Supplementary appendix
This appendix has been provided by the authors to give readers additional information about
their work.
1
1
2
3
12
Untargeted next-generation sequencing-based first-line diagnosis of infection in
immunocompromised adults: a multicentre, blinded, prospective study
Online supplemental materials
Table of Contents
List of tables..........................................................................................................................................3
List of investigators................................................................................................................................4
Supplementary methods.......................................................................................................................8
Supplementary tables..........................................................................................................................13
Supplementary figures........................................................................................................................31
2
4
5
6
7
8
9
10
11
12
13
14
15
34
List of tables
Supplementary Table 1. Conventional microbiological methods used in the standard procedures.............................................................................................................................................................13Supplementary Table 2. Concordant identifications by UNGS and standard procedures at inclusion.............................................................................................................................................14Supplementary Table 3. Divergent identifications by UNGS and standard procedures at inclusion.............................................................................................................................................................23Supplementary Table 4. Clinically-relevant viruses and bacteria in samples with concordant results for UNGS and standard procedures......................................................................................28Supplementary Table 5. Comparison of UNGS at inclusion with standard procedures running for 30 days for clinically-relevant viruses, bacteria and fungi..............................................................29
3
16
1718192021222324252627
56
List of investigators
Country Investigator Affiliation(s) Qualification
France Perrine Parize Paris Descartes University, Sorbonne
Paris Cité, Necker Pasteur Center for
Infectious Diseases and Tropical
Medicine, Necker Enfants Malades
University Hospital, Institut Imagine,
Paris
MD
France Erika Muth PathoQuest, Paris
France Clémence Richaud Department of Microbiology,
European Georges Pompidou Hospital,
Assistance Publique-Hôpitaux de
Paris, Université Paris Descartes, Paris
MD
France Marlène Gratigny PathoQuest, Paris MSc
France Benoit Pilmis Paris Descartes University, Sorbonne
Paris Cité, Necker Pasteur Center for
Infectious Diseases and Tropical
Medicine, Necker Enfants Malades
University Hospital, Institut Imagine,
Paris
MD
France Arnaud Lamamy PathoQuest, Paris TECH
France Jean-Luc Mainardi Department of Microbiology,
European Georges Pompidou Hospital,
Assistance Publique-Hôpitaux de
Paris, Université Paris Descartes, Paris
MD, PhD
France Justine Cheval PathoQuest, Paris MSc
France Louise de Visser PathoQuest Paris MSc
France Florence Jagorel PathoQuest, Paris TECHN
4
28
78
Country Investigator Affiliation(s) Qualification
France Laura Ben Yahia PathoQuest, Paris TECHN
France Geraldine Bamba PathoQuest, Paris TECHN
France Myriam Dubois PathoQuest, Paris B.Eng
France Olivier Join-Lambert Paris Descartes University, Sorbonne
Paris Cité, Laboratory of Microbiology,
Necker Enfants Malades University
Hospital, Paris
MD, PhD
France Marianne Leruez-Ville Paris Descartes University, Sorbonne
Paris Cité, Laboratory of Microbiology,
Necker Enfants Malades University
Hospital, Paris
MD, PhD
France Xavier Nassif Paris Descartes University, Sorbonne
Paris Cité, Laboratory of Microbiology,
Necker Enfants Malades University
Hospital, Paris
MD, PhD
France Agnes Lefort University Paris-Diderot, Hospital
Beaujon, Clichy
MD, PhD
France Fanny Lanternier Paris Descartes University, Sorbonne
Paris Cité, Necker Pasteur Center for
Infectious Diseases and Tropical
Medicine, Necker Enfants Malades
University Hospital, Institut Imagine,
Paris
MD, PhD
France Felipe Suarez Hematology Department, Necker
Hospital, Paris Descartes - Sorbonne
Paris Cité University, INSERM U1163
CNRS ERL8654, Imagine Institute,
Paris
MD, PhD
5910
