st-visium-datasets 0.1.0

No project description provided

st-visium-datasets

Spatial Transcriptomics Atlas for 10XGenomics datasets in huggingface datasets format :huggingface:

Installation

TODO

Usage

This library provides a common interface for 10XGenomics 'Gene Expression' datasets based on huggingface.

This library provides a single dataset: visium with different configurations

from st_visium_datasets import load_visium_dataset
import torch

ds = load_visium_dataset() # default lodas 'all' config
# or ds = load_visium_dataset("human")

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

torch_ds = ds.with_format("torch", device=device)
# cf: https://huggingface.co/docs/datasets/v2.15.0/en/use_with_pytorch

Availlable configs

name	number_of_spots_under_tissue	number_of_genes_detected
all	344961	1651817
human	192976	863878
human-heart	8482	40491
human-lymph-node	8074	48178
human-kidney	5936	18068
human-colorectal	9080	18077
human-skin	3458	18069
human-prostate	14334	68215
human-ovary	15153	77975
human-brain	27696	110543
human-large-intestine	6276	39440
human-spinal-cord	5624	39951
human-cerebellum	4992	17355
human-brain-cerebellum	9984	41163
human-lung	10053	36143
human-breast	38063	217985
human-colon	25771	72225
mouse	151985	787939
mouse-olfactory-bulb	2370	38664
mouse-kidney	6000	57469
mouse-brain	123254	614201
mouse-kidney-brain	2805	19399
mouse-mouse-embryo	12877	38808
mouse-lung-brain	4679	19398