--- orphan: true --- # Use cases The use cases are longer analyses built on the steps of the {doc}`tutorials <../tutorials/index>`. Each starts from a question, builds a cohort for it and runs the analysis to a result. - {doc}`training_cohort` prepares a cohort for machine learning: one table across datasets, a split that keeps each experiment in one fold, and a manifest that records the split. It runs on a laptop in a few minutes. - {doc}`visium_atlas/visium-atlas` integrates every Visium section in the catalog, about 4,000 of them, into one atlas and asks whether the spots group by species or by organ. It needs a large GPU node and has its own environment. - {doc}`loader_atera` writes the six whole Atera H&E slides of a catalog release as a loader collection on a Slurm cluster and measures how fast the spatialbatch loader reads 224 px training crops from it, on CPU and into a GPU. - {doc}`loader_training` trains a model on H&E crops from those slides in PyTorch, with a spatial split and one slide held out. ```{toctree} :hidden: training_cohort visium_atlas/visium-atlas loader_atera loader_training ```