Country Investigator Affiliation(s) Qualification
France Olivier Lortholary Paris Descartes University, Sorbonne
Paris Cité, Necker Pasteur Center for
Infectious Diseases and Tropical
Medicine, Necker Enfants Malades
University Hospital, Institut Imagine,
Paris
MD, PhD
France Marc Lecuit Paris Descartes University, Sorbonne
Paris Cité, Necker Pasteur Center for
Infectious Diseases and Tropical
Medicine, Necker Enfants Malades
University Hospital, Institut Imagine,
Paris, and Institut Pasteur, Biology of
Infection Unit, Inserm U1117,
Pathogen Discovery Laboratory, Paris
MD, PhD
France Marc Eloit PathoQuest, Paris and Institut Pasteur,
Biology of Infection Unit, Inserm
U1117, Pathogen Discovery
Laboratory, Paris
DVM, PhD
France Emmanuel Guérot Groupe HEGP MD
France Juliette Pavie Groupe HEGP MD
France Georgia Malamut Groupe HEGP MD, PhD
France Bruno Landi Groupe HEGP MD
France Adrien Michon Groupe HEGP MD
France Isabelle Pierre Groupe HEGP MD
France Romain Guillemain Groupe HEGP MD
France Elisabeth Fabre Groupe HEGP MD
France Stéphane Oudard Groupe HEGP MD, PhD
France Alexis Ferré Groupe HEGP MD
61112
Country Investigator Affiliation(s) Qualification
France Sarah Roussel Groupe HEGP MD
France Jean Pastre Groupe HEGP MD
France Eric Thervet Groupe HEGP MD, PhD
France Christophe Legendre Groupe Necker MD
France Anne Scemla Groupe Necker MD
France Olivier Hermine Groupe Necker MD, PhD
France David Lebeaux Groupe Necker MD
France Claire Aguilar Groupe Necker MD
7
29
1314
Supplementary methods
Sample preparation and sequencing
Samples were first spiked with DNA and RNA controls composed of one RNA, and one DNA,
bacteriophage at a final concentration of 105 and 106 genomes/ml, respectively, to allow for
quality control of the extraction, amplification and sequencing steps. Nucleic acids (DNA and
RNA) were extracted from plasma samples and other biological fluids, which had been
previously treated with a cocktail of nucleases to reduce contamination by free nucleic acids,
using the QIAamp Cador pathogen kit (Qiagen Cat. No. 54104, Hilden, Germany). The extraction
of bacterial nucleic acids was performed on whole blood using the MolYsis-5 reagents (Cat. No.
D-321-050, Molzym, Bremen, Germany). Extraction of nucleic acids was controlled by
quantitative real-time q(RT-)PCR based on the TaqMan® technology against target sequences
within bacteriophages introduced as internal controls. First-strand cDNA synthesis was
performed by using commercially available reagents (SuperScript III First-Strand synthesis
system for RT-PCR kit, Life Technologies Cat. No. 18080-051, Marly le Roi, France). Nucleic acids
were random-amplified by a multiple displacement amplification (MDA) reaction using the
bacteriophage Phi29 DNA polymerase (Qiagen Cat. No. 207043, Breme,, Germany).
Amplification of viral and bacterial nucleic acids was controlled by the same RT-PCR against
internal control sequences (bacteriophages), as described above. High molecular weight DNA
resulting from MDA was fragmented with a Covaris M220 ultrasonicator (Woburn, USA) at a
power of 50 W and at 200 cycles per burst for 160 seconds. Fragmented DNA was further
purified by Agencourt AMPure XP beads (Beckman Coulter, Cat. No. A63880-881, Villepinte,
France) and end-repaired (Life Technologies Cat. No. 4471252, Marly le Roi, France). Adapters
were ligated and DNA was nick-repaired (Ion Xpress Barcode Adapters kit, Life Technologies
Cat. No. 4471250, 4474009, 4474517-21). DNA was then purified with Agencourt AMPure XP
beads. This unamplified library was sized to 200 nucleotides using Solid Phase Reversible
Immobilization (SPRI) beads (Beckman Coulter, Cat. No. B23317-18). The library was then
amplified by PCR using Ion Torrent reagents (Life Technologies, ThermoFischer, Waltham,
8
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
1516
USA). After purification of DNA with AMPure XP beads, and quantification and size verification
on an Agilent Bioanalyzer 2100 instrument, the library was sequenced on an Ion Proton
instrument (Life Technologies, ThermoFischer, Waltham, USA ) using a minimum of 40 million
reads per sample according to the manufacturer’s instructions.
Sequence filtering, mapping, and scoring
Various filtering steps were applied to each sequence; nucleotides at the extremities were
trimmed according to their quality. For size trimming, sequence length had to be greater than
60 nucleotides. The human sequences were then selectively suppressed by mapping against the
human genome (hg19, Burrows–Wheeler transform algorithm modified by PathoQuest SAS).
