ECCMID 2016 - How to build actionable virulome databases

João André Carriço,
Microbiology Institute and Instituto de Medicina Molecular,
Faculty of Medicine, University of Lisbon
jcarrico@fm.ul.pt twitter: @jacarrico
Session SY024 Controversies in interpreting whole genome sequence data
26th ECCMID, Amsterdam, Netherlands
7-12 April 2016

Virulence Factors:
 Class of gene products
 Help pathogens to invade the host and
evade specific host’s defensive mechanisms
 Enhance the pathogen’s potential to cause
disease

Virulence Factors (example):
 Bacterial toxins (Endotoxins and Exotoxins)
 Adherence factors (Pili)
 Cell surface carbohydrates and proteins that protect a
bacterium (Streptococcal M Protein)
 Hydrolytic enzymes that may contribute to the
pathogenicity of the bacterium (hyaluronidase)
 Factors to compete with host nutrient uptake
(Siderophores)
Sources:VFDB / Medical Microbiology. 4th edition. (http://www.ncbi.nlm.nih.gov/books/NBK7627/)

Virulome
Core genome
Accessory
genome
Mobilome

 VFDB (http://www.mgc.ac.cn/VFs/main.htm)
 Pathosystems Resource Integration Center (PATRIC)
VF (https)://www.patricbrc.org/)
 Victors (http://www.phidias.us/victors/)
 PHI-Base (http://www.phi-base.org/)
 MvirDB (http://mvirdb.llnl.gov/ )
Criteria for choice:
 Focused mainly on virulence factors DB (as defined in the first slide)
 excludes Antibiotic resistance databases (CARD, ARDB,ARGO, RAC,…)

* Created to facilitate the screening of HTS data
Database last update:
Tue Feb 23 22:05:25
2016

• 6 NIAID priority genera:
• Mycobacterium
• Salmonella
• Escherichia
• Shigella
• Listeria
• Bartonella
• 1572VFs
• 1071 articles
• Use of controlled vocabulary
• IntegratesVFDB andVictorsVF information
• PATRIC supports:
• Genome annotation
• Comparative Genomics
• Transcriptomics
• Pathways
• Host-pathogen interaction
• Disease-related information
• Database last update:
• March 2016
Pathosystems Resource Integration Center

• 5177Virulence Factors
• 126 Pathogens (class/#sp/#VFs):
• Gram + 15 1160
• Gram – 36 3488
• Virus 54 179
• Parasites 13 105
• Fungi 8 245
• Last DB Update: 27/8/2014

• pathogenicity, virulence and effector genes
• Fungal
• Oomycete
• bacterial pathogens
• Hosts:
• Animal
• Plant
• Fungal
• Insect hosts.

• Biodefense focused
• Last update 2007??
• Data still available for download..

 All the databases have:
 manually curated data
 links for the original publication
 However manual curation is a huge caveat
due to the sustainability of the process

 Querying annotation in the the website
 Selecting species of interest, and browsing
the website
 BLAST query for DNA or Protein

 Download the gene/protein databases and
use them as templates for searching own
data

 With HTS several core genome /whole genome MLST schemas are becoming available/being
developed:
 Neisseria sp.
 Campylobacter sp.
 Staphylococcus aureus
 Legionella pneumophila
 Listeria monocitogenes
 Enterococcus faecium
 Mycobacterium tuberculosis
 Acinetobacter baumannii
 Salmonella enterica
 E.coli
 ….
 Loci in these schemas can be annotated / linked to the Virulence Factor DBs for automatic
allele annotation through these systems
Seqsphere+
http://pubmlst.org/
http://bigsdb.web.pasteur.fr/
https://enterobase.warwick.ac.uk/
Bionumerics 7.5

 So far we have seen what is available
How can we design
actionable virulome databases ?
Actionable: able to be done or acted on; having practical value
New Oxford American Dictionary

 Available databases still lack interfaces for
programmatic access :
 RESTful APIs would allow:
▪ easy automatic querying from scripts without the need
of web interfaces or downloads
▪ Database updates by authorized groups (distributed
curation effort)
APIs : Application Programming Interfaces

 Existing DBs reuse each others datasets without true
database interoperability: need for common ontologies
(controlled vocabularies already exist but are not used by
all)
 Ontologies and computer readable data formats (json-
ld or RDF) can allow for true database interoperability
allowing bioinformaticians to extract the targeted
information from a single query reaching multiple
databases

Trends Microbiol 17, 279–285 (2009).

 Major problems of databases
 Manual curation still a necessity
 Academic model for sustainability of a resource:
lack of funding leads to “dead” databases

 Existing virulome databases provide a wealth of data
 A large part of the availableVF data overlaps between DBs.
The overlap largely depends of the last database update and
what was included.
 They are always aWork in Progress , heavily relying in
manual curation
 Novel HTS based techniques such as cg/wgMLST can use
this databases to annotate schemas and provide a much
richer picture ofVF diversity at DNA/Protein level.

 UMMI Members
 Mário Ramirez
 José Melo-Cristino
 EFSA INNUENDO Project (https://sites.google.com/site/innuendocon/)
 Mirko Rossi
 FP7 PathoNGenTrace (http://www.patho-ngen-trace.eu/):
 Dag Harmsen (Univ. Muenster)
 Stefan Niemann (Research Center Borstel)
 Keith Jolley, James Bray and Martin Maiden (Univ. Oxford)
 Joerg Rothganger (RIDOM)
 Hannes Pouseele (Applied Maths)
 Genome Canada IRIDA project (www.irida.ca)
 Franklin Bristow, Thomas Matthews, Aaron Petkau, Morag Graham and Gary Van Domselaar (NLM , PHAC)
 Ed Taboada and Peter Kruczkiewicz (Lab Foodborne Zoonoses, PHAC)
 Fiona Brinkman (SFU)
 William Hsiao (BCCDC)
INTEGRATED RAPID INFECTIOUS DISEASE ANALYSIS

ECCMID 2016 - How to build actionable virulome databases

More Related Content

What's hot (20)

Similar to ECCMID 2016 - How to build actionable virulome databases (20)

Recently uploaded (20)

ECCMID 2016 - How to build actionable virulome databases