Upload
others
View
7
Download
0
Embed Size (px)
Citation preview
From takahē and microarrays to the Reverend Bayes
mEpiLab
Nigel French, Patrick Biggs and Jonathan Marshall
IDReC • An interdisciplinary
research centre at Massey University.
• IDReC brings together 6 research groups each with knowledge and expertise in different aspects of infectious disease research.
IDReC
Epidemiology and molecular epidemiology
Population genetics and
disease ecology
Mathematical modelling and
statistics Public health
Microbiology
IDReC’s Six Principal Investigators Professor Nigel French IDReC Director mEpiLab – Palmerston North Molecular epidemiology and veterinary public health
Professor Paul Rainey Rainey Lab, Auckland Max Planck Institute, Germany Evolutionary genetics
Professor Martin Hazelton Statistics and Bioinformatics Group – Palmerston North Statistics and spatial modelling
Professor Tim Carpenter Epicentre – Palmerston North Veterinary epidemiology and economics
Professor Mick Roberts Institute of Information & Mathematical Sciences - Auckland Mathematical biology and infectious disease modelling
Professor Jeroen Douwes Centre for Public Health Research - Wellington Public health and epidemiology
Scientific Advisory Group
Molecular Epidemiology and Public Health Laboratory
– Epidemiology, evolution and control of infectious diseases
• Particularly zoonotic diseases • Campylobacter, Salmonella, E. coli, Leptospira,
Cryptosporidium, Giardia, and multidrug resistant bacteria
– Located in the Hopkirk Research Institute
Zoonoses and Emerging Infectious Diseases
• ~60% of ~1500 human infectious diseases are zoonotic
• ~75% of human emerging infectious diseases are zoonotic
• 33% of zoonoses are transmissible between humans
• Many zoonoses don’t affect their animal host
7
Zoonoses in perspective
• Emerging and re-emerging zoonoses – Global pandemics, high mortality (SARS, H5N1,
H1N1, Nipah…..)
– Major burden to health services and economy
• Outbreaks grab attention, but many persisting – Rabies kills 55,000 per year
Single and multi-host pathogens
• Most infectious agents are multi-host – Particularly emerging ones – Subset of these are the zoonoses
• Single host agents relatively rare – Subset of these are the solely human to
human spread agents (anthroponoses) – Evolved once population reached critical size
Emerging Infectious Disease: why the increase?
• Human risk factors – Population density,
urbanisation and growth – Increased global travel – Poverty – Changing dietary habits
• Animal risk factors – Farm animals and
wildlife • Habitat destruction • Global distribution of food • Climate change
Zoonoses figure very prominently in New Zealand. Source: ESR Ltd
Pathways and pathogens
E. coli O157 ~ 150 cases
Salmonella DT160 ~ 250 cases
Campylobacter ~ 8,000 cases
Cryptosporidium ~ 800 cases
Giardia ~ 1000 cases Leptospira ~ 100 cases
Molecular epidemiology: tools • Microbiology • Molecular biology
– PCR – Sequencing – Microarray
• Genomics and bioinformatics • Modelling
– Space and time – Case-control / risk factor – Evolution – Source attribution
Campylobacteriosis
Source: Olsen et al Campylobacter. 3rd ed. Washington DC: ASM Press; 2008 .
Campylobacter from space... Google Earth 2000-10 110,000 cases
C. Jewell
Manawatu sentinel site 2005-2012 • Identify genotypes common to
particular sources • Modelling (reservoir attribution)
Campylobacter reservoir attribution
• Multilocus sequence
typing • Population genetics
approach • Evolutionary modelling • Used to find out source
of human infections Human
Wild bird
Chicken Sheep
Cow
Water
Reservoir attribution (Mullner et al, Sears et al)
Est $40M saving per annum to NZ economy
Control of Campylobacter in poultry
Changing epidemiology presents new challenges
0
1
2
3
4
5
6
7
00 00 00 00 01 01 01 01 02 02 02 02 03 03 03 03 04 04 04 04 05 05 05 05 06 06 06 06 07 07 07 07 08 08 08 08 09 09 09 09
1 3 5 7 9 11 1 3 5 7 9 11 1 3 5 7 9 11 1 3 5 7 9 11 1 3 5 7 9 11 1 3 5 7 9 11 1 3 5 7 9 11 1 3 5 7 9 11 1 3 5 7 9 11 1 3 5 7 9 11
Urban
Rural no cattle
Rural high cattle
Poultry intervention
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Seasonality and dairy density
Urban
High dairy rural Young children During calving season Direct contact
Hand mouth behaviour Epidemiology changed
Different set of policy and decision makers Different stakeholders
0
5
10
15
20
25
30
Jan-
97Ju
n-97
Nov
-97
Apr-
98Se
p-98
Feb-
99Ju
l-99
Dec-
99M
ay-0
0O
ct-0
0M
ar-0
1Au
g-01
Jan-
02Ju
n-02
Nov
-02
Apr-
03Se
p-03
Feb-
04Ju
l-04
Dec-
04M
ay-0
5O
ct-0
5M
ar-0
6Au
g-06
Jan-
07Ju
n-07
Nov
-07
Apr-
08Se
p-08
Feb-
09Ju
l-09
Dec-
09M
ay-1
0O
ct-1
0M
ar-1
1Au
g-11
Jan-
12
VTEC
Number of cases
Date
E. coli O157 VTEC/STEC (mainly O157)
Major impact on export trade Super eight
P. Jaros H. Irshad B. Govan
A. Cookson A. Reynolds
