Output from prediction with locpix_points and dSTORM data on epiregulin of response to panitumumab treatment for metastatic colorectal cancer

Output from using ClusterNet (https://github.com/oubino/locpix_points) combined with logistic regressions on data in https://doi.org/10.5281/zenodo.21339937. ClusterNet is a graph neural network acting on clusters and their spatial arrangement in a point cloud [1]. We used k = 96 in k-means clustering in this case. Contains configuration files, scripts, interactive notebooks, intermediate data processing and results and final prediction of patient response to treatment. See oubino/locpix_points at working and locpix_points/paper at working · oubino/locpix_points for details. [1] Umney O, Slaney H, Williams CJM, Quirke P, Peckham M, Curd AP. ClusterNet: Classifying Single-Molecule Localization Microscopy Datasets with Graph-Based Deep Learning of Supracluster Structure. Small Science. 2025;5(12):e202500255.