Implements exhaustive grid search and cross-validation using Snakemake's paramspace interface to parallelize embedding and clustering steps for different parameter combinations. We aggregate results into a single table and then process that table with a Jupyter notebook to identify the optimal distance threshold per method and optimal method-specific parameters for t-SNE and UMAP.
Implements exhaustive grid search and cross-validation using Snakemake's paramspace interface to parallelize embedding and clustering steps for different parameter combinations. We aggregate results into a single table and then process that table with a Jupyter notebook to identify the optimal distance threshold per method and optimal method-specific parameters for t-SNE and UMAP.