1.
Web applications for rapid
microbial taxonomy
identification
Ole Lund
Center for Genomic Epidemiology

2.
Whole genome sequence based
Diagnostics
Infectious diseases are responsible for >25% of
all global deaths
An increasing number of infectious diseases have
a global epidemiology (e.g. SARS, avian flu,
influenza, Salmonella etc.).
Rapid detection, identification and exchange of
comparable information between public health
laboratories globally, are crucial to avoid or
control global and local spread.

3.
Sample Antibiotic
resistance
Culturing ID Typing
1-2 days 1-2 days 1-2 days 1 – several weeks
Routine microbial diagnostic

4.
Sample Culturing
ID
Resistance
Typing
+
Much
more
1-2 days ½-1 day
Whole genome sequence based diagnostic

5.
Bacterial genomics
• Sequencing a bacterial genome cost ~$100 (on
a desk top sequencer)
• Equipment will cost less than $100 000
• In Denmark 1 million clinical microbiology
isolates are handled each year
– EU/USA ~100 million
– Globally ~ 1 billion (10 billion needed)
• Future limiting factor will not be sequencing
but handling the sequences

6.
K-mer based method works well for
species identification
Benchmarking of methods for genomic taxonomy. Larsen MV, Cosentino S, Lukjancenko O, Saputra D, Rasmussen S, Hasman
H, Sicheritz-Pontén T, Aarestrup FM, Ussery DW, Lund O. J Clin Microbiol. 2014 May;52(5):1529-39.

7.
K-mers: Not a new idea

8.
Multilocus Sequence Typing of Total Genome Sequenced Bacteria. Larsen MV, Cosentino S, Rasmussen S, Friis C, Hasman H,
Marvig RL, Jelsbak L, Pontén TS, Ussery DW, Aarestrup FM, Lund O. J Clin Microbiol. 2012 Apr;50(4):1355-61.
MLST typing

9.
Genotyping using whole-genome sequencing is a realistic alternative to surveillance based on phenotypic antimicrobial susceptibility testing. Zankari E, Hasman H, Kaas RS, Seyfarth AM, Agersø Y, Lund O, Larsen MV,
Aarestrup FM. J Antimicrob Chemother. 2013 68:771-7.
Identification of acquired antimicrobial resistance genes. Zankari E, Hasman H, Cosentino S, Vestergaard M, Rasmussen S, Lund O, Aarestrup FM, Larsen MV. J Antimicrob Chemother. 2012 67:2640-4.
Antimicrobial resistance

10.
Pheno typing by machine learning
• Earlier methods are all based on alignment to a
database of genes with known (pheno-) types.
• Andreatta et al. took a radically different approach and
sorted genomes of Gamma-Proteobacteria into
pathogenic or non-pathogenic, and looked for gene
families that were statistically associated with either
pathogenic or non-pathogenic bacteria (Andreatta et
al. 2010).
• First example of using machine learning techniques to
determine the phenotype from WGS.
• Extended to work for all species of bacteria and using
raw sequencing data as input (Cosentino et al. 2013).

11.
User Statistics
Until now: 280.000 submissions

12.
To be added:
Upload to public repositories
(SRA/ENA)

13.
102.000 sequences analyzed in 16.000 submissions

14.
Phylogeny of the isolates by the NDtree method.
Joensen K G et al. J. Clin. Microbiol. 2014;52:1501-1510

15.
3.0
0508R6762_IonTorrent_2
0508R6707_IonTorrent_1
0508R6707_MiSeq_2
0507R6701_HiSeq
0508R6701_IonTorrent_1
0508R6701_IonTorrent_2
0508R6701_MiSeq_2
0508R6762_MiSeq_1
0508R6762_IonTorrent_1
0508R6762_HiSeq
NCTC_13348_IonTorrent_2
0508R6707_IonTorrent_2
NCTC_13348_IonTorrent_1
0508R6762_MiSeq_2
NCTC_13348_MiSeq_1
NCTC_13348_MiSeq_2
0508R6707_HiSeq
0508R6707_MiSeq_1
0508R6701_MiSeq_1
NCTC_13348_MiSeq_3
0.5
0508R6762_IonTorrent_2
0508R6707_MiSeq_1
0508R6707_IonTorrent_1
NCTC_13348_MiSeq_1
NCTC_13348_MiSeq_3
0508R6701_MiSeq_1
0508R6762_MiSeq_2
0508R6701_IonTorrent_1
0508R6707_HiSeq
0508R6762_HiSeq
NCTC_13348_MiSeq_2
0508R6707_IonTorrent_2
NCTC_13348_IonTorrent_1
0508R6701_IonTorrent_2
0508R6762_MiSeq_1
0508R6762_IonTorrent_1
NCTC_13348_IonTorrent_2
0508R6701_MiSeq_2
0508R6707_MiSeq_2
0507R6701_HiSeq
0508R6762_IonTorrent_2
0508R6762_HiSeq
0507R6701_HiSeq
0508R6707_IonTorrent_1
0508R6701_MiSeq_1
0508R6707_IonTorrent_2
0508R6707_MiSeq_2
NCTC13348_Miseq1
0508R6707_MiSeq_1
0508R6701_MiSeq_2
0508R6707_HiSeq
NCTC13348_IonTorrent1
NCTC13348_IonTorrent2
0508R6762_MiSeq_2
0508R6762_MiSeq_1
0508R6762_IonTorrent_1
NCTC13348_Miseq3
0508R6701_IonTorrent_1
NCTC13348_Miseq2
0508R6701_IonTorrent_2
SNPtree CSIPhylogeny NDtree
Close
reference
Remote
reference
Salmonella
Typhimurium
DT104
PLoS One. 2014 Aug 11;9(8):e104984.