Sequences of Anelloviridae were filtered using a subtractive comparison against a proprietary
database of anelloviruses (provided by PathoQuest SAS, version October 2013, Burrows–
Wheeler transform algorithm modified by PathoQuest SAS, available on request). Remaining
sequences were then mapped against a proprietary database of viruses and bacteria created
and maintained by PathoQuest SAS (version mid-2013, database composition available on
request, Burrows–Wheeler transform algorithm modified by PathoQuest SAS). This database
was built using all available long length reference genome sequences downloaded from NCBI at
the date of the database (July 2013). We only focused on viral and bacterial species of clinical
interest. Species selection was performed using bibliography (for bacteria) plus International
Committee on Taxonomy of Viruses (ICTV) classification for viruses. Database, species list and
sequences in FASTA format are available on request. Mapping results were deposited in a
proprietary and dedicated data warehouse that stores and organises all alignment-related
information for each potential target. The bioinformatics pipeline, database schemes and expert
rules used for the analysis are proprietary. These tools were developed using industrial-level
quality processes and were fully tested (unit test and acceptance tests).
The scoring methodology used various metrics including genome coverage, genome
distribution, sequencing depth, and alignment quality.These metrics are tightly associated.
9
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
1718
Thus, a high depth of sequencing on a high titrated pathogen in the sequencing library is
necessary to obtain a good genome coverage according to the Lander-Waterman theory. Our
scoring for each hit (i.e. pathogen) was the cumulative effect of these different metrics
(coverage percentage, number of unique bloc/part of genome identified [i.e. genome
fragmentation], genome distribution, p-value estimation.). We sought to normalise the score but
this presented a significant challenge: our test is untargeted and we had to deal with the
variable nature of different pathogens’ genomes (e.g. circular, DNA, RNA, and segmented) and a
heterogeneous amount of background noise (the portion of the sequencing library that targets
the host genome rather than pathogens).
As the pathogen capture is random (extraction and amplification), the proportion of a pathogen
in the sequencing library is not correlated to its initial titration. The only way to calibrate our
score and determine a significance threshold was to perform in silico simulation and biological
mock sample generation. Based on our experiments, we determined that a score below 100
should not be considered as a positive hit, but likely as a non-specific alignment with extremely
low genome coverage and biased read alignment localisation. Scores between 100 and 1000 are
in the grey zone, i.e. the genome was correctly identified but could lead to invalid taxonomic
assignment or unprecise taxonomy at the genus level. Scores above 1000 correspond to a well-
defined genome (high coverage or high genome coverage distribution) with non-ambiguous
alignments. The aim of the score is therefore to provide a way to discriminate taxonomic
assignment quality. It is not possible to use the score to select or emphasise relevant organisms
for infections. The score remains a good indicator for background noise detection or nucleic acid
contamination identification. Two types of bacteria always have a high score: commensal
bacteria located on the skin (e.g. Propionibacterium acnes) and reagent contaminants (e.g.
Escherichia coli).
The score was generated according to the following formula:
score=genom ecoverage∗genomecoverage segment number∗genomedistribution∗100
10
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
1920
Genome coverage: percentage of nucleotides of the genome that are covered by at least
one read or alignment
Genome coverage segment number: number of continuous regions of the genome where
all nucleotides are covered by at least one read or alignment
Genome distribution: metric that measures the dispersion of all alignments along the
considered genome
The genome distribution metric depends on the genome coverage segment number and we
assumed that the coverage metric follows a discrete uniform distribution. We used this
distribution to position expected block positions (attractors) on the targeted genome. We then
computed the consistency between observed and expected block positions using a Pearson’s
chi-squared test (see supplementary figure 1).
Consequently, the genome distribution variable is a good indicator of bias regarding genome
coverage. This metric is directly correlated to the number of blocks, so it cannot be computed if
the genome coverage is too low or too high. If the genome coverage was <4 blocks, the value
was set to 0.01. If the genome coverage was > 50%, the value was set to 1.0. In this case, the
binomial distribution may be approximated by a normal distribution. Supplementary Figure 2
presents three genome coverage organisations that generate a low consistency metric, an
intermediate score and a perfect accordance with the uniform distribution. The score will
increase accordingly, reflecting the confidence in the taxonomic assignment for this
genome/target.
Python codes for scoring procedures are available on demand.
