tools, requiring additional
than reference mapping, and is usually also slower and com- expertise and computational infrastructure. High-performance
putationally more intensive. However, de novo assembly is computing environments are capable of data analysis paral-
useful when the pathogen is poorly understood or a good ref- lelization, thus increasing analysis throughput and decreasing
erence does not exist, or one suspects insertions, deletions, or analysis time of large datasets.
repetitions to be present [16]. In addition, de novo assembly is
commonly used for detection and assembly of novel pathogens, Advanced Bioinformatic Analyses of Viral Genome Sequence Data: Seeing
of pathogens that exhibit horizontal gene transfer, or assembly the Forest from the Trees
of nonchromosomal elements, such as bacterial plasmids Advanced characterization of viral genomes in surveillance
[24, 61]. For instance, in LaBreck et al [24] de novo assembly and research for public health spans an array of analyses. From
was used for characterization of a novel Staphylococcus au- simpler analyses, such as screening for phenotypically impor-
reus plasmid carrying a gene with decreased susceptibility to tant viral mutations which may confer influenza or HIV anti-
chlorhexidine, a biocide used in healthcare facilities. Some viral resistance, desktop tools such as MEGA and Geneious, and
commonly used tools for de novo assembly include Velvet, some web-based tools (ie, tools at the Influenza Virus Resource
Minia, ABySS, SOAPdenovo, and SPAdes [67–70]. If gaps and HIV Drug Resistance Database) have been used to rapidly
in the genome occur, in silico gap closure can be attempted estimate the frequency of these phenotypes in a given season or
using a combination of tools such as Bandage, Mauve, CLC region [9, 10, 77]. On the other hand, more advanced genomic
Workbench, and EDGE [71, 72]. Lastly, combining reference characterizations usually require more expertise and com-
mapping and de novo assembly can be a good strategy for putational power (servers or high-performance computers).
increasing the accuracy of the overall genome and identifying Although user-friendly pipelines have been made available for
changes in the pathogen [4, 16, 61]. rapid advanced analyses (EDGE, Galaxy, Nextstrain), training
Consensus postassembly curation and quality control can in specific software would be recommended for laboratories
be sometimes neglected, but this is a critically important that wish to undertake comprehensive phylodynamic analyses
step in the assembly of a pathogen genome, regardless of that leverage the relatively fast evolutionary rate of RNA viruses
whether reference mapping or de novo assembly is used for to gain critical epidemiological insights into viral epidemics

(Figure 2) [78]. While many phylodynamic software packages
Box 1. Considerations required before performing and
are increasingly flexible and powerful, several important factors interpreting advanced viral genomic analysesa
need to be taken into account to accurately infer and interpret
evolutionary epidemiological analyses (Box 1). 1. Method of case surveillance (active vs passive) and bias
Performed and interpreted correctly, viral phylodynamic introduced by sampling skew by years/locales
analyses can clarify the putative origins and early spread of major 2. Biases from unsampled asymptomatic cases
viral epidemics. For instance, recent Bayesian analyses of HIV-1 3. Role of pathogen serotype/subtype, biospecimen
genomes sequenced by contemporary NGS on historical sera type, and time point of sampling in whole-genome
sampled from early US HIV cases indicated the role of New York sequencing quality
City as an early hub of HIV-1 dissemination in North America, 4. Availability and quality of background reference se-
and emphasized how whole-genome sequence data was critical in quence data
resolving such epidemiological insights into the early HIV pan- 5. Determination of recombinant or reassortant

demic [79]. Phylodynamic approaches have also been used to de- sequences, and their role in phylogenetic interpretation
termine the spatial origins of the 2009 H1N1 pandemic, the period 6. Identification of reference or study sequences with an im-
of cryptic transmission of the 2016 Zika virus Florida outbreak, plausible amount of evolution relative to sampling time
and whether local H7N9 outbreaks in China have been seeded 7. Alignment quality, including decisions to include non-
from single versus multiple introductions of strains [29, 80, 81]. coding regions
Beyond reconstructions of epidemic histories, Bayesian statistical 8. Choice of tree inference methods (distance or character
frameworks can also test the role of population size and human based, time scaled or unconstrained) and their compu-
movement in virus spread, and can identify predictors of epi- tational demands
demic growth and peaks in viral populations. Lemey et al [17, 82] 9. Choice of nucleotide substitution model with or
pioneered these extended Bayesian phylodynamic analyses with a without demographic and clock model assumptions
study demonstrating that A/H3N2 influenza virus spread correlates 10. Sensitivity analyses using datasets adjusted for sam-
with air travel on an international scale, but is best explained by ge- pling skew
ographic distance on finer spatial scales. More recently, a similar 11. Rationale for tree rooting, including the availability of a
approach has indicated that rabies virus spread in Africa correlates suitable outgroup
with human population density and connectivity [83]. 12. Choice of statistical support method for phylogeny
While these and other whole-genome analyses have been able nodes and branches
to resolve the patterns and predictors of viral spread, one major 13. Whether confounding should be considered in
challenge has been reconstructing such epidemic dynamics on any associations examined between genotype and
very fine spatial and temporal scales, which may offer the most phenotype
relevance to public health response. Consensus whole-genome Statistical approaches to confirming association between
sequences typically have insufficient variability to distinguish be- phenotype and genotype
tween infecting strains sampled within 2 weeks of each other [84]. a
This list is representative, context dependent, and not
With deep NGS sequencing, additional intrahost viral variant in- exhaustive.
formation is available. Transmission of viral minor variants, that
often do not make it into the consensus genome, has been observed,
which can be used to resolve more granular viral transmissions coverage for genetic characterization results in which one can
[16, 85–87]. In Gire et al [16] patterns of intrahost and interhost have confidence is of utmost importance. Gene calling can be
variation gave insights into the transmission and epidemiology performed in a variety of ways, including RAST or using NCBI
of the 2014 Ebola epidemic. Recent analytical frameworks and services at the time of full genome submission [91]. Results
tools that leverage both within-host and between-host sequence of multiple annotation tools can be compared for accuracy
variability have been developed [88, 89]. However, a key technical and completeness and, if necessary, merged using BEACON
challenge with these analyses is the accuracy of intrahost variant [92]. Beyond gene calling, further analysis of chromosomal
calling, requiring deep NGS coverage, careful experimental de- sequences includes in silico multilocus sequence typing for
sign, and appropriate controls to distinguish PCR and sequencing strain typing using various online resources specific for each
errors from true within-host genetic variation [86, 90]. pathogen, SNP-based phylogenetic analysis for epidemiologic
investigations (BAGA), and prophage characterization using
Advanced Bioinformatic Analyses of Bacterial Sequence Data: Resistance, tools such as PHASTER [93–95]. Specialty gene characteriza-
Virulence, and Extrachromosomal Replicons tion, such as characterization of resistance and virulence factor
Advanced characterization of bacterial organisms can be very genes, is another type of advanced characterization that can be
challenging, and obtaining the necessary depth and breadth of performed for bacterial pathogens, for both chromosomal and

