Code and Data for 'Spatio-temporal spillover risk of yellow fever in Brazil'

This repository contains the code and data necessary to reproduce the analysis from:

Kaul, RajReni B., Michelle V. Evans, Courtney C. Murdock, John M. Drake. 2018. Spatio-temporal spillover risk of yellow fever in Brazil. Parasite & Vectors 11, 488. https://doi.org/10.1186/s13071-018-3063-6

Dependencies

The majority of this code is to be run in R, and was initially run on the following version:

R version 3.4.3 (2017-11-30)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 14.04.5 LTS

Some scripts to download data will need to run via the command line, and were initially run on Linux machines.

Overall File Structure

• data_raw: raw environmental, demographic, and case data. This includes remote sensed environmental data that has been extracted for each municipality
• data_clean: cleaned environmental data, processed via R scripts in data_processingScripts
• data_processingScripts: cleans data, moving it from raw to clean folder
• code: code to run analysis and create manuscript figures
• data_out: results of analysis, predictions by municipality x month and variable importance data

Data

Raw data is located in the data_raw folder. Meta-data on how they were downloaded and processed can be found in the files:

• YF_metadata.txt
• demo_metadata.txt
• EnvCovar-metadata.md

All scripts used to download raw data are in the data_processingScripts folder, and noted in the metatdata files. Most should be run in the following order: download, extract, process, combine.

Once each individual variables has been processed, it is saved in the data_clean folder. The final full datasets are created via the GiantDataFrameMaker.R file. This creates TestingDataSpat2.rds and TrainingDataSpat2.rds. The data is further split into the National and Regional Models in the nhpSplit.R file in the data_processingScripts folder.

Final datasets

There are six final datasets that are used in our analysis, three training and three testing:

Training:

• data_clean/Mosi_Final/TrainingDataLowNHP.rds : training data for LRR model
• data_clean/Mosi_Final/TrainingDataHighNHP.rds : training data for HRR model
• data_clean/Mosi_Final/TrainingDataLowNHP.rds : training data for National model

Testing:

• same as training but named Testing*