Ecosystem goods and services: Water supply and purification
Important for public health
…and recreation
Water-borne protozoan infections ~ 1800 cases per year in NZ Major global issue
Water quality and safety
Second order properties
+Deb Prattley
Alex Grinberg Mark Stevenson
Rima Shrestha Ben Phiri
Eve Pleydell Julanda Al Malwy Pete Lockhart
Anthony Pita
Whole genome: 80% of sample to be sequenced
Sequenced on Illumina MiSeq system
16S rRNA V4 region PCR amplification
18S rRNA PCR amplification; region to be determined
DNA extracted from water samples
Metagenomics
Leptospirosis: an occupationally acquired zoonosis in NZ • 81 notified cases in 2010
– underdiagnosed – 75 reported occupation
• 41 farmers • 14 meat industry
– Half hospitalised
RWNZ, Massey and leptospirosis.... a one health approach
Fang Fang, Anou Dreyfus, Emilie Vallee, Alison Harland Peter Wilson, Julie Collins-Emerson, Anne Midwinter
Wildlife and Emerging Infectious Disease
• Wildlife important source – Human encroachment,
urbanisation – Habitat destruction – Climate change
Nipah virus
Salmonella DT160
Conservation, translocation and wildlife health • Major challenges
– Declining habitats – Small, fragmented populations
• Low immune repertoire • Immunologically naive
– Translocation • Host and its parasites • Transmission of infectious agents • Treatments may reduce biodiversity • Stress host and alter immune system
Supershedders and superspreaders
Kyle Richardson and Dan Tompkins
Population biology and evolution of zoonotic pathogens in NZ
• Low diversity • Unique genotypes • Population structure shaped by:
– Hitchhiking – immigration, people and animals • Isolation, recent colonisation by man
– Subsequent evolution
EID 7, 767
Campylobacter: Mechanisms of evolution
• Recombination more important than mutation – Natural transformation – Transduction
• Using MLST and whole genome data
C. jejuni C. coli
EM algorithm
Next (Now) generation sequencing
• Human and poultry isolates, same genotype, time and space.
• 83 genes differed, 55 with amino acid differences.
• 96.7% were imported via recombination. • Recombination much more important
than mutation for generating divergence.
ST 474 flaA 14
Non-homologous recombination
2011
Livestock and wildlife populations Barbara Binney
Cattle
Modelling demographic history
Tim Vaughan Alexei Drummond Paul Fearnhead Dennis Prangle
Campylobacter source attribution: water Most bacteria from water in Manawatu associated with wildlife – even in dairy catchments
Water birds
Sheep Cattle
New Zealand water rail lineages
Nearest to common ancestor
NJ tree glnA allele
Phil Carter, ESR
7,000yrs
Divergence time of “C. aotearoa” and other NZ lineages
100yrs
500yrs
10yrs
Analysis in BEAST
Dr Patrick Biggs Genomic epidemiology and public health
Bacterial sequence typing • Goal is to use typing methods of bacteria to inform molecular
and/or genomic epidemiology • Molecular bacterial typing methods
– “Generation 1”: MLST 1990’s – “Generation 2”: rMLST 2010’s
• MLST (multilocus sequence typing) – Used sequencing technology of the time to sequence portions of
housekeeping genes
• rMLST (ribosomal MLST) – Uses high throughput sequencing of whole genomes to analyse many
more genomic loci – Higher resolution
MLST in Campylobacter jejuni/coli • Both species share same
MLST scheme – Other schemes for other
bacteria
• PCR highly conserved genes
• 7 genes used • Used to define:
– ST = sequence type – unique pattern of 7 genes
– CC = clonal complex – group of related STs
The PubMLST website • Lists all genomes analysed by Prof Martin Maiden’s
group in Oxford, UK • Other sites exist for other bacteria • For each scheme,
this shows: – Allele definitions,
number and amplification conditions
– Isolates for that scheme
Ribosomal MLST (rMLST) • Proposed in 2012 as typing scheme to integrate
genealogy and typing – Synthetic – needs whole genome sequence for
analysis • Use of the 53 genes encoding bacterial ribosome
subunits (rps) – Present in all bacteria – Distributed around chromosome – Under stabilising selection
• More resolution than: – MLST approaches – 16S rRNA approaches
Microbiology (2012) 158, 1005-1015
rMLST for Campylobacter jejuni/coli • Only 52 of the 53 genes
found – No rpmD gene
• Few genes distributed in clusters around genome
• Genomes uploaded and analysed on a remote copy (UK) of the BIGSdb database set up for our project in collaboration with Prof Martin Maiden’s group in Oxford, UK
rMLST for 62 C. jejuni/coli genomes
• SplitsTree image based on allele numbers (n =52)
• Maximum distance is therefore 52
Host type: Env water/wild birds Livestock/human gut
The core genome • Genes present in ALL
strains are core genes – A proxy for analysing
evolution
• Genes present in a subset of strains are accessory genes – Virulence? – Niche adaptation?