plasmid sequences. Importantly, extrachromosomal sequences parasites Plasmodium, Babesia, Theileria, Toxoplasma, Emeria,
in bacterial pathogens, such as plasmids, can contribute to an and Cryptosporidium can undergo both mitotic and meiotic
observed phenotype or disease, but their accurate identifica- reproduction. For example, Plasmodium parasites are haploid
tion can be challenging. For instance, plasmid sequences can for the majority of their life cycle in the vertebrate host and
be present at varying depths of coverage when compared to diploid for a brief time in a mosquito vector [104]. As a result,
the chromosomal sequences, due to varying copy numbers. In the parasite genome structure is very diverse and highly com-
addition, plasmids in some organisms can contain multiple in- plex. In addition, the parasite population structure can differ
sertion sequences and exhibit substantial modularity, making vastly based on geography and local transmission intensity
an accurate and closed assembly more difficult [24]. For char- [105–107]. The genome sizes of parasites are large (eg, mul-
acterization of antibiotic resistance genes, the Resistance Gene tiple Mbp in size) have extreme repetitive AT-rich or GC-rich
Identifier from the Comprehensive Antibiotic Resistance regions, can vary in size even within the same species, and
Database (CARD) is commonly used [1, 96]. To characterize include extrachromosomal organelle DNA (eg, mitochon-

virulence factor genes, ShortBRED offers analyses with a drial and plastid) [108, 109]. This complexity poses challenges
customized database from the Virulence Factor Database [97, for whole-genome sequencing and bioinformatics analysis,
98]. Specialty gene characterization is currently implemented including generating high-quality, standardized data sets,
using these tools and databases in EDGE for a user-friendly ex- which are needed to accurately assemble genomes, identify
perience [59, 99]. Pathosystems Resource Integration Center polymorphisms, and obtain reliable population genetic sig-
(PATRIC) is another useful tool for specialty gene profiling, nals. Nonetheless, efforts to generate Plasmodium falciparum
and can aid users in producing high-quality visualizations and genome data from different geographies, including the mos-
comparisons of genomes [100, 101]. quito vector, are underway [110]. Other sequencing studies
In the case of bacterial pathogens, which often exhibit hori- for Plasmodium vivax, Leishmania donovani, Trypanosoma
zontal gene transfer, typically all these analyses are employed in brucei, and Toxoplasma gondii have now also been completed
an exploratory manner to gain insights into pathogenic poten- [111–114]. Combined, these studies are improving our under-
tial, treatment options, and transmission patterns. For instance, standing of lineage-specific changes of these parasites in ways
a recent genetic investigation of clinical Klebsiella pneumoniae that were not possible using older sequencing technologies.
isolates from 2 hospitals in South Africa involved characteriza- Perhaps most noteworthy is the impact NGS and bioinfor-
tion and comparisons of chromosomal and plasmid sequences, matics recently had in identifying the P. falciparum locus in-
to include virulence factors, antibiotic resistance determinants, volved in resistance to artemisinin [2].
and prophages. In this study, specific antibiotic resistance genes However, unlike viral and bacterial sequencing, which can be
were characterized and enumerated amongst isolates to deter- used as a routine public health surveillance tool due to their less
mine the circulating antibiotic resistance potential within the complex and smaller genome sizes, parasite genome sequencing
hospital system and the relative contribution of various β-lactam is better suited for research studies focused on understanding
resistance genes. The study also examined relatedness of isolates parasite populations. The results of these studies, such as the
within wards and between hospitals and identified evidence of identification of the artemisinin-resistance marker, can be used
clonal spread of extended spectrum β-lactamase–producing to address public health needs in a timely fashion (eg, routine
K. pneumoniae from one facility to another as a likely conse- artemisinin molecular marker surveillance) [115]. Once suit-
quence of the hospital referral system, thereby using in-depth able molecular markers are identified using sequencing, they
genetic characterization to provide compelling motivation for can rapidly be used in a targeted, multilocus, deep amplicon
increased screening, disinfection, and infection control meas- sequencing approach for routine molecular surveillance of
ures [102]. Importantly, for diagnostic purposes, the NGS and parasitic diseases. For example, this approach can be used to
bioinformatics analyses are usually combined with other assays, characterize all currently known P. falciparum associated drug
such as laboratory resistance testing, to improve diagnosis and resistance markers, for up to 380 patient samples, using a single
treatment accuracy. sequencing assay [116]. More recently, the same method was
used for the detection of multiple blood-borne parasites, using
Advanced Bioinformatic Analyses of Parasites: New Tools for 18S rRNA targeted sequencing, in a single test [117]. This
Understanding Complex Genomes of Ancient Diseases targeted deep amplicon sequencing (TADS) approach may pro-
While rapid advances in NGS and bioinformatics are vide a scalable, cost effective, and deployable tool for a routine
transforming routine public health virology and microbiology molecular surveillance of parasitic diseases in a public health
work, the adoption of similar methods in parasitology has been laboratory. Developing new, standardized, and validated bio-
limited mostly to research-based studies [103]. One of the main informatics analysis tools for parasitic diseases will be critical
reasons for this is the complexity introduced by sexual repro- before these TADS can be adopted into practical clinical and
duction, which is only found in eukaryotes. The apicomplexan public health applications.

