This study aims to evaluate pathogen transmission rates under various simulated healthcare conditions using both traditional culturing and culture-independent metagenomic methods. Researchers designed experiments to directly and indirectly transmit pathogens between surfaces and hands at different concentrations, with and without washing. Preliminary results found direct transmission was more efficient than indirect transmission, and washing reduced transmission most on cotton. Further bioinformatics analysis will characterize detection limits and identify pathogen genes.
1. Validating Metagenomic
Analyses through Simulated
Direct and Indirect Healthcare-
Related Pathogen Transmissions
Krista Ternus
Katharina Weber, Nicolette Albright, Gene
Godbold, Veena Palsikar, Danielle LeSassier,
Nicole Westfall, Kathleen Schulte, Curt Hewitt
SFAF Meeting • 23 May 2019
2. Funding Acknowledgement
2
This work was
supported
by the Centers of
Disease Control
and Prevention’s
investments to
combat antibiotic
resistance
under award
number 200-2018-
75D30118 C02922
3. Research Study Objectives
3
• Evaluate how healthcare-associated pathogen transfer
rates change under a variety of simulated conditions
• Compare traditional culturing methods with the
culture-independent methods of shotgun
metagenomics and metranscriptomics for pathogen
detection and characterization
• Assess how pathogen abundance impacts detection of
virulence and antimicrobial resistance genes
• Identify potential signatures of pathogen viability
www.cdc.gov/cdiff/index.html
stock/123RF.com
4. 4
Two Types of Culture-Independent Methods
• Funding is incremental and reactive • Funding is upfront and proactive
• Works at inflexible, known locations • Works at virtually any location
• Strong signal in specific circumstances • Weak signals can be lost
• Know your goal ahead of time • Decide your goal in the moment
• Limited utility • Broad utility
• Currently less expensive if already in place • Very adaptable to emerging, new use cases
https://upload.wikimedia.org/wikipedia/commons/2/2e/2015-03-
16_14_11_28_Old_phone_booth_at_the_Northeastern_Nevada_Museum_in_Elko%2C_Nevada.JPG
Methods
developed for
specific pathogens
(e.g., amplicons)
Pathogen
agnostic methods
(e.g., metagenomics)
https://pixabay.com/p-2464968/?no_redirect
5. 5
Things that Change…
Funding to maintain
reference databases
Funding to detect
specific pathogens
https://upload.wikimedia.org/wi
kipedia/commons/e/eb/Zika-
information-virus.jpg
Emerging antimicrobial
resistance
Reference database
content
“Unique” regions
of microbial
genomes
Synthetic biology
techniques
https://www.arl.army.mil/www/articles/
2903/image.2.large.jpg
Pathogen detection
platforms and chemistries
Pathogen threat lists
http://www.cell.com/pb-assets/journals/research/cell-host-
microbe/online-now/S1931-3128(17)30554-1.pdf
SNPs underlying
primer sites
8. 8
Analysis of Bacterial Isolate Sequences
• Illumina MiSeq 2x75bp Nextera XT DNA sequencing of
eight skin microbiome “background” bacterial isolates
and two pathogens
• Raw reads came from isolates sequenced in house and
from previous studies, as indicated by sequence
accession numbers in CDC AR Isolate Bank
• Reads were processed with Trimmomatic to achieve
high quality scores and visualized with FastQC/MultiQC
• Trimmed reads were assembled with SPAdes, evaluated
with QUAST statistics and reference alignments, and
predicted gene content with prokka to verify strain
identification
• Custom curated databases, along with SRST2 and
ABRicate default databases, will be used to evaluate the
presence of antimicrobial and virulence genes
Plots Generated with FastQC and MultiQC
9. 9
“Ideal Mixtures” of DNA and RNA Sequences
• Represents the best case scenario of 100% transfer and recovery
• Microbial concentrations:
– Background organisms at equal amounts of 105 CFU/mL
– Pathogens at equal amounts of 106 CFU/mL
• These are the “high” pathogen concentrations for subsequent transfer events
• Lab methods:
– All eight background microorganisms and eight pathogens were
independently cultured and pooled, followed by sample processing
and library prep
– For RNA samples, ribosomal RNA depletion was performed with
MICROBExpress™ Bacterial mRNA Enrichment Kit (Thermo), then
reverse transcribed to cDNA and converted to double stranded DNA
using the SuperScript™ Double-Stranded cDNA Synthesis Kit
– Both mixtures were sequenced on the MiSeq with Nextera XT 2x75bp
CDC/Janice Carr
stock/123RF.com
https://microbewiki.kenyon.edu/inde
x.php/File:Brevibacteriumlinens.jpg
10. 10
