Tabula Muris (droplet) prepared for PIONEER: annotated subset with a fixed train/validation/test split

A prepared copy of the Tabula Muris droplet dataset: 55,656 cells x 23,433 genes of raw integer counts, 56 annotated cell types across 12 tissues. Contents are the standard Tabula Muris droplet annotation plus exactly one added column, obs['train_mode'], which fixes the train/validation/test split so that everyone reproduces identical numbers. Cells were restricted to those carrying a cell_ontology_class annotation and to >= 500 detected genes. Original source: The Tabula Muris Consortium, "Single-cell transcriptomics of 20 mouse organs creates a Tabula Muris", Nature 562, 367-372 (2018). doi:10.1038/s41586-018-0590-4. Processed counts and metadata: figshare doi:10.6084/m9.figshare.5968960 Raw data: GEO accession GSE109774. The source data is distributed by The Tabula Muris Consortium under the Creative Commons Attribution 4.0 International licence (CC BY 4.0, https://creativecommons.org/licenses/by/4.0/), and this derived dataset is released under the same licence. Changes made relative to the source: cells restricted to those with a cell_ontology_class annotation and >= 500 detected genes; a train_mode column added recording a fixed train/validation/test split. Expression values were not modified.