CONCLUSIONS Supplement sponsorship. This supplement is sponsored by
The field of NGS is rapidly evolving, with constant improvement WRAIR, LANL, USAMRIID, PUCP (Pontificia Universidad
of sequencing chemistries leading to better outputs and reduced Catolica del Peru), USAFSAM, NIH.
error rates and cost. However, this also makes system standardiza- Potential conflicts of interest. All authors: No reported
tion for routine public health surveillance and pathogen discovery conflicts of interest. All authors have submitted the ICMJE
challenging. While new technologies and protocols can open new Form for Disclosure of Potential Conflicts of Interest. Conflicts
research avenues and improve surveillance activities, the valida- that the editors consider relevant to the content of the manu-
tion and implementation of these methods, including quality as- script have been disclosed.
surance standards, in public health laboratories can take multiple
References
years. Thus, it is important that each public health program plans
1. Schmidt K, Mwaigwisya S, Crossman LC, et al. Identification
for the long term, both in terms of capital and human investment,
of bacterial pathogens and antimicrobial resistance directly
when transitioning to these rapidly evolving approaches. In ad-

from clinical urines by nanopore-based metagenomic
dition to the rapid NGS laboratory advances, data analytics and
sequencing. J Antimicrob Chemother 2017; 72:104–14.
bioinformatics have seen explosive growth in the last decade, pri-
2. Ariey F, Witkowski B, Amaratunga C, et al. A molecular
marily fueled by the advances in computational power and access
marker of artemisinin-resistant Plasmodium falciparum
to cloud services. This has produced a myriad of bioinformatics
malaria. Nature 2014; 505:50–5.
tools, and with them various challenges, making standardization
3. Maljkovic Berry I, Melendrez MC, Li T, et al. Frequency of
of analysis workflows and tools difficult. Nonetheless, as the field
influenza H3N2 intra-subtype reassortment: attributes and
further matures and larger collaborations are established, syn-
implications of reassortant spread. BMC Biol 2016; 14:117.
chronization and standardization of these methodologies are
4. Maljkovic Berry I, Eyase F, Pollett S, et al. Global outbreaks
likely to occur naturally. While the initial investments in this field
and origins of a chikungunya virus variant carrying
may seem daunting and costly, the return on investment can be
mutations which may increase fitness for Aedes aegypti:
achieved within a few short years and it will drastically improve
revelations from the 2016 Mandera, Kenya outbreak. Am J
the ability of a laboratory to respond faster and better to infec-
Trop Med Hyg 2019; 100:1249–57.
tious disease public health threats.
5. Faber M, Faber ML, Papaneri A, et al. A single amino acid
change in rabies virus glycoprotein increases virus spread
Notes and enhances virus pathogenicity. J Virol 2005; 79:14141–8.
Disclaimer. Material has been reviewed by the Walter Reed 6. Chen EC, Yagi S, Kelly KR, et al. Cross-species transmis-
Army Institute of Research. There is no objection to its presentation sion of a novel adenovirus associated with a fulminant
and/or publication. The opinions or assertions contained herein pneumonia outbreak in a new world monkey colony. PLoS
are the private views of the author, and are not to be construed Pathog 2011; 7:e1002155.
as official, or as reflecting true views of the Department of the 7. Goldstein T, Anthony SJ, Gbakima A, et al. The discovery
Army, Department of the Navy, Department of Defense, Centers of Bombali virus adds further support for bats as hosts of
for Disease Control and Prevention, nor the US Government. ebolaviruses. Nat Microbiol 2018; 3:1084–9.
Use of trade names is for identification only and does not imply 8. Grard G, Fair JN, Lee D, et al. A novel rhabdovirus associ-
endorsement by the Centers for Disease Control and Prevention/ ated with acute hemorrhagic fever in central Africa. PLoS
the Agency for Toxic Substances and Disease Registry, the Public Pathog 2012; 8:e1002924.
Health Service, the US Department of Health and Human 9. Toledo-Rueda W, Rosas-Murrieta NH, Munoz-Medina JE,
Services, or the Department of the Army, Department of the Gonzalez-Bonilla CR, Reyes-Leyva J, Santos-Lopez G.
Navy and Department of Defense. K. A. B.-L. is an employee of Antiviral resistance markers in influenza virus sequences in
the US Government and R. G. J. is a military service member. Mexico, 2000–2017. Infect Drug Resist 2018; 11:1751–6.
This work was prepared as part of their official duties. Title 17 10. Wensing AM, van de Vijver DA, Angarano G, et al.;
U.S.C. §105 provides that ‘Copyright protection under this title SPREAD Programme. Prevalence of drug-resistant HIV-1
is not available for any work of the United States Government.’ variants in untreated individuals in Europe: implications for
Title 17 U.S.C. §101 defines a US Government work as a work clinical management. J Infect Dis 2005; 192:958–66.
prepared by a military service member or employee of the US 11. Salje H, Lessler J, Maljkovic Berry I, et al. Dengue diversity
Government as part of that person’s official duties. across spatial and temporal scales: local structure and the
Financial support. This work was supported by the Armed effect of host population size. Science 2017; 355:1302–6.
Forces Health Surveillance Branch and its Global Emerging 12. Quick J, Loman NJ, Duraffour S, et al. Real-time, portable
Infections Surveillance Section, FY2018 ProMIS (grant number genome sequencing for Ebola surveillance. Nature 2016;
P0108_18_WR). 530:228–32.

