Quantifying plasmid movement in drug-resistant Shigella species using phylodynamic inference

Authors

• Nicola F. Müller^1,2,†
• Ryan R. Wick³
• Louise M. Judd⁴
• Deborah A. Williamson⁵
• Trevor Bedford^2,6
• Benjamin P. Howden^3,4
• Sebastián Duchêne^3,7,‡
• Danielle J. Ingle^3,‡,†

¹Division of HIV, ID and Global Medicine, University of California San Francisco, CA, USA
²Vaccine and Infectious Disease Division, Fred Hutchinson Cancer Center, Seattle, WA, USA
³Department of Microbiology and Immunology at the Peter Doherty Institute for Infection and Immunity, The University of Melbourne, Melbourne, VIC, AUS
⁴Centre for Pathogen Genomics, University of Melbourne (Doherty Institute)
⁵Department of Infectious Diseases at the Peter Doherty Institute for Infection and Immunity, The University of Melbourne, Melbourne, VIC, AUS
⁶Howard Hughes Medical Institute, Seattle, WA, USA
⁷EDID unit, Department of Computational Biology, Institut Pasteur, Paris, France

^†Corresponding authors: nicola.felix.mueller@gmail.com, danielle.ingle@unimelb.edu.au
^‡These authors contributed equally to this work

---

Abstract

The 'silent pandemic' of antimicrobial resistance (AMR) represents a significant global public health threat. These AMR genes are often carried on mobile elements, such as plasmids. The horizontal movement of these plasmids allows AMR genes, and the resistance they confer to key therapeutics, to disseminate throughout a population. However, quantifying the movement of plasmids remains challenging with existing computational approaches.

Here, we introduce a novel method for reconstructing and quantifying the movement of plasmids in bacterial populations over time. To do this, we model the co-evolution of chromosomal and plasmid DNA using a joint coalescent and plasmid transfer process in a Bayesian phylogenetic network framework. Our approach captures differences in the evolutionary histories of plasmids and chromosomes to identify instances where plasmids likely moved between bacterial lineages, while rigorously accounting for uncertainty in the data.

We apply this new method to a five-year dataset of Shigella, exploring transfer rates for five different plasmids that harbor various AMR and virulence profiles. In doing so, we reconstruct the co-evolution of the large Shigella virulence plasmid with the chromosomal DNA, quantify the higher transfer rates of three small plasmids that circulate among lineages of Shigella sonnei, and describe the recent dissemination of a multidrug-resistant plasmid between S. sonnei and S. flexneri lineages. Notably, this plasmid appears to have transferred in multiple independent events alongside an overall increase in its prevalence since 2010.

Our approach provides a powerful framework for understanding the evolutionary dynamics of plasmids carrying AMR genes as they are introduced, circulate, and are maintained in bacterial populations.

---

Repository Structure

.
├── Software/               # Analysis software tools (JAR files)
├── Applications/          # Real-world data analyses
│   └── Shigella/         # Shigella analysis pipeline and results
├── Simulations/          # Simulation studies
│   ├── PlasmidEvol/      # Plasmid evolution simulations
│   └── Rates/            # Rate parameter simulations
└── Validation/           # Method validation tests

---

Software Tools

The Software/ directory contains custom Java tools for plasmid phylodynamic analysis:

• PlasmidTreeMapper.jar - Maps plasmid trees onto chromosomal phylogenies
• TransferCounter.jar - Counts and analyzes plasmid transfer events
• LossCounter.jar - Identifies and quantifies plasmid loss events
• Summarizer.jar - Summarizes posterior distributions from MCMC analyses
• CoalPT.jar - Coalescent and plasmid transfer inference
• LTT.jar - Lineage-through-time analysis

Requirements

• Java 8 or higher
• BEAST 2.7+ (for XML-based analyses)
• R 4.0+ (for visualization and data analysis)
• MATLAB (for XML generation scripts)

---

Applications

Shigella Analysis

The Applications/Shigella/ directory contains the complete analysis pipeline for Shigella data:

Key Scripts

MATLAB Scripts:

• createXml.m - Generate BEAST XML files for individual analyses
• createSonneiFlexXml.m - Generate XMLs for S. sonnei and S. flexneri analyses
• createChromosomePlasmidTrees.m - Create tanglegram analyses
• createSupplementalReassortmentXml.m - Generate supplemental reassortment analyses
• splitFlex.m - Split S. flexneri data for analysis
• compute.m - Main computational pipeline

R Scripts:

• plotShigellaData.R - Main visualization of Shigella results
• plotLBIandClusters.R - Plot local branching index and cluster analysis
• plotDensityTree.R - Create density tree visualizations
• plotSupplementalAnalyses.R - Generate supplemental figures
• densitreecopy.R - Additional density tree utilities

Python Notebooks:

• plasmid_vis.ipynb - Interactive plasmid visualization
• tanglegram.ipynb - Chromosome-plasmid tanglegram generation
• ChromosomePlasmidTanglegrams.ipynb - Advanced tanglegram analyses

Data Structure

• data/ - Contains sequence alignments and phylogenetic trees
  • Sonnei/ - S. sonnei genomic data
  • Flex/ - S. flexneri genomic data
  • pksr100/ - Plasmid sequence data
• combined/ - Results from combined chromosome-plasmid analyses
• xmls/ - BEAST XML configuration files
• supplementalxmls/ - Supplemental analysis configurations

---

Simulations

Plasmid Evolution Simulations (Simulations/PlasmidEvol/)

Simulate plasmid transfer and evolution under various evolutionary scenarios.

Key files:

• makeXmls.m - Generate simulation XML files
• runSimulations.sh - Execute simulation pipeline
• runInferences.sh - Run inference on simulated data
• countSNPs.m - Count SNPs in simulation outputs
• plotRates.R - Visualize inferred rates

Rate Validation (Simulations/Rates/)

Validate rate parameter estimation under known conditions.

Usage:

cd Simulations/Rates/
bash runSimulations.sh
bash runInferences.sh
Rscript plotRates.R

---

Validation

The Validation/ directory contains tests to validate the method's accuracy.

Test cases:

• Serial sampling with 5 taxa and 5 plasmids
• Serial sampling with 5 taxa and 10 plasmids

Run validation:

cd Validation/
beast testAll_serial5taxon5plasmids.xml
beast testAll_serial5taxon10plasmids.xml
Rscript plotValidation.R