Measures taken to prevent or minimise contamination
Untargeted next-generation sequencing (UNGS) is a highly sensitive technique that can
potentially detect minute quantities of nucleic acid. A number of measures were therefore
implemented in the workflow to reduce the risk of contamination from each of the following
different sources:
11
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
2122
1. Pre-amplification contamination from the laboratory environment: Cross-contamination
is a major issue for UNGS of libraries prepared by random amplification. Indeed, each
individual nucleic acid molecule has the potential to be amplified to the extent that it
will be detected by the sequencing. Therefore, the utmost precautions were taken
regarding the working environment and sample handling during the process (e.g.
nucleic acid decontamination before and after handling the samples).
2. Contamination by nucleic acids present in laboratory reagents: Although traces of
nucleic acid contamination may be present in research products such as nucleic acid
purification spin columns, the majority comes from enzymes. Every single enzyme used
in the described procedures (with the exception of nucleases which are degrading
nucleic acids) contains traces of the bacterial genome of the host in which it was
produced. There is currently no nucleic acid-free enzyme commercially available that
could be used in our methodology. Consequently, the enzyme contamination was
assessed, in order to exclude it from the results. For this reason, Escherichia coli hits
were excluded from the report as it is the preferred host for the expression of
recombinant proteins such as enzymes.
3. Sample cross-contamination: The risk of cross-contamination occurs during the
handling of plates or tubes containing several samples. This is in particular a risk for
samples tagged with the same barcode for sequencing. During this study, an interval of
2-3 months was therefore observed between using the same barcode.
12
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
2324
Supplementary tables
13
157
158
2526
Supplementary Table 1. Conventional microbiological methods used in the standard
procedures
Microbiological method/test
Bacterial diagnosis Blood culture and culture of others samples on appropriate media
(bacterial identification by MALDI-TOF mass spectrometry)
Serological diagnosis
Urine: Soluble Legionella and pneumococcal antigen detection
Respiratory samples: PCR Mycoplasma pneumoniae, Chlamydia
pneumoniae, Bordetella pertussis/parapertussis, Legionella
pneumophila
Urogenital samples: PCR Neisseria gonorrhoeae, Mycoplasma spp.,
Chlamydia trachomatis
Stool samples: PCR Clostridium difficile
PCR (any sample): 16S rRNA gene sequencing, PCR Bartonella
henselae, Helicobacter pylori, Staphylococcus aureus, Streptococcus
pyogenes, Kingella kingae, Listeria monocytogenes, Neisseria
meningitidis
Viral diagnosis Serological diagnosis
PCR influenza A/B, rhinovirus, enterovirus, RSV A/B,
metapneumovirus, parainfluenzae 1,2,3,4, coronavirus, HSV 1/2, VZV,
JC virus, BK virus, CMV, EBV, parvovirus B19, norovirus, adenovirus,
measles-virus, hepatitis virus A/B/C/Delta/E
Fungal diagnosis Direct examination
Culture and fungal identification by MALDI-TOF mass spectrometry
Galactomannan antigen detection in blood, detection of cryptococcal
antigen, 1,3 B-D glucan detection in blood
PCR: Aspergillus fumigatus, Pneumocystis jirovecii, Toxoplasma gondii,
Microsporidium
Abbreviations: CMV, HHV5, Cytomegalovirus; HSV, HHV1/2, Herpes Simplex Virus 1/2; MALDI-TOF, Matrix Assisted Laser Desorption Ionization – Time-of-Flight; RSV, Respiratory Syncytial Virus; VZV, Varicella Zoster Virus.