13. Stewart-Ibarra AM, Ryan SJ, Kenneson A, et al. The 29. Grubaugh ND, Ladner JT, Kraemer MUG, et al. Genomic
burden of Dengue fever and Chikungunya in Southern epidemiology reveals multiple introductions of Zika virus
Coastal Ecuador: epidemiology, clinical presentation, and into the United States. Nature 2017; 546:401–5.
phylogenetics from the first two years of a prospective 30. Fernandez-Garcia MD, Volle R, Joffret ML, et al. Genetic
study. Am J Trop Med Hyg 2018; 98:1444–59. characterization of Enterovirus A71 circulating in Africa.
14. Faria NR, Azevedo RDSDS, Kraemer MUG, et al. Zika virus Emerg Infect Dis 2018; 24:754–7.
in the Americas: early epidemiological and genetic findings. 31. Mongkolrattanothai K, Naccache SN, Bender JM, et al.
Science 2016; 352:345–9. Neurobrucellosis: unexpected answer from metagenomic
15. Faria NR, Quick J, Claro IM, et al. Establishment and cryptic next-generation sequencing. J Pediatric Infect Dis Soc
transmission of Zika virus in Brazil and the Americas. 2017; 6:393–8.
Nature 2017; 546:406–10. 32. Graf EH, Simmon KE, Tardif KD, et al. Unbiased detec-
16. Gire SK, Goba A, Andersen KG, et al. Genomic surveillance tion of respiratory viruses by use of RNA sequencing-based

elucidates Ebola virus origin and transmission during the metagenomics: a systematic comparison to a commercial
2014 outbreak. Science 2014; 345:1369–72. PCR panel. J Clin Microbiol 2016; 54:1000–7.
17. Lemey P, Rambaut A, Bedford T, et al. Unifying viral ge- 33. Dickson RP, Singer BH, Newstead MW, et al. Enrichment of
netics and human transportation data to predict the global the lung microbiome with gut bacteria in sepsis and the acute
transmission dynamics of human influenza H3N2. PLoS respiratory distress syndrome. Nat Microbiol 2016; 1:16113.
Pathog 2014; 10:e1003932. 34. Pendleton KM, Erb-Downward JR, Bao Y, et al. Rapid
18. Gargis AS, Kalman L, Lubin IM. Assuring the quality pathogen identification in bacterial pneumonia using real-
of next-generation sequencing in clinical microbiology time metagenomics. Am J Respir Crit Care Med 2017;
and public health laboratories. J Clin Microbiol 2016; 196:1610–2.
54:2857–65. 35. McGann P, Bunin JL, Snesrud E, et al. Real time application
19. Chin CS, Sorenson J, Harris JB, et al. The origin of the Haitian of whole genome sequencing for outbreak investigation -
cholera outbreak strain. N Engl J Med 2011; 364:33–42. what is an achievable turnaround time? Diagn Microbiol
20. Au KF, Underwood JG, Lee L, Wong WH. Improving Infect Dis 2016; 85:277–82.
PacBio long read accuracy by short read alignment. PLoS 36. Faria NR, Sabino EC, Nunes MR, Alcantara LC, Loman NJ,
One 2012; 7:e46679. Pybus OG. Mobile real-time surveillance of Zika virus in
21. Jain M, Fiddes IT, Miga KH, Olsen HE, Paten B, Akeson M. Brazil. Genome Med 2016; 8:97.
Improved data analysis for the MinION nanopore se- 37. Salter SJ, Cox MJ, Turek EM, et al. Reagent and labora-
quencer. Nat Methods 2015; 12:351–6. tory contamination can critically impact sequence-based
22. Goodwin S, Gurtowski J, Ethe-Sayers S, Deshpande P, microbiome analyses. BMC Biol 2014; 12:87.
Schatz MC, McCombie WR. Oxford Nanopore sequencing, 38. Qin XC, Shi M, Tian JH, et al. A tick-borne segmented RNA
hybrid error correction, and de novo assembly of a eukary- virus contains genome segments derived from unsegmented
otic genome. Genome Res 2015; 25:1750–6. viral ancestors. Proc Natl Acad Sci U S A 2014; 111:6744–9.
23. Graham RM, Doyle CJ, Jennison AV. Real-time investiga- 39. Yu JM, Li JS, Ao YY, Duan ZJ. Detection of novel viruses in
tion of a Legionella pneumophila outbreak using whole ge- porcine fecal samples from China. Virol J 2013; 10:39.
nome sequencing. Epidemiol Infect 2014; 142:2347–51. 40. Korkusol A, Takhampunya R, Hang J, et al. A novel fla-
24. LaBreck PT, Rice GK, Paskey AC, et al. Conjugative transfer vivirus detected in two Aedes spp. collected near the
of a novel staphylococcal plasmid encoding the biocide re- demilitarized zone of the Republic of Korea. J Gen Virol
sistance gene, qacA. Front Microbiol 2018; 9:2664. 2017; 98:1122–31.
25. Karikari TK. Bioinformatics in Africa: the rise of Ghana? 41. Holmes EC, Rambaut A, Andersen KG. Pandemics: spend
PLoS Comput Biol 2015; 11:e1004308. on surveillance, not prediction. Nature 2018; 558:180–2.
26. Pollett S, Leguia M, Nelson MI, et al. Feasibility and effec- 42. Lee HK, Lee CK, Tang JW, Loh TP, Koay ES. Contamination-
tiveness of a brief, intensive phylogenetics workshop in a controlled high-throughput whole genome sequencing for
middle-income country. Int J Infect Dis 2016; 42:24–7. influenza A viruses using the MiSeq sequencer. Sci Rep
27. Kwong JC, Mercoulia K, Tomita T, et al. Prospective whole- 2016; 6:33318.
genome sequencing enhances national surveillance of 43. Kumar A, Murthy S, Kapoor A. Evolution of selective-
Listeria monocytogenes. J Clin Microbiol 2016; 54:333–42. sequencing approaches for virus discovery and virome
28. Wilson MR, Naccache SN, Samayoa E, et al. Actionable diag- analysis. Virus Res 2017; 239:172–9.
nosis of neuroleptospirosis by next-generation sequencing. 44. Conceição-Neto N, Zeller M, Lefrère H, et al. Modular ap-
N Engl J Med 2014; 370:2408–17. proach to customise sample preparation procedures for

