Search for: All records

Abstract Understanding the drivers of seasonal disease outbreaks remains a fundamental challenge in disease ecology. Periodic outbreaks can be driven by several seasonally varying factors, including pulses of susceptible individuals through births, changes in host behaviour and social aggregation and variation in host immunity. However, when these potential drivers overlap temporally, isolating their relative contributions to outbreak patterns becomes challenging.We studied Hendra virus, a zoonotic pathogen with seasonal spillovers from bats to horses and humans. Multiple seasonal factors have been hypothesized to drive Hendra virus transmission, including food shortages, birth pulses and changes in host aggregation, but their temporal overlap has made identifying primary drivers difficult.We conducted a 4‐year longitudinal study ofPteropusbats to test whether seasonal birth pulses and the resulting influx of susceptible juveniles drive Hendra virus transmission. Using a Bayesian ageing model, we aged sexually immature bats and placed them into birth cohorts. We used our age predictions to model how viral shedding and antibody responses changed as bats aged. We trackedBartonellaspp. Infection—a bacterial pathogen requiring close contact for transmission—as an indicator of transmission opportunities within each cohort for comparison.We found no evidence that seasonal birth pulses of immunologically naïve juveniles drove Hendra virus transmission. Two out of three cohorts showed substantially reduced maternal antibody transfer compared to the 2018 cohort, with seroprevalence near zero at our earliest sampling timepoints and showed no clear evidence of synchronized seroconversion. Furthermore,Bartonellainfection rates were consistent across cohorts, indicating that opportunities for pathogen transmission remained consistent across cohorts despite varying viral shedding patterns.Our findings demonstrate that birth pulses alone cannot explain observed patterns of Hendra virus outbreaks. These results highlight the importance of using multiple lines of evidence to evaluate competing mechanisms underlying seasonal disease dynamics, particularly when potential drivers coincide temporally. 

Abstract Substantial global attention is focused on how to reduce the risk of future pandemics. Reducing this risk requires investment in prevention, preparedness, and response. Although preparedness and response have received significant focus, prevention, especially the prevention of zoonotic spillover, remains largely absent from global conversations. This oversight is due in part to the lack of a clear definition of prevention and lack of guidance on how to achieve it. To address this gap, we elucidate the mechanisms linking environmental change and zoonotic spillover using spillover of viruses from bats as a case study. We identify ecological interventions that can disrupt these spillover mechanisms and propose policy frameworks for their implementation. Recognizing that pandemics originate in ecological systems, we advocate for integrating ecological approaches alongside biomedical approaches in a comprehensive and balanced pandemic prevention strategy. 

Abstract The COVID-19 pandemic and its aftermath are the most significant socio-economic crises in modern history. The pandemic’s devastating impacts have prompted urgent policy and regulatory action to reduce the risks of future spillover events and pandemics. Stronger regulatory measures for the trade of wildlife are central to discussions of a policy response. A variety of measures, including broad bans on the trade and sale of wildlife to banning specific species for human consumption are among a suite of discussed options. However, the wildlife trade is diverse, complex, and important for the livelihoods of millions of people globally. We argue that reducing the risk of future pandemics stemming from the wildlife trade must follow established principles of governance which include being equitable, responsive, robust, and effective. We demonstrate how incorporating these principles will support the development of context-specific, culturally sensitive, and inclusive responses that recognize the on-the-ground complexity of disease emergence and the social-ecological systems in which the wildlife trade occurs. 

Changes in the quality and quantity of food consumed can affect the health of hosts, their ability to control infections and potentially shape the likelihood of pathogen spillover. Dietary shifts have been proposed as one of the factors driving spillovers of zoonotic viruses from bats to humans. In this study, we examined how diet composition alters the immune response to viral shedding and the risk of spillover by developing a mechanistic model fitted to experimental data of Jamaican fruit bats infected with influenza A virus H18N11 and fed different diets. The model selected from alternative immune and metabolic relationships showed that the coupled effects of citrulline and tumour necrosis factor alpha (TNF) affected the control of viral shedding with parameters that varied with diet. Bats on the suboptimal fat diet appeared to control shedding more successfully than bats on suboptimal sugar or optimal protein diets. Yet, bats on the optimal diet could potentially cause lower hazard of spillover because of reduced food consumption, suggesting fewer and/or shorter visits at the feeding sites and thus transmission to secondary hosts. This study provides a parsimonious explanation of the barriers that affect viral shedding by reservoir hosts and the consequences for the hazard of spillover. 