Number of Reads from Ideal Mixtures Mapped to Isolate Assemblies*
0
1,000,000
2,000,000
3,000,000
4,000,000
5,000,000
6,000,000
7,000,000
8,000,000
9,000,000
Ideal DNA Mix Ideal cDNA Mix
NumberofReads
Data Generated with BWA and SAMtools
*Note: These mappings are
not to unique genomic regions
11. 11
Isolate Genomes Contained in Ideal Mixtures
MashScreenIdentity
Data Generated with Mash Screen
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
Ideal DNA Mix Ideal cDNA Mix
14. 14
Preliminary Results from Direct Transmission Scenarios
• The simulated hand washing did not have a strong effect on
pathogen presence or viability
• Possibly because the force of water was not included in our simulation
• Pathogen spike-in levels had an impact on CFU count and the
amount of detectable pathogen genomes available
• More bioinformatics analyses will be performed to further evaluate LoD
and gene-based detection and characterization
• Microbial concentrations in direct and indirect scenarios:
• High = Pathogens at equal amounts of 106 CFU/mL
• Low = Pathogens at equal amounts of 102 CFU/mL
• Background organisms at equal amounts of 106 CFU/mL
15. 15
Direct Contact Scenario with
High Spike-In and No Wash
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 2000 4000 6000 8000 10000 12000 14000
MashScreenIdentityMetric
CFU/mL
Direct Contact Scenario with
High Spike-In and Wash
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 1000 2000 3000 4000 5000 6000
MashScreenIdentityMetric
CFU/mL
16. 16
Direct Contact Scenario with
Low Spike-In and No Wash
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 1000 2000 3000 4000 5000 6000
MashScreenIdentityMetric
CFU/mL
Direct Contact Scenario with
Low Spike-In and Wash
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 1000 2000 3000 4000 5000 6000
MashScreenIdentityMetric
CFU/mL
19. 19
Preliminary Results from Indirect Transmission Scenarios
• Direct transfer is
more efficient than
indirect, especially at
high pathogen
concentrations
• Cotton transferred
the least and was
most impacted by
washing (i.e.,
simulated laundry)
• Nitrile and stainless
steel showed slightly
higher pathogen
transmission rates
than cotton
27. Ideal DNA and RNA Mixtures
27
Category Sample Source Raw Reads
Trimmed
Reads
Expected
GC%
Observed
GC%
8 Background +
8 Pathogens
DNA Mixture ATCC and CDC 7,229,710 6,882,852 48% 50%
8 Background +
8 Pathogens
RNA Mixture ATCC and CDC 5,645,200 5,313,848 48% 52%
Editor's Notes
Advantages of amplicon sequencing:
After initial development costs have been invested, it is more cost effective to run per reaction
Lower limits of detection than shotgun sequencing
Advantages of shotgun metagenomics and metatranscriptomics:
- Reduces dependence on inflexible, pathogen-specific methods and reference databases
- No prior knowledge needed about microbial targets
- Potential to identify co-infections and unexpected disease causes or virulence/AMR genes
- Could shorten time to answer if the pathogens take a long time to culture
New technologies are likely to improve agnostic pathogen detection methods, eventually eliminating the need for amplicon-based detection.
False negatives currently are a challenge for shotgun metagenomics because pathogen or gene-specific sequences may not always be present at high enough quantities in environmental samples to confidently determine that a pathogen or antimicrobial resistance gene is present. This study aims to better understand the limits of detection in metagenomics by evaluating the relationship between colony forming units and detectable pathogen and gene-specific sequences under a variety of environmental transmission scenarios.
More agnostic pathogen detection methods will make changing with the times easier!
Mash Screen “Identity” = Fraction of bases shared between the assembled isolate and the mixture, which is estimated from the fraction of their shared k-mers
The simulated hand washing did not have a strong effect on pathogen presence or viability, possibly because the force of water was not included in our simulation.
Pathogen spike-in levels had an impact on CFU count and the amount of detectable pathogen genomes available, with fewer viable pathogens in the low spike-in compared to the high.
Some pathogens seem to be more easily transmitted than others, and this will be described in more detail in the future.
This is an example of the CDC taxonomic assignment disagreeing with NCBI’s taxonomic assignment at the species level.
This is an example of the CDC taxonomic assignment disagreeing with NCBI’s taxonomic assignment at the species level.