**Multi-scale dynamic imaging reveals that cooperative motility behaviors promote efficient predation in bacteria** *Sara Rombouts, Anna Mas, Antoine Le Gall, Jean-Bernard Fiche, Tâm Mignot, Marcelo Nollmann* https://www.biorxiv.org/content/10.1101/2022.12.20.521001v1 ## UMAP embedding Uniform Manifold Approximation and Projection (UMAP) was employed for dimensionality reduction and visualization. A normalization step was performed on the input local density and cluster size features using the StandardScaler from scikit-learn. The UMAP algorithm was executed using a Number of Neighbors of 100 and a Minimum Distance of 0. UMAP's default values were used for other parameters, such as the distance metric (Euclidean) and the number of dimensions in the projected space (2D visualization). The UMAP projection results were visualized using Datashader. The resulting embedding is found in the `UMAP embedding` folder with filename: `00_trained_on_exp8.pkl`. This file can be loaded using the Pickle package in python.