16.
Controlled evolution
Johanne Ahrenfeldt, Submitted

17.
Phylogenetic tree using neighbor
joining
Johanne Ahrenfeldt, Submitted

18.
Outbreak analysis of billions of strains:
Real-time tracking of all microbial genomes
• OX values
• O10
– Number of earlier isolates (from within the last
year) with less than 10 SNP differences to the
current isolate
• Do not need to be updated
• Mapped genomes may be stores as binary files
• Search can/should be restricted to those that cluster to
the same template

19.
Evergreen Trees
• User submitted samples
compared against all close-
matching sample clusters
• Ever growing trees from the
clusters
• Users can see all previous
samples their sample is closely
related to

20.
Global Data Exchange
Global repositories*
* Providers you will bet your life on will provided High bandwidth programic access to
deposition/retrieval forever: SRA/ENA/??
Hospital
Food safety agency
National CDC
Analysis www servers
Sequence + Meta data
Animal health

21.
Sample
ID
Resistance
Type
+
Every
thing
½-1 day
Metagenomic based diagnostic

22.
Sample
ID
Resistance
Type
+
Every
thing
minutes
Metagenomic based diagnostic with non batch
mode sequencing (nanopore technologies)

23.
Rapid whole genome sequencing for the detection and characterization of microorganisms directly from clinical samples. Hasman H, Saputra D, Sicheritz-Ponten T, Lund O, Svendsen CA, Frimodt-Møller N, Aarestrup
FM. J Clin Microbiol. 2014 Jan;52(1):139-46

24.
It is not important to know where you
are but where you are not
• Analysis of absence/presence of specific
strains/species may be more important for
diagnosis of infectious diseases than the
general composition of the microbiome
normally associated with metagenomics

25.
Whole genome sequencing
• Is it a game changer in the combat against infectious
diseases
• Game changer? - what is new with WGS?
– Typing with ultimate resolution (bar epigenetics?)
• Resolution = 1/mutation rate = 1 year
– Can (soon) be done in a day
– Instant deep pheno-typing (e. g. resistance/virulence
genes)
– With falling prices surveillance may be ubiquitous
• Everything is under constant surveillance
– People, animals, planes, places, doorhandles …
– Information can be shared instantly around the globe

26.
Transmission do not have to be zero
• But R0:
– The number of secondary infections that a case on
the average give rise to
• Have to be below 1

27.
Game changer?
• Can WGS + IT be used to set R0 to less than 1
for some pathogens in some areas?
• Which are the best cases?

28.
Thanks
DTU Systems Biology/CBS/Lund group
Mette Voldby Larsen
Martin Thomsen
Johanne Ahrenfeldt
Vanessa Jurtz
Jose L. Bellod Cisneros
Johanne Ahrenfeldt
Anna Maria Malberg Tetzschner
Ex members
Salvatore Consentiono
Student helpers
Jamie Neubert Pedersen
Valentin Ibanez
Rosa Allesøe
Camilla Lemvigh
DTU Systems Biology/CBS
Dave Ussery
Thomas Ponten
Dhany Saputra
Simon Rasmussen
Thomas Nordahl Petersen
DTU DMAC
Laurent Gautier
Marlene Dalgaard
DTU Food
Frank Aarestrup
Henrik Hasman
Rene S. Hendriksen
Shinny Leekitcharoenphon
Rolf Sommer Kaas
Marlene Hansen
Katrine Grimstrup Joensen
Oksana Lukjancenko
Copenhagen University/CMP
Thor Theander
Michael Alifrangis
Sidsel Nag
KCMC Moshi, Tanzania
Gibson Kibiki
Happiness Kumburu
Tolbert Sonda