metadata
license: apache-2.0
tags:
- biology
size_categories:
- 10M<n<100M
Mouse-Genecorpus-20M
Write: Keita Ito
Abstract
This repository contains the datasets of mouse-Genecorpus-20M and evaluation datasets to analysis. Mouse-Genecorpus-20M is large-scale pretraining corpus comprised of ~21 million mosue single cell transcriptomes from a broad range of tissues from publicly available data. This corpus is used for pretraining mouse-Geneformer, a pretrained transformer model that enables context-aware predictions in settings with limited data in network biology.
[bioRxiv]
Citation
If you find this repository is useful. Please cite the following references.
@article{Ito2024.09.09.611960,
author = {Ito, Keita and Hirakawa, Tsubasa and Shigenobu, Shuji and Fujiyoshi, Hironobu and Yamashita, Takayoshi},
title = {Mouse-Geneformer: A Deep Leaning Model for Mouse Single-Cell Transcriptome and Its Cross-Species Utility},
journal = {bioRxiv},
year = {2024},
URL = {https://www.biorxiv.org/content/early/2024/09/13/2024.09.09.611960}
}
How to use
You need to download datasets on your disk you want to use. After doing, you program in python as follow to use its :
from datasets inport load_from_disk
dataset = load_from_disk("/path/to/mouse-Genecorpus-20M/MLM-re_All_mouse_tokenize_dataset.dataset/")