FuhaiLiAiLab/OmniCellTOSG_Dataset

OmniCellTOSG是一个大规模的文本-组学信号图(TOSG)数据集，用于单细胞学习。它集成了分片的表达矩阵、图拓扑（完整/内部/PPI边缘）和文本实体元数据（名称、描述、序列），以及可选的预计算嵌入。它支持图感知预训练和下游任务，如细胞类型注释、疾病状态和性别分类。

OmniCellTOSG is a large-scale Text–Omic Signaling Graph (TOSG) dataset for single-cell learning. It integrates sharded expression matrices, graph topology (full/internal/PPI edges), and textual entity metadata (names, descriptions, sequences) with optional precomputed embeddings. It supports graph-aware pretraining and downstream tasks such as cell-type annotation, disease status, and gender classification.