14
159
160
161162
2728
Supplementary Table 2. Concordant identifications by UNGS and standard procedures at inclusion
Patient
identification
number
Immunodeficiency Clinical
presentation
and/or site of
infection
Pathogen
identified
Microbial
identification
by standard
procedures on
a sample also
tested by UNGS
Microbial
identification
by standard
procedures on
a sample not
tested by UNGS
Microbial
identification
by UNGS
Likelihood of
infectiona
1 HIV infection Fever Virus HIV 1 0 1 1
3 Chemotherapy Febrile
neutropenia
No diagnosisb 0 0 0 1
5 HSCT Febrile
neutropenia
No diagnosis 0 0 0 1
7 SOT Pleurisy No diagnosis 0 0 0 0
8 SOT Skin lesion No diagnosis 0 0 0 0
9 SOT Pneumonia No diagnosis 0 0 0 1
10 Other Pneumonia No diagnosis 0 0 0 1
11 PI Pneumonia No diagnosis 0 0 0 0
12 HSCT Fever No diagnosis 0 0 0 1
14 SOT Pneumonia No diagnosis 0 0 0 1
15
163
2930
Patient
identification
number
Immunodeficiency Clinical
presentation
and/or site of
infection
Pathogen
identified
Microbial
identification
by standard
procedures on
a sample also
tested by UNGS
Microbial
identification
by standard
procedures on
a sample not
tested by UNGS
Microbial
identification
by UNGS
Likelihood of
infectiona
163132
Patient
identification
number
Immunodeficiency Clinical
presentation
and/or site of
infection
Pathogen
identified
Microbial
identification
by standard
procedures on
a sample also
tested by UNGS
Microbial
identification
by standard
procedures on
a sample not
tested by UNGS
Microbial
identification
by UNGS
Likelihood of
infectiona
15 Chemotherapy Sub-cutaneous
abscess
Staphylococcus
aureus
1 0 1 1
20 SOT Arthritis No diagnosis 0 0 0 0
21 Other Pneumonia No diagnosis 0 0 0 1
22 Chemotherapy Febrile
neutropenia +
pneumonia
No diagnosis 0 0 0 1
23 Other Fever No diagnosis 0 0 0 1
28 SOT Uveitis Human Herpes
Virus type 5
(CMV)
1 0 1 1
30 Chemotherapy Skin lesion No diagnosis 0 0 0 1
31 Chemotherapy Febrile No diagnosis 0 0 0 1
173334
Patient
identification
number
Immunodeficiency Clinical
presentation
and/or site of
infection
Pathogen
identified
Microbial
identification
by standard
procedures on
a sample also
tested by UNGS
Microbial
identification
by standard
procedures on
a sample not
tested by UNGS
Microbial
identification
by UNGS
Likelihood of
infectiona
neutropenia
183536
Patient
identification
number
Immunodeficiency Clinical
presentation
and/or site of
infection
Pathogen
identified
Microbial
identification
by standard
procedures on
a sample also
tested by UNGS
Microbial
identification
by standard
procedures on
a sample not
tested by UNGS
Microbial
identification
by UNGS
Likelihood of
infectiona
32 Other Fever No diagnosis 0 0 0 1
34 Other Fever No diagnosis 0 0 0 1
35 HSCT Febrile
neutropenia
No diagnosis 0 0 0 1
36 PI Pneumonia No diagnosis 0 0 0 1
38 Chemotherapy Febrile
neutropenia
No diagnosis 0 0 0 1
39 SOT Skin infection Human Herpes
Virus type 3 (VZV)
1 0 1 1
40 SOT Skin infection Pseudomonas sp. 1 0 1 1
43 PI Pneumonia No diagnosis 0 0 0 1
44 Chemotherapy Pneumonia No diagnosis 0 0 0 1
46 SOT Fever No diagnosis 0 0 0 0
193738
Patient
identification
number
Immunodeficiency Clinical
presentation
and/or site of
infection
Pathogen
identified
Microbial
identification
by standard
procedures on
a sample also
tested by UNGS
Microbial
identification
by standard
procedures on
a sample not
tested by UNGS
Microbial
identification
by UNGS
Likelihood of
infectiona
48 HSCT Flu like
syndrome
Influenza virus A 0 0 0 1
53 HSCT Disseminated
infection (lung,
muscle, skin)
Nocardia sp. 0 0 0 1
57 SOT Pneumonia No diagnosis 0 0 0 1
58 PI Fever No diagnosis 0 0 0 0
59 SOT Pneumonia Pseudomonas
aeruginosa
1 0 1 1
60 HSCT Liver lesion
(GVH)
No diagnosis 0 0 0 0