viral metagenomics: a reproducible protocol for virome [published online ahead of print 28 April, 2015]. Food
analysis. Sci Rep 2015; 5:16532. Environ Virol doi:10.1007/s12560-015-9194-4.
45. Metsky HC, Matranga CB, Wohl S, et al. Zika virus evolu- 58. Flygare S, Simmon K, Miller C, et al. Taxonomer: an inter-
tion and spread in the Americas. Nature 2017; 546:411–5. active metagenomics analysis portal for universal pathogen
46. Allander T, Emerson SU, Engle RE, Purcell RH, Bukh J. A detection and host mRNA expression profiling. Genome
virus discovery method incorporating DNase treatment Biol 2016; 17:111.
and its application to the identification of two bovine parvo- 59. Li PE, Lo CC, Anderson JJ, et al. Enabling the democratiza-
virus species. Proc Natl Acad Sci U S A 2001; 98:11609–14. tion of the genomics revolution with a fully integrated web-
47. Gu W, Crawford ED, O’Donovan BD, et al. Depletion of abun- based bioinformatics platform. Nucleic Acids Res 2017;
dant sequences by hybridization (DASH): using Cas9 to remove 45:67–80.
unwanted high-abundance species in sequencing libraries and 60. Kilianski A, Carcel P, Yao S, et al. Pathosphere.org: path-
molecular counting applications. Genome Biol 2016; 17:41. ogen detection and characterization through a web-based,