0
500
1000
1500
2000
2500
3000
3500
4000
1 2 3 4 5 6 7 8 9 10 11 12 13
num
ber o
f clu
ster
s
number of genes in a cluster
Core genome estimation from 13 NZ E. coli O157:H7 genomes
Accessory Character Core
Core genome for 76 C. jejuni/coli genomes
• Core genome NeighborNet using 548 genes with length range < 20 aa and variable sites.
• 148,046 aa sites with 36,528 variable (24.7%).
rMLST for Escherichia coli
• All 53 genes present • Significant clusters
of genes in the genome
• 13 O157:H7 genomes analysed: – 7 bovine – 6 human
rMLST for 13 NZ E. coli genomes • rMLST
NeighborNet image from 13 New Zealand E. coli O157:H7 genomes, with 21 reference genomes (3 non-O157 and 18 O157)
Core genome for 13 NZ E. coli genomes
• Core genome NeighborNet using 2216 genes with length range < 20 aa and variable sites present.
• 734,563 aa sites with 19,530 variable (2.65%). • In addition, 178,620 aa sites in a further 758 genes are invariant.
Integration of genotype with phenotype for E. coli
• Use of the Omnilog system
• Brown SBI type 5 with stx2c
• Sucrose non-fermenter
• Other studies (Lee et al. AEM 2012) have shown stx2c strains to be ‘stress sensitive’
Tessy George
Dr Jonathan Marshall
epiclustR
A Bayesian model for detecting outbreaks of disease from
notification data
NZ campylobacteriosis cases
NZ campylobacteriosis cases
epiclustR goals
• Model the underlying temporal trend.
• Model the underlying spatial trend.
• Identify potential ‘outbreaks’ as short periods of localised increased risk over and above these trends.
• Produce output for visualising these ‘outbreaks’ suitable for surveillance.
epiclustR models the expected number of cases within a particular area at a particular time using The rate is then split into time, space and space-time effects
epiclustR statistical model
epiclustR statistical model The random effects Rt and Ui ‘soak up’ the primary temporal and spatial variation. Rt Change in risk this week is similar to the change in risk last week. Ui Risk in a spatial location is similar to the risk in neighbouring locations. What is left over are the short periods of localized increased risk.
The influence of affluence
epiclustR detecting potential outbreaks
• Group fine spatial units into coarser regions.
• Associate a binary variable with each region, indicating whether a localized outbreak occurred.
• Prior information for these variables may be specified according to disease behaviour.
• The model output for these variables gives the probability of an outbreak.
epiclustR output
epiclustR output
epiclustR output
Raw milk cluster
Raw milk cluster
Raw milk cluster
epiclustR summary
• Identifies underlying spatial and temporal trends
• Identifies potential localised outbreaks over and above this trend.
• Produces output that can be visualised in a number of ways.
• May be a useful supplement to the current NZ disease surveillance tools.
Conclusion • Molecular epidemiology and public health – we work on:
• Understanding epidemiology of major zoonoses • Pathogen evolution (emerging infectious disease) • Control and prevention • Food safety • Protecting export markets • Ecosystem health (water supply) • Conservation
• ... Under ‘One Health’ model • In collaboration with ESR, AgResearch, Landcare, University of
Otago and members of the Allan Wilson Centre
Acknowledgements • mEpiLab
– Julie Collins-Emerson, Anne Midwinter, Jonathan Marshall, Eve Pleydell, Deb Prattley, Patrick Biggs
– Rebecca Pattison, Rukhshana Akhter, Errol Kwan, Lynn Rogers, Isabel Li, Sarah Moore, Angie Reynolds, Neville Haack
– mEpiLab PhD students – Tessy George
• Martin Hazelton, Chris Jewell • Brett Gartrell, Laryssa Howe • ESR - Phil Carter, Sharla McTavish, Elaine Moriarty , Stephen On,
ERL team • AgResearch –Adrian Cookson, Gale Brightwell, Helen Withers, Bryce
Buddle • University of Oxford: Martin Maiden, Danny Wilson, Sam Sheppard • MidCentral Public Health, Tui Shadbolt • Petra Mullner • MedLab Central • Landcare: Dan Tompkins • MPI – Peter van de Logt, Steve Hathaway, Donald Campbell, Roger
Cook • University of Otago: A/Prof Michael Baker, Simon Hales, Aparna Lal,
Ann Sears • VUW: Nicky Nelson • University of Tasmania: Barbara Holland • Lancaster University: Dennis Prangle and Paul Fearnhead • University of Warwick: Simon Spencer
Funding