* Added support for heatmaps (`workflows.overexpressed_gene_heatmap()`)
* Improved data preprocessing (`preprocess.preprocess_data()`) that allows better reporting of QC metrics
* Support for graph-based clustering using the Leiden algorithm, by running Scanpy's implementation using the Monet latent space model (`workflows.graph_based_clustering()`)
* Latent space model simplified => Now everything is built around the `PCAModel` class
* Nearest-neighbor aggregation function optimized (now uses scikit-learn `NearestNeighbor` class)
* Support for processing data produced with STARsolo's Velocyto option
* New concepts of **workflows** that provide a simplified API to streamline common analysis tasks (try `from monet import workflows as flow`), including:
* Data preprocessing (e.g., `workflows.preprocess_10x_v3()`)
* Visualization (e.g., `workflows.tsne()`)
* Clustering (e.g., `workflows.graph_based_clustering()`)