{"Abstract":[" LUMINEX Wildlife Disease Analysis Pipeline\n\n  Overview\n\n  Bayesian analysis pipeline for "Cohorts of immature Pteropus bats show interannual variation in Hendra virus serology"\n\nSummary \n\n  Prerequisites\n\n  Software Requirements\n\n  - R (≥ 4.0.0)  - Stan (≥ 2.21)\n\n  Required R Packages\n\n  # Core packages  install.packages(c("rstan", "tidyverse", "here", "loo", "bayesplot"))  # Additional packages    install.packages(c("RColorBrewer", "cowplot", "boot", "see", "factoextra", "bestNormalize", "LaplacesDemon", "ggpubr", "plyr", "see", "pscl"))    # Core utility functions (CRITICAL DEPENDENCY)  source(here("R", "useful_functions.R"))  # Bayesian analysis functions  source(here("R", "paper_theme.R"))      # Plotting themes    Hardware Requirements\n\n  - Storage: ≥5GB free space  - CPU: Multi-core processor recommended\n\n  Data Requirements\n\n  /raw_sharable folder\n\n  Execution Workflow\n\n  ⚠️ CRITICAL: Execute in This Order\n\n  Phase 1: Core Data Processing\n\n  1. Initial Data Cleaning  source("R/create_datasets_for_stan_part_1.R")    - Runtime: ~5-10 minutes    - Outputs: data_for_cohort_model.csv, luminex_igg_igm.csv  2. Serology Classification  rmarkdown::render("R/mixture_model_final.R")    - Runtime: ~30-60 minutes (Stan model fitting)    - Outputs: serology_prob.csv  3. Age/Cohort Assignment  rmarkdown::render("R/cohort_model_2025_05_12.Rmd")    - Runtime: ~1-3 hours (complex Stan models)    - Outputs: age_predictions.csv & a species comparison analysis   4. Dataset Integration  source("R/create_datasets_for_stan_part_2_06_30_24.R")    - Runtime: ~10-15 minutes    - Outputs: All analysis-ready datasets with time-alive variables\n\n  Phase 2: Primary Analyses\n\n  5. Prevalence Smoothing Analysis  rmarkdown::render("R/gaussian_smooth_prevalence_06_30_24.Rmd")    - Runtime: ~10+ hours (multiple Stan models)    - Key outputs: Prevalence curves, model comparisons    - Additional outputs:       -Basic prevalence smoothing (4 pathogens × 4 cohorts)      -Site-specific analysis (8-cohort models)      -Sex-stratified analysis      -Stringent cohort cutoff analysis      -Multiple cutoff threshold testing (5 different cutoffs)      -Date-based modeling      -Batch effect testing      -Adult vs juvenile comparisons  6. Logistic Regression Analysis  source("R/logistic_models_fig_3.R")  source("R/logistic_models_fig_2.R")    - Runtime: ~30-60 minutes    - Outputs: Figure 2 & 3 plots\n\n  Phase 3: Supporting Analyses (Optional)\n\n  7. Additional Analyses (run as needed):    - adult_prevalence_curves.Rmd - Adult dynamics    - PCA_new_analysis.R - Multivariate analysis    - additional_figures.R - Supplementary figures\n\n  Key Functions (useful_functions.R)\n\n  - fit_4cohort_model() - Bayesian 4-cohort prevalence model  - compile_stan_results() - Extract and format Stan results  - create_time_sequence() - Generate prediction timepoints  - compute_loo_cv() - Leave-one-out cross-validation  - plot_parameter_diagnostics() - Model diagnostic plots\n\n  Troubleshooting\n\n  Common Issues\n\n  Stan compilation errors:  # Recompile Stan models  rstan_options(auto_write = TRUE)  options(mc.cores = parallel::detectCores())\n\n  Memory issues:  - Reduce Stan iterations: ITER = 1000 instead of ITER = 2000  - Run analyses sequentially, not in parallel\n\n  Missing dependencies:  # Load all utility functions  source(here("R", "useful_functions.R"))  source(here("R", "paper_theme.R"))\n\n  Resume from Saved Results\n\n  Many scripts save intermediate results:  # Check for existing model fits  if(file.exists("model_results.RData")) {    load("model_results.RData")  } else {    # Run full analysis  }\n\n  Output Structure\n\n  Luminex_figs/r_figs/     # All generated figures  Data_for_publication/    # Final analysis datasets  stan/                    # Stan model files  R/                       # Analysis scripts\n\n  Expected Runtime\n\n  Full pipeline: 10-20 hours on modern hardware  \n\n \n\n "]} 

Data and code for 'Diet composition affects bat viral shedding with potential consequences for pathogen spillover' 

