-
Notifications
You must be signed in to change notification settings - Fork 0
5. Model setup
These model setup instructions apply to both models.
To successfully get all the large files from GitHub, you may first need to have the LFS installed in your local machine or on the "idmodeling2" server (the other two don't have git-lfs available). For more information, you could visit this website. After the LFS is ready, you can git clone the master branch.
Your project directory should have the following folders for the simulations to run:
configs/diagnostics/input_data/input_data/confirmation_rate/input_data/incidence/input_data/shapefiles/input_data/worldpop/montagu/output_final/output_raw/packages/packages/ocvImpactsbatch_scripts/scripts/summarize_outputs/
Among them, input_data/shapefiles, output_final/, and output_raw/ could be empty but they must exist. We will create new functions in ocvImpact package that can generate such directories automatically in the near future.
Our directories should also have certain files ready before the simulation.
Please note that the .tif files and .zip file are very large, you need to use run git lfs install before cloning or pulling. Please refer to this website for more information.
input_data/log1d_obs.rdsinput_data/log2d_obs.rdsinput_data/confirmation_rate/parameters.csvinput_data/incidence/VIMC-47-countries-for-cholera-modelling.csvinput_data/incidence/afro_2010-2016_lambda_5k.zipinput_data/incidence/afro_2010-2016_lambda_5k_mean.tifinput_data/incidence/who_case_repo_source.csvinput_data/worldpop/ppp_2020_1km_Aggregated.tifinput_data/worldpop/ppp_2020_5km_Aggregated.tifinput_data/IHME_GBD_2019_DISABILITY_WEIGHTS_Y2020M010D15.XLSXinput_data/locations_todeletelater.csvinput_data/ocv_ve_overtime.csvinput_data/region_country.csvinput_data/who_cfrs.csv-
input_data/incidence/afro_2010-2016_lambda_5k.tiforinput_data/incidence/afro_2010-2016_lambda_5k.zip input_data/incidence/afro_2010-2016_lambda_5k_mean.tif-
input_data/incidence/afro_2016-2020_lambda_5k.tiforinput_data/incidence/afro_2016-2020_lambda_5k.zipfor the 202310gavi-4 touchstone -
input_data/incidence/afro_2016-2020_lambda_5k_mean.tiffor the 202310gavi-4 touchstone input_data/incidence/VIMC-47-countries-for-cholera-modelling.csvinput_data/incidence/who_case_repo_source.csvmontagu/201910gavi-5/stochastic_template_params.csvmontagu/202110gavi-3/stochastic_template_params.csv-
input_data/drc_custom_coverage_2024_2026.csv, for the DRC case study -
input_data/drc_custom_targeting_2024_2026.rds, for the DRC case study -
input_data/shapefiles/DRC_custom_shapefile/custom_shapefile.rds, a health zone shapefile for DRC -
input_data/shapefiles/DRC_custom_shapefile/country_shapefile.rds, a country-level shapefile for DRC for the DRC Case Study
Each set of VIMC core project model runs from the VIMC will have a unique run name (touchstone) (e.g., 201910gavi-5, 202110gavi-3, and 202310gavi-4) and a set of standard demographic files associated with that run. These demographic files will be automatically downloaded from the Modellers contribution portal on the VIMC Montagu website by our ocvImpact package. These demographic files associated with the 202110gavi-3 run name are used for the surveillance project.
Specifically for the VIMC core project, the files for both genders (combined) and in long format will be downloaded and saved in montagu/<enter-run-name-here>/. The files required for this model will be:
- Population: Interpolated, 1-year
- Population: Total
- Mortality: Life Expectancy at Birth
Each VIMC core project model run will also have a set of coverage templates (one per scenario), burden templates (central and stochastic), and a stochastic parameters template. These will also be automatically acquired from the same Modellers contribution portal described above and placed in montagu/<enter-run-name-here>/ except for the stochastic parameters template, which will be pre-generated and placed under the montagu/201910gavi-5/ directory or the montagu/202110gavi-3/ directory.
Files from external sources are also required.
Country-level and admin2 level shapefiles for each modeled country will be downloaded from GADM.org to input_data/shapefiles as sf objects in RDS file format when load_shapefile_by_country is called for the first time. Subsequent calls to use the same shapefile will then load an already existing file. If you want to refresh the shapefiles you are using, you should manually delete the relevant files from input_data/shapefiles and use the load_shapefile_by_country function again.
Raster-level population data (1km) for the whole world should be acquired from WorldPop. We recommend the ``Unconstrained global mosaic'' at 1 km resolution from 2020. This file should be saved and manually placed in input_data/worldpop.
After we check that all the directories and files are in place, we could use cd <the gavi_vimc_cholera directory> to change the current working directory to within the gavi_vimc_cholera folder.
Then, we unzip a file needed for the model using cd input_data/incidence/ and unzip -o afro_2016-2020_lambda_5k.zip, and don't forget to get back to the main directory afterward. The reason why we need to do it is that GitHub has a limit for the file size, we had to zip it before uploading it.
Next, we modify scripts/montagu_handle.R to input our own credentials for the VIMC Montagu API.