"Cohorts of immature Pteropus bats show interannual variation in Hendra virus serology

LUMINEX Wildlife Disease Analysis Pipeline Overview Bayesian analysis pipeline for "Cohorts of immature Pteropus bats show interannual variation in Hendra virus serology" Summary Prerequisites Software Requirements - R (≥ 4.0.0) - Stan (≥ 2.21) Required R Packages # 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 - Storage: ≥5GB free space - CPU: Multi-core processor recommended Data Requirements /raw_sharable folder Execution Workflow ⚠️ CRITICAL: Execute in This Order Phase 1: Core Data Processing 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 Phase 2: Primary Analyses 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 Phase 3: Supporting Analyses (Optional) 7. Additional Analyses (run as needed): - adult_prevalence_curves.Rmd - Adult dynamics - PCA_new_analysis.R - Multivariate analysis - additional_figures.R - Supplementary figures Key Functions (useful_functions.R) - 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 Troubleshooting Common Issues Stan compilation errors: # Recompile Stan models rstan_options(auto_write = TRUE) options(mc.cores = parallel::detectCores()) Memory issues: - Reduce Stan iterations: ITER = 1000 instead of ITER = 2000 - Run analyses sequentially, not in parallel Missing dependencies: # Load all utility functions source(here("R", "useful_functions.R")) source(here("R", "paper_theme.R")) Resume from Saved Results Many scripts save intermediate results: # Check for existing model fits if(file.exists("model_results.RData")) { load("model_results.RData") } else { # Run full analysis } Output Structure Luminex_figs/r_figs/ # All generated figures Data_for_publication/ # Final analysis datasets stan/ # Stan model files R/ # Analysis scripts Expected Runtime Full pipeline: 10-20 hours on modern hardware