Synchronized seasonal excretion of multiple coronaviruses coincides with high rates of coinfection in immature bats. This repo contains instructions and source code for reproducing the statistical analyses in the manuscript. Repo Contents scripts: contains the source .R and .stan files to reproduce the anaysis. Each file is detailed below in the specific sections corresponding to the statistical analyses. data: contains the raw source data and model generated output. figures: contains the final output figures from the manuscript. These can be recreated with the CovOZ_Figures_Submission_Clean.R script. 1. System Requirements Hardware Requirements Our source code requires only a standard computer. Much of the Markov chain Monte Carlo code is run in parallel so a computer with ample memory and multiple cores can be advantageous. The runtimes below are generated using a macbook with the recommended specs (64 GB RAM, 8 cores at 2.7 GHz). The code will also work on linux or windows computer. Software Requirements Reproducing the statistical analyses requires a current version of R and stan. We use version 4.4.1 of R and version 2.32.2 of stan. Package dependencies and versions Users will need the following packages install the following packages to execute the code. Our versions are effective October 1, 2024 tidyverse 2.0.0 lubridate 1.9.3 stringr 1.5.1 rstan 2.32.6 cowplot 1.1.3 ggtext 0.1.2 jpeg 0.1-10 scales 1.3.0 tictoc 1.2.1 2. Installation Guide Running the analysis requires: installing R. Depending on wifi speeds, installing R usually takes a few minutes. installing stan. Depending on wifi speeds, installing stan usually takes a few minutes. installing the necessary R packages (listed above). Depending on wifi speeds, installing packages usually takes about 30 seconds per package. 3. Demo This source code is not an R package with a formal demo, but rather source code is included for the various analyses in section 4. 4. Instructions for Use 4.1 Coinfection Analysis Runs chi-squared tests on coinfections of beta 2d.iv and beta 2d.v. Generates summary statistics, test statistics, and p-values from manuscript. input files: individual_variant_covariates.csv script file: coinfection_final.R run time: approximately 1 second 4.2 Individual Level Dynamics of Infection: Dynamic Binary Regression Runs individual level dynamic binary regression models. Produces output file that can recreate figures. input files: individual_variant_covariates.csv script files: logistic_curves_final.R GP_regression.stan output files: logistic_curve_out.RData run time: approximately 66 minutes 4.3 Dynamics of Circulation at the Population Level Runs combined (individual and pooled data) dynamic models. Produces output file that can recreate figures. input files: combined_out_variant.csv script files: cluster_curves_final.R GP_withLL.stan output files: cluster_curves.csv run time: approximately 25 minutes 4.4 Manuscript Figures Combined script that uses output files created by previous scripts to recreate all figures in the manuscript. input files: model_output/cluster_curves.csv combined_out_variant.csv individual_variant_covariates.csv model_output/logistic_curve_out.RData script files: CovOZ_Figures_Submission_Clean.R output files: Figure2_final.png Figure3_final.png Figure4_A-D_final.png Figure6_AP.png Figure7.png SIFigure8.png SIFigure9.png run time: approximately 16 seconds 4.5 Model Comparison Integrated Compares LOOIC values for sets of model frameworks. input files: combined_out_variant.csv script files: Pred_Comparisons.R GP_withLL.stan output files: preds.RData run time: approximately 2 hours 4.6 Model Comparison Individual Compares LOOIC values for sets of model frameworks. input files: combined_out_variant.csv script files: logistic_curves_loo.R GP_regression.stan GP_regression_add.stan GP_regression_interact.stan output files: logistic_curve_loo_age.RData logistic_curve_loo_age_add_sex.RData logistic_curve_loo_age_interact_sex.RData run time: approximately 6:45 hours 

Objectives We aim to estimate geographic variability in total numbers of infections and infection fatality ratios (IFR; the number of deaths caused by an infection per 1,000 infected people) when the availability and quality of data on disease burden are limited during an epidemic. Methods We develop a noncentral hypergeometric framework that accounts for differential probabilities of positive tests and reflects the fact that symptomatic people are more likely to seek testing. We demonstrate the robustness, accuracy, and precision of this framework, and apply it to the United States (U.S.) COVID-19 pandemic to estimate county-level SARS-CoV-2 IFRs. Results The estimators for the numbers of infections and IFRs showed high accuracy and precision; for instance, when applied to simulated validation data sets, across counties, Pearson correlation coefficients between estimator means and true values were 0.996 and 0.928, respectively, and they showed strong robustness to model misspecification. Applying the county-level estimators to the real, unsimulated COVID-19 data spanning April 1, 2020 to September 30, 2020 from across the U.S., we found that IFRs varied from 0 to 44.69, with a standard deviation of 3.55 and a median of 2.14. Conclusions The proposed estimation framework can be used to identify geographic variation in IFRs across settings.