Protocol data (Python version)

We published 3 protocols illustrating how MetaNeighbor can be used to quantify cell type replicability across single cell transcriptomic datasets. The data files included here are needed to run the Python version of the protocols available on Github (https://github.com/gillislab/MetaNeighbor-Protocol) in Jupyter (.ipynb) notebook format. To run the protocols, download the protocols on Github, download the data on Figshare, place the data and protocol files in the same directory, then run the notebooks in Jupyter. Briefly:- biccn_hvg.h5ad contains a single cell transcriptomic dataset published by the Brain Initiative Cell Census Network (in AnnData format). It combines data from 7 datasets obtained in the mouse primary motor cortex (https://www.biorxiv.org/content/10.1101/2020.02.29.970558v2). Note that this dataset only contains highly variable genes.- biccn_gaba.h5ad: same dataset as biccn_hvg.h5ad, but restricted to GABAergic neurons. The dataset contains all genes common to the 7 BICCN datasets (not just highly variable genes).- hemberg.h5ad contains a merged version of 4 human pancreas single cell transcriptomic datasets made available in a standardized form by the Hemberg lab.- tasic.h5ad contains a single cell transcriptomic dataset of neurons from the mouse primary visual cortex, as published in Tasic et al. 2016.- go_mouse.mtx, go_mouse_col_labels.txt, go_mouse_row_labels.txt: mouse gene ontology annotations, stored as a one-hot encoded matrix.

本团队发布了三套操作方案，用于演示如何利用MetaNeighbor量化不同单细胞转录组数据集间的细胞类型可复现性。本文附带的数据文件可用于运行GitHub平台（https://github.com/gillislab/MetaNeighbor-Protocol）上发布的、基于Jupyter（.ipynb）笔记本格式的Python版操作方案。若需运行这些操作方案，请先从GitHub下载对应方案文件，再从Figshare下载数据集文件，将数据集与操作方案文件置于同一目录下，随后在Jupyter环境中运行笔记本文件即可。简要说明如下：
- biccn_hvg.h5ad：包含由脑计划细胞普查网络（Brain Initiative Cell Census Network）发布的单细胞转录组数据集，格式为AnnData。该数据集整合了7项小鼠初级运动皮层相关研究的转录组数据（来源链接：https://www.biorxiv.org/content/10.1101/2020.02.29.970558v2）。请注意，此数据集仅包含高度可变基因。
- biccn_gaba.h5ad：与biccn_hvg.h5ad为同一数据集，但仅保留γ-氨基丁酸能神经元（GABAergic neurons）的数据。该数据集涵盖了7项BICCN数据集共有的全部基因（而非仅高度可变基因）。
- hemberg.h5ad：整合了4项人类胰腺单细胞转录组数据集，该数据集由亨伯格实验室（Hemberg lab）以标准化格式发布。
- tasic.h5ad：包含小鼠初级视觉皮层神经元的单细胞转录组数据集，相关研究成果发表于Tasic等人2016年的论文中。
- go_mouse.mtx、go_mouse_col_labels.txt、go_mouse_row_labels.txt：小鼠基因本体（Gene Ontology）注释文件，以独热编码矩阵（one-hot encoded matrix）格式存储。