48. Li D, Li Z, Zhou Z, et al. Direct next-generation sequencing open source informatics platform. BMC Bioinformatics
of virus-human mixed samples without pretreatment is fa- 2015; 16:416.
vorable to recover virus genome. Biol Direct 2016; 11:3. 61. Hang J, Forshey BM, Kochel TJ, et al. Random amplification
49. Mate SE, Kugelman JR, Nyenswah TG, et al. Molecular ev- and pyrosequencing for identification of novel viral genome
idence of sexual transmission of Ebola virus. N Engl J Med sequences. J Biomol Tech 2012; 23:4–10.
2015; 373:2448–54. 62. National Center for Biotechnology Information. Basic local
50. Yang Y, Walls S, Gross SM, Schroth GP, Jarman RG, Hang J. alignment search tool. https://blast.ncbi.nlm.nih.gov/Blast.
Targeted sequencing of respiratory viruses in clinical cgi. Accessed 6 June, 2019.
specimens for pathogen identification and genome-wide 63. Li H. Aligning sequence reads, clone sequences and as-
analysis. In: Moya A, Brocal BP, eds. The human virome sembly contigs with BWA-MEM. arXiv:1303.3997. 2013.
methods and protocols. New York, NY: Humana Press, https://arxiv.org/abs/1303.3997. Accessed 6 June, 2019.
2018. 64. Langmead B, Salzberg SL. Fast gapped-read alignment with
51. Briese T, Kapoor A, Mishra N, et al. Virome capture Bowtie 2. Nat Methods 2012; 9:357–9.
sequencing enables sensitive viral diagnosis and compre- 65. Koren S, Treangen TJ, Hill CM, Pop M, Phillippy AM.
hensive virome analysis. MBio 2015; 6:e01491–15. Automated ensemble assembly and validation of microbial
52. Cummings MJ, Tokarz R, Bakamutumaho B, et al. Precision genomes. BMC Bioinformatics 2014; 15:126.
surveillance for viral respiratory pathogens: virome capture 66. Viral Diseases Branch, Walter Reed Army Institute of
sequencing for the detection and genomic characterization Research ngs_mapper. https://github.com/VDBWRAIR/
of severe acute respiratory infection in Uganda [published ngs_mapper. Accessed 6 June, 2019.
online ahead of print 7 August, 2018]. Clin Infect Dis doi: 67. Zerbino DR, Birney E. Velvet: algorithms for de novo short
10.1093/cid/ciy656. read assembly using de Bruijn graphs. Genome Res 2008;
53. Wylie TN, Wylie KM, Herter BN, Storch GA. Enhanced 18:821–9.
virome sequencing using targeted sequence capture. 68. Simpson JT, Wong K, Jackman SD, Schein JE, Jones SJ,
Genome Res 2015; 25:1910–20. Birol I. ABySS: a parallel assembler for short read sequence
54. Paskey AC, Frey KG, Schroth G, Gross S, Hamilton T, data. Genome Res 2009; 19:1117–23.
Bishop-Lilly KA. Enrichment post-library preparation 69. Luo R, Liu B, Xie Y, et al. SOAPdenovo2: an empirically
enhances the sensitivity of high-throughput sequencing- improved memory-efficient short-read de novo assembler.
based detection and characterization of viruses from com- Gigascience 2012; 1:18.
plex samples. BMC Genomics 2019; 20:155. 70. Bankevich A, Nurk S, Antipov D, et al. SPAdes: a new ge-
55. Lalremruata A, Jeyaraj S, Engleitner T, et al. Species and nome assembly algorithm and its applications to single-cell
genotype diversity of Plasmodium in malaria patients from sequencing. J Comput Biol 2012; 19:455–77.
Gabon analysed by next generation sequencing. Malar J 71. Wick RR, Schultz MB, Zobel J, Holt KE. Bandage: interactive
2017; 16:398. visualization of de novo genome assemblies. Bioinformatics
56. Guan M, Hall JS, Zhang X, et al. Aerosol transmission of 2015; 31:3350–2.
gull-origin Iceland subtype H10N7 influenza A virus in 72. Darling AE, Mau B, Perna NT. progressiveMauve: multiple
ferrets [published online ahead of print 17 April, 2019]. J genome alignment with gene gain, loss and rearrangement.
Virol doi: 10.1128/JVI.00282-19. PLoS One 2010; 5:e11147.
57. Ogorzaly L, Walczak C, Galloux M, Etienne S, Gassilloud B, 73. Jia P, Li F, Xia J, et al. Consensus rules in variant detection
Cauchie HM. Human adenovirus diversity in water samples from next-generation sequencing data. PLoS One 2012;
using a next-generation amplicon sequencing approach 7:e38470.

74. Kraemer MUG, Cummings DAT, Funk S, et al. 89. Skums P, Zelikovsky A, Singh R, et al. QUENTIN: recon-
Reconstruction and prediction of viral disease epidemics. struction of disease transmissions from viral quasispecies
Epidemiol Infect 2018; 1–7. genomic data. Bioinformatics 2018; 34:163–70.
75. Woolhouse ME, Rambaut A, Kellam P. Lessons from Ebola: 90. McCrone JT, Woods RJ, Martin ET, Malosh RE, Monto AS,
improving infectious disease surveillance to inform out- Lauring AS. Stochastic processes constrain the within and
break management. Sci Transl Med 2015; 7:307rv5. between host evolution of influenza virus. Elife 2018; 7:pii:
76. Pollett S, Melendrez MC, Maljkovic Berry I, et al. e35962.
Understanding dengue virus evolution to support epidemic 91. Aziz RK, Bartels D, Best AA, et al. The RAST server: rapid
surveillance and counter-measure development. Infect annotations using subsystems technology. BMC Genomics
Genet Evol 2018; 62:279–95. 2008; 9:75.
77. Kumar S, Stecher G, Tamura K. MEGA7: molecular evolu- 92. Kalkatawi M, Alam I, Bajic VB. BEACON: automated tool
tionary genetics analysis version 7.0 for bigger datasets. Mol for Bacterial GEnome Annotation ComparisON. BMC

