“基于最优传输的单细胞数据集成算法”测试数据集-小鼠脾脏单细胞多组学数据

Mouse_spleen_processed.h5ad 数据集提供了小鼠脾脏单细胞多组学数提供了 scRNA-seq 和 scATAC-seq 两种模态，且两者为非配对。scRNA 部分含 4,382 个细胞与 13,575 个基因，scATAC 部分经 SnapATAC 处理后含 3,166 个细胞与 19,410 个基因。该数据用于验证算法在多模态异质性与不平衡样本条件下的表现。该研究设计了“unbalanced matching” 实验，通过人为删除部分细胞类型评估模型在不平衡条件下的鲁棒性。The Mouse_spleen_processed.h5ad dataset provides single-cell multi-omics data of mouse spleen, including two unpaired modalities: scRNA-seq and scATAC-seq. The scRNA-seq subset contains 4,382 cells and 13,575 genes, while the scATAC-seq subset processed via SnapATAC contains 3,166 cells and 19,410 genes. This dataset is used to validate algorithm performance under conditions of multi-modal heterogeneity and imbalanced samples. The study designed an "unbalanced matching" experiment, where partial cell types were artificially removed to evaluate the model's robustness under imbalanced conditions.

搜集汇总

背景与挑战

背景概述

该数据集为小鼠脾脏单细胞多组学数据，包含非配对的scRNA-seq和scATAC-seq两种模态，分别涵盖数千个细胞和基因。它用于测试基于最优传输的数据集成算法在多模态异质性与不平衡样本条件下的鲁棒性，并设计了不平衡匹配实验以评估模型表现。