Prioritizing newborn screening sites in Uganda using HbS and PfPR with greedy selection

Release: Uganda SCD newborn screening co-risk siting pipeline (v0.1.0)

Highlights

• End-to-end workflow to build an HbS × PfPR co-risk raster, rank newborn SCD screening sites with a deterministic greedy rule, and produce coverage, equity, and collocation diagnostics for Uganda.
• Clean structure with configs, CLI scripts, and Make targets for repeatable runs.
• Ready for phase planning, equity monitoring, and facility collocation analysis.

What's included

• src/co_risk/make_co_risk.py — align input rasters and write corisk.tif
• src/co_risk/greedy_sites.py — greedy site selection at a fixed service radius
• src/co_risk/coverage_metrics.py — coverage curve and 60/70/80% milestones
• src/co_risk/equity_metrics.py — district Lorenz inputs and Gini coefficient
• src/co_risk/collocate_sites.py — collocate ranked sites to facilities using 60-minute catchments or nearest facility
• configs/config.yaml — paths, CRS, grid, selection radius, catchment settings
• README.md, Makefile, environment.yml, requirements.txt
• .github/workflows/ci.yml — minimal CI (lint/compile)

Data requirements (place in data/raw/)

• hbs_surface.tif (EPSG:32636), HbS allele frequency, 0–1
• pfpr_mean_2015_2024.tif (EPSG:32636), decade mean PfPR2–10, 0–1
• districts.geojson with stable district IDs and names
• facilities.parquet with id,name,lat,lon and optional readiness fields
• Optional isochrones.gpkg with 60-minute travel-time polygons

Install and run

# conda
conda env create -f environment.yml
conda activate uganda-scd-screening

# or venv + pip
python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt

# run the full pipeline
make all

If you use this repository, please cite it as below.