61 SOT Pneumonia Stenotrophomonas
maltophilia
1 0 1 1
203940
Patient
identification
number
Immunodeficiency Clinical
presentation
and/or site of
infection
Pathogen
identified
Microbial
identification
by standard
procedures on
a sample also
tested by UNGS
Microbial
identification
by standard
procedures on
a sample not
tested by UNGS
Microbial
identification
by UNGS
Likelihood of
infectiona
62 Other Fever No diagnosis 0 0 0 0
63 HSCT Febrile
neutropenia
No diagnosis 0 0 0 1
64 Other Pneumonia No diagnosis 0 0 0 1
65 Chemotherapy Pneumonia No diagnosis 0 0 0 1
66 Chemotherapy Pneumonia No diagnosis 0 0 0 1
68 Other Fever No diagnosis 0 0 0 1
69 Other Myositis No diagnosis 0 0 0 1
70 PI Pneumonia No diagnosis 0 0 0 1
71 SOT Pneumonia Streptococcus
anginosus
1 0 1 1
72 Other Laryngitis No diagnosis 0 0 0 1
76 Other Pneumonia No diagnosis 0 0 0 1
214142
Patient
identification
number
Immunodeficiency Clinical
presentation
and/or site of
infection
Pathogen
identified
Microbial
identification
by standard
procedures on
a sample also
tested by UNGS
Microbial
identification
by standard
procedures on
a sample not
tested by UNGS
Microbial
identification
by UNGS
Likelihood of
infectiona
77 SOT Febrile
neutropenia
No diagnosis 0 0 0 1
81 Chemotherapy Meningitis No diagnosis 0 0 0 0
83 SOT Liver lesion No diagnosis 0 0 0 1
86 Chemotherapy Colitis No diagnosis 0 0 0 1
88 Chemotherapy Febrile
neutropenia
No diagnosis 0 0 0 1
89 PI Pneumonia No diagnosis 0 0 0 1
90 Chemotherapy Fever No diagnosis 0 0 0 1
91 Other Fever No diagnosis 0 0 0 1
92 PI Febrile
adenomegalies
No diagnosis 0 0 0 0
94 HSCT Febrile No diagnosis 0 0 0 1
224344
Patient
identification
number
Immunodeficiency Clinical
presentation
and/or site of
infection
Pathogen
identified
Microbial
identification
by standard
procedures on
a sample also
tested by UNGS
Microbial
identification
by standard
procedures on
a sample not
tested by UNGS
Microbial
identification
by UNGS
Likelihood of
infectiona
neutropenia
96 Other Fever No diagnosis 0 0 0 1
98 Chemotherapy Fever No diagnosis 0 0 0 1
101 Other Surgical site
infection
Finegoldia magna 1 0 1 1
102 Other Skin infection No diagnosis 0 0 0 1
103 Other Fever No diagnosis 0 0 0 1
Abbreviations: CMV, Cytomegalovirus; HSCT, hematopoietic stem cell transplantation; PI, primary immunodeficiency; SOT, solid organ transplantation; UNGS, untargeted next-
generation sequencing.
a Evaluated at the end of the trial using all available information.
b No diagnosis: No clinically-relevant viruses or bacteria were identified as being the agents responsible for the patient’s symptoms after consideration of all the available data and
documentation by the expert panel of clinicians.
23
164
165
166
167
168
4546
Supplementary Table 3. Divergent identifications by UNGS and standard procedures at
inclusion
24
169
170
4748
Pathogen Microbial
identification by
standard
procedures on a
sample also
tested by UNGS
Microbial
identification by
standard
procedures on a
sample not
tested by UNGS
Microbial
identification
by UNGS
Likelihood of
infectiona
Candida albicans 1 0 0 1
Aspergillus sp. 0 1 0 1
Aspergillus sp. 0 1 0 1
Clostridium sp. 0 1 0 1
Nocardia farcinica 0 1 0 1
Proteus mirabilis 0 1 0 1
Human Herpes
Virus type 1
(HSV1)
0 1 0 1
Candida albicans 0 1 0 1
Rhinovirus A 0 0 1 1
Porphyromonas
gingivalis
0 0 1 1
Adenovirus B 0 0 1 1
Pseudomonas sp. 0 0 1 1
Microsporidium sp. 0 1 0 1
Microsporidium sp. 0 1 0 1
Enterobacter
aerogenes
0 1 0 1
Klebsiella
pneumoniae
0 1 0 1
Pseudomonas sp. 0 0 1 1
Pseudomonas sp. 0 0 1 1
Pseudomonas sp. 0 0 1 1
254950
Pathogen Microbial
identification by
standard
procedures on a
sample also
tested by UNGS
Microbial
