Datasets:
metadata
license: cdla-permissive-2.0
task_categories:
- tabular-classification
pretty_name: hnoca_cells30
tags:
- tabular
- singlecell
- classification
- Biomedical
- Benchmark
- rna-seq
configs:
- config_name: hnoca_cells30
data_files:
- split: train
path: data/train.parquet
- split: validation
path: data/val.parquet
- split: test
path: data/test.parquet
dataset_info:
description: >-
Single-cell RNA-seq data from the HNOCA dataset (He et al., Integrated
Organoids). The dataset has been subsetted to only include the top 30 most
frequent cell types, and subsampled down to 100001 total cells while
preserving class proportions. 22000 genes were uniformly randomly selected.