Biol Evol 2016; 33:1870–4. Genomics 2015; 16:616.
78. Pybus OG, Fraser C, Rambaut A. Evolutionary epidemi- 93. Millar EV, Rice GK, Elassal EM, et al. Genomic charac-
ology: preparing for an age of genomic plenty. Philos Trans terization of USA300 methicillin-resistant Staphylococcus
R Soc Lond B Biol Sci 2013; 368:20120193. aureus (MRSA) to evaluate intraclass transmission and re-
79. Worobey M, Watts TD, McKay RA, et al. 1970s and ‘patient currence of skin and soft tissue infection (SSTI) among
0’ HIV-1 genomes illuminate early HIV/AIDS history in high-risk military trainees. Clin Infect Dis 2017; 65:461–8.
North America. Nature 2016; 539:98–101. 94. Arndt D, Marcu A, Liang Y, Wishart DS. PHAST,
80. Mena I, Nelson MI, Quezada-Monroy F, et al. Origins of the PHASTER and PHASTEST: Tools for finding prophage
2009 H1N1 influenza pandemic in swine in Mexico. Elife in bacterial genomes [published online ahead of print
2016; 5:pii: e16777. 25 September, 2017]. Brief Bioinform doi: 10.1093/bib/
81. Liu J, Xu J, Liu L, et al. Sudden emergence of human bbx121.
infections with H7N9 avian influenza A virus in Hubei 95. Mottawea W, Duceppe MO, Dupras AA, et al. Salmonella
province, central China. Sci Rep 2018; 8:2486. enterica prophage sequence profiles reflect genome diver-
82. Suchard MA, Lemey P, Baele G, Ayres DL, Drummond AJ, sity and can be used for high discrimination subtyping.
Rambaut A. Bayesian phylogenetic and phylodynamic data Front Microbiol 2018; 9:836.
integration using BEAST 1.10. Virus Evol 2018; 4:vey016. 96. Jia B, Raphenya AR, Alcock B, et al. CARD 2017: expan-
83. Dellicour S, Rose R, Faria NR, et al. Using viral gene sion and model-centric curation of the comprehensive
sequences to compare and explain the heterogeneous antibiotic resistance database. Nucleic Acids Res 2017;
spatial dynamics of virus epidemics. Mol Biol Evol 2017; 45:D566–73.
34:2563–71. 97. Kaminski J, Gibson MK, Franzosa EA, Segata N, Dantas G,
84. Pybus OG, Tatem AJ, Lemey P. Virus evolution and trans- Huttenhower C. High-specificity targeted functional pro-
mission in an ever more connected world. Proc Biol Sci filing in microbial communities with ShortBRED. PLoS
2015; 282:20142878. Comput Biol 2015; 11:e1004557.
85. Stack JC, Murcia PR, Grenfell BT, Wood JL, Holmes EC. 98. Chen L, Zheng D, Liu B, Yang J, Jin Q. VFDB 2016: hier-
Inferring the inter-host transmission of influenza A virus archical and refined dataset for big data analysis–10 years
using patterns of intra-host genetic variation. Proc Biol Sci on. Nucleic Acids Res 2016; 44:D694–7.
2013; 280:20122173. 99. Philipson C, Davenport K, Voegtly L, et al. Brief pro-
86. Zanini F, Brodin J, Thebo L, et al. Population genomics of tocol for EDGE bioinformatics: analyzing microbial and
intrapatient HIV-1 evolution. Elife 2015; 4:pii: e11282. metagenomic NGS data. Bio-Protocol 2017; 7:e2622.
87. Fischer GE, Schaefer MK, Labus BJ, et al. Hepatitis C virus 100. Davis JJ, Boisvert S, Brettin T, et al. Antimicrobial re-
infections from unsafe injection practices at an endoscopy sistance prediction in PATRIC and RAST. Sci Rep 2016;
clinic in Las Vegas, Nevada, 2007–2008. Clin Infect Dis 6:27930.
2010; 51:267–73. 101. Mao C, Abraham D, Wattam AR, et al. Curation, inte-
88. Wymant C, Hall M, Ratmann O, et al. PHYLOSCANNER: gration and visualization of bacterial virulence factors in
inferring transmission from within- and between-host PATRIC. Bioinformatics 2015; 31:252–8.
pathogen genetic diversity [published online ahead of print 102. Founou RC, Founou LL, Allam M, Ismail A, Essack SY.
23 November, 2017]. Mol Biol Evol doi: 10.1093/molbev/ Whole genome sequencing of extended spectrum
msx304. β-lactamase (ESBL)-producing Klebsiella pneumoniae

isolated from hospitalized patients in KwaZulu-Natal, of resistance to severe malaria in a region of ancient bal-
South Africa. Sci Rep 2019; 9:6266. ancing selection. Nature 2015; 526:253–7.
103. Kirchner S, Power BJ, Waters AP. Recent advances in 111. Neafsey DE, Galinsky K, Jiang RH, et al. The malaria
malaria genomics and epigenomics. Genome Med 2016; parasite Plasmodium vivax exhibits greater genetic di-
8:92. versity than Plasmodium falciparum. Nat Genet 2012;
104. Guttery DS, Holder AA, Tewari R. Sexual development 44:1046–50.
in Plasmodium: lessons from functional analyses. PLoS 112. Downing T, Imamura H, Decuypere S, et al. Whole genome
Pathog 2012; 8:e1002404. sequencing of multiple Leishmania donovani clinical isolates
105. Anderson TJ, Haubold B, Williams JT, et al. Microsatellite provides insights into population structure and mechanisms
markers reveal a spectrum of population structures in the of drug resistance. Genome Res 2011; 21:2143–56.
malaria parasite Plasmodium falciparum. Mol Biol Evol 113. Goodhead I, Capewell P, Bailey JW, et al. Whole-genome
2000; 17:1467–82. sequencing of Trypanosoma brucei reveals introgression