identification by
standard
procedures on a
sample not
tested by UNGS
Microbial
identification
by UNGS
Likelihood of
infectiona
265152
Pathogen Microbial
identification by
standard
procedures on a
sample also
tested by UNGS
Microbial
identification by
standard
procedures on a
sample not
tested by UNGS
Microbial
identification
by UNGS
Likelihood of
infectiona
Pseudomonas sp. 0 0 1 1
Aspergillus sp. 1 0 0 1
Pseudomonas sp. 0 0 1 1
Proteus mirabilis 0 1 1 1
Pseudomonas sp. 0 0 1 1
Influenza virus A 0 0 0 1
Campylobacter
jejuni
0 1 0 1
Human rhinovirus 0 1 0 1
Enterococcus
faecium
0 1 0 1
Staphylococcus
haemolyticus
0 1 0 1
Stenotrophomonas
maltophilia
0 1 0 1
Staphylococcus
warneri
0 0 1 1
Pseudomonas sp. 0 0 1 1
Nocardia sp. 0 0 0 1
Escherichia coli 0 1 0 1
Streptococcus
pneumoniae
0 1 0 1
275354
Pathogen Microbial
identification by
standard
procedures on a
sample also
tested by UNGS
Microbial
identification by
standard
procedures on a
sample not
tested by UNGS
Microbial
identification
by UNGS
Likelihood of
infectiona
Streptococcus
intermedius
0 0 1 1
Pseudomonas sp. 0 0 1 1
Escherichia coli 1 0 0 1
Pseudomonas sp. 0 0 1 1
Streptococcus
pneumoniae
0 0 1 1
Pseudomonas sp. 0 0 1 1
Pseudomonas sp. 0 0 1 1
Enterococcus
faecium
0 1 0 1
Pseudomonas sp. 0 0 1 1
Pseudomonas sp. 0 0 1 1
Pseudomonas sp. 0 0 1 1
Pseudomonas sp. 0 0 1 1
Stenotrophomonas
maltophilia
0 0 1 1
Streptococcus
pneumoniae
0 0 1 1
Acinetobacter
baumannii
0 0 1 1
Pseudomonas sp. 0 0 1 1
285556
Pathogen Microbial
identification by
standard
procedures on a
sample also
tested by UNGS
Microbial
identification by
standard
procedures on a
sample not
tested by UNGS
Microbial
identification
by UNGS
Likelihood of
infectiona
Acinetobacter
baumannii
0 0 1 1
Streptococcus
pneumoniae
0 0 1 1
Bacteroides fragilis 0 0 1 1
Finegoldia magna 0 0 1 1
Anaerococcus
prevotii
0 0 1 1
Streptococcus
parasanguinis
0 0 1 1
CMV 1 0 0 1
Acinetobacter
baumannii
0 0 1 1
Aspergillus sp. 0 1 0 1
Abbreviations: CMV, HHV5, Cytomegalovirus; UNGS, untargeted next-generation sequencing.
a Evaluated at the end of the trial using all available information.
29
171
172
173
5758
Supplementary Table 4. Clinically-relevant viruses and bacteria in samples with
concordant results for UNGS and standard procedures
Number of diagnoses
(N=9)
All Antimicrobial
treatment
Viruses
HIV 1
CMV 1 Ganciclovir IV
VZV 1 Aciclovir IV
All Active
antimicrobial
therapy
Inactive
antimicrobial
therapy
No therapy
Bacteria
Staphylococcus aureus 1 1 0 0
Pseudomonas spp. 2 1 1
Stenotrophomonas
maltophilia
1 1
Streptococcus spp. 1 1
Finegoldia magna 1 1
Abbreviations: CMV, HHV5, Cytomegalovirus; HIV, Human Immunodeficiency Virus; UNGS, untargeted next-
generation sequencing; VZV, Varicella Zoster Virus.
30
174
175
176
177
178
5960
Supplementary Table 5. Comparison of UNGS at inclusion with standard procedures
running for 30 days for clinically-relevant viruses, bacteria and fungi
Standard procedures-
positivea
Standard procedures-
negative
Total no. of
patients (%)
UNGS-positive 13b 23 36 (36)c
UNGS-negative 13 52b 65 (64)
Total no. of patients (%) 26 (26)c 75 (74) 101
Abbreviations: CI, confidence interval; SP, standard procedures; UNGS, untargeted next-generation sequencing.
a Positive: patient with a microbiological diagnosis.
b UNGS and SP were concordant for 65 of 101 patients (Kappa test=0.17 [95% CI,-0.02 to 0.37]).
c The detection rate of CRVB by UNGS and SP was not significantly different based on the McNemar test: P=0.133.
31
179
180
181
182
183
184
185
6162
Supplementary figures
Figure S1. Genome distribution metric calculation
Figure S2. Three different genome coverage and distribution models
32
186
187
188
189
190
191
192
6364