Interplay between climate, childhood mixing, and population-level susceptibility explains a sudden shift in RSV seasonality in Japan

---

This repository contains all code and data used for the analysis.

• R: contains an R script for simulating the SIRS model
• analysis: contains an R script for comparing changes in transmission rate before and after 2017
• data: contains time series data for RSV cases and number of children attending childcare facilities as well as data on population sizes by prefecture
• doc: contains latex files for the manuscript
• figure: contains R scripts for plotting all figures
  • figure1.R: R script for generating figure 1 and figure S1
  • figure_comb_sirs_npi.R: R script for generating figure 2
  • figure_comb_sirs_change.R: R script for generating figure 3
  • figure_childcare.R: R script for generating figure 4
  • figure_map.R: R script for generating figure S2
  • figure_joint_climate.R: R script for generating figure S3
  • figure_ryukyu_sirs_change.R: R script for generating figure S4
• script: contains an R script for processing data
• simulate_sirs: contains R scripts for simulating the fitted SIRS model across different islands to explore the relationship between seasonal forcing and center of gravity
  • simulate_sirs_honshu.R: R script for simulating epidemic dynamics in Honshu island
  • simulate_sirs_kyushu.R: R script for simulating epidemic dynamics in Kyushu island
  • simulate_sirs_ryukyu.R: R script for simulating epidemic dynamics in Ryukyu island
  • simulate_sirs_shikoku.R: R script for simulating epidemic dynamics in Shikoku island
  • simulate_sirs_interpolate.R: R script for simulating epidemic dynamics using interpolated transmission rates
• stanfit_sirs: contains R scripts for fitting SIRS model using Stan
• stanfit_sirs2: contains R scripts for fitting SIRS model that allows for changes in transmission using Stan
• stanmodel: constrains stan scripts for deterministic models

---

• R scripts in stanfit_sirs and stanfit_sirs2 folders can be run independently as standalone files; these files need to be run first to generate rda files for model fits.
• R scripts in simulate_sirs folder can be run after all stan models have been fitted. These scripts will generate rda files that contain a summary of analyses of fitted models.
• R scripts in figure folder need to be run after all models have been fitted and analyzed.

---

All code was run on M2 MacBook Pro, 2023. Each model fit takes <10 minutes to run. All other code will take much less time than model fitting.