106. Volkman SK, Sabeti PC, DeCaprio D, et al. A genome- between subspecies that is associated with virulence. MBio
wide map of diversity in Plasmodium falciparum. Nat 2013; 4:pii: e00197-13.
Genet 2007; 39:113–9. 114. Kissinger JC, Gajria B, Li L, Paulsen IT, Roos DS. ToxoDB:
107. Miotto O, Amato R, Ashley EA, et al. Genetic architec- accessing the Toxoplasma gondii genome. Nucleic Acids
ture of artemisinin-resistant Plasmodium falciparum. Nat Res 2003; 31:234–6.
Genet 2015; 47:226–34. 115. World Health Organization. WHO updates on artemisinin
108. Otto TD, Böhme U, Sanders M, et al. Long read assemblies resistance and ACT efficacy. Geneva: WHO, 2017.
of geographically dispersed Plasmodium falciparum 116. Talundzic E, Ravishankar S, Kelley J, et al. Next-generation
isolates reveal highly structured subtelomeres. Wellcome sequencing and bioinformatics protocol for malaria
Open Res 2018; 3:52. drug resistance marker surveillance. Antimicrob Agents
109. Carlton JM, Adams JH, Silva JC, et al. Comparative geno- Chemother 2018; 62:pii: e02474-17.
mics of the neglected human malaria parasite Plasmodium 117. Flaherty BR, Talundzic E, Barratt J, et al. Restriction en-
vivax. Nature 2008; 455:757–63. zyme digestion of host DNA enhances universal detection
110. Band G, Rockett KA, Spencer CC, Kwiatkowski DP; of parasitic pathogens in blood via targeted amplicon deep
Malaria Genomic Epidemiology Network. A novel locus sequencing. Microbiome 2018; 6:164.

Jiz 286

Uploaded by

Copyright:

Available Formats

Jiz 286

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Jiz 286

Uploaded by

Copyright:

Available Formats

The Journal of Infectious Diseases

Next Generation Sequencing and Bioinformatics

Downloaded from https://academic.oup.com/jid/article/221/Supplement_3/S292/5586940 by guest on 04 April 2022

S292 • JID 2020:221 (Suppl 3) • Maljkovic Berry et al

Downloaded from https://academic.oup.com/jid/article/221/Supplement_3/S292/5586940 by guest on 04 April 2022

NGS and Bioinformatics Methodologies • JID 2020:221 (Suppl 3) • S293

Downloaded from https://academic.oup.com/jid/article/221/Supplement_3/S292/5586940 by guest on 04 April 2022

S294 • JID 2020:221 (Suppl 3) • Maljkovic Berry et al

Downloaded from https://academic.oup.com/jid/article/221/Supplement_3/S292/5586940 by guest on 04 April 2022

NGS and Bioinformatics Methodologies • JID 2020:221 (Suppl 3) • S295

Downloaded from https://academic.oup.com/jid/article/221/Supplement_3/S292/5586940 by guest on 04 April 2022

S296 • JID 2020:221 (Suppl 3) • Maljkovic Berry et al

Downloaded from https://academic.oup.com/jid/article/221/Supplement_3/S292/5586940 by guest on 04 April 2022

QC: Library quality & quantity

Controls: spiked DNA, control pathogen,

Wet Lab Workflow Considerations:

• Study goal, detection or genome assembly

NGS and Bioinformatics Methodologies • JID 2020:221 (Suppl 3) • S297

Downloaded from https://academic.oup.com/jid/article/221/Supplement_3/S292/5586940 by guest on 04 April 2022

S298 • JID 2020:221 (Suppl 3) • Maljkovic Berry et al

Quality Control and Preprocessing*

Pathogen Discovery Genome Construction Advanced Analysis

2 De novo assembly, contig 1 2 Gene calling/Annotation, alignment,

Downloaded from https://academic.oup.com/jid/article/221/Supplement_3/S292/5586940 by guest on 04 April 2022

1 Taxonomer, EDGE, Pathosphere, 1 1

*Computational requirements will depend on sequencing platform data output size.

NGS and Bioinformatics Methodologies • JID 2020:221 (Suppl 3) • S299

Downloaded from https://academic.oup.com/jid/article/221/Supplement_3/S292/5586940 by guest on 04 April 2022

S300 • JID 2020:221 (Suppl 3) • Maljkovic Berry et al

Downloaded from https://academic.oup.com/jid/article/221/Supplement_3/S292/5586940 by guest on 04 April 2022

NGS and Bioinformatics Methodologies • JID 2020:221 (Suppl 3) • S301

Downloaded from https://academic.oup.com/jid/article/221/Supplement_3/S292/5586940 by guest on 04 April 2022

S302 • JID 2020:221 (Suppl 3) • Maljkovic Berry et al

Downloaded from https://academic.oup.com/jid/article/221/Supplement_3/S292/5586940 by guest on 04 April 2022

NGS and Bioinformatics Methodologies • JID 2020:221 (Suppl 3) • S303

Downloaded from https://academic.oup.com/jid/article/221/Supplement_3/S292/5586940 by guest on 04 April 2022

S304 • JID 2020:221 (Suppl 3) • Maljkovic Berry et al

Downloaded from https://academic.oup.com/jid/article/221/Supplement_3/S292/5586940 by guest on 04 April 2022

NGS and Bioinformatics Methodologies • JID 2020:221 (Suppl 3) • S305

Downloaded from https://academic.oup.com/jid/article/221/Supplement_3/S292/5586940 by guest on 04 April 2022

S306 • JID 2020:221 (Suppl 3) • Maljkovic Berry et al

Downloaded from https://academic.oup.com/jid/article/221/Supplement_3/S292/5586940 by guest on 04 April 2022

NGS and Bioinformatics Methodologies • JID 2020:221 (Suppl 3) • S307

You might also like