Topological Skeletonization Scripts for Assessing Estimated Malaria Parasite Migration Contours Under Sparse Sampling

Abstract

Malaria parasite gene flow simulation models that utilize georeferenced genomic data have become key tools for inferring parasite migration patterns and diversity through simulation-based approaches. However, the accuracy of the models’ outputs can be compromised when sampling locations are sparse, leading to uncertainty in the results. We have developed a sample filtering framework that applies topological skeletons for a contour region and kernel density estimation based on malaria parasite sampling locations to assess malaria parasite sample density for each contour and filter contour regions with lower sample density. The Topological Skeletonization Scripts package provides two reproducible Jupyter notebooks that convert Effective-Migration-Surface (EEMS) contour rasters into vector-ready topological skeletons. The topo_ske3spots.ipynb script operates on a simulated hexagon-shaped barrier example scenario produced using msprime, a population genetic simulator of ancestry and DNA sequence evolution, while the topo_ske_cam.ipynb notebook processes empirical Plasmodium falciparum malaria parasites collected in Cambodia from 2008 to 2013 (data archived as part of the MalariaGen Community Project). Each notebook implements automated thresholding, morphological cleaning, medial-axis skeletonization (scikit-image), and exports the results as aligned GeoTIFFs, GeoJSON polylines, and diagnostic plots. Developed in Python 3.9 and tested on macOS 10.15 under a Miniconda-managed environment, the workflows utilize open-source libraries (GDAL/rasterio, GeoPandas/Shapely, NumPy, OpenCV).