arcinstitute/Perturb-Sapiens

Perturb Sapiens是一个不断发展的AI预测单细胞扰动响应的数据库，代表了第一个人类全器官扰动细胞图谱。该数据集通过使用后训练的Stack模型（一个用于单细胞生物学的上下文学习基础模型）生成。数据来源包括Parse/OpenProblems PBMC扰动数据和Tabula Sapiens查询数据。数据集包含超过1.03亿个细胞，涵盖28个人体组织和40个细胞类别，以及111种药物条件和90种细胞因子条件。数据集结构分为药物和细胞因子两个子目录，每个子目录包含相应的.h5ad文件。数据集旨在用于药物反应预测、细胞因子信号研究和假设生成等用途。Perturb Sapiens is an evolving database of AI-predicted single-cell perturbation responses, representing the first human whole-organism atlas of perturbed cells. The dataset is generated using the post-trained Stack model, an in-context learning foundation model for single-cell biology. Data sources include Parse/OpenProblems PBMC perturbation data and Tabula Sapiens query data. The dataset contains over 103 million cells, covering 28 human tissues and 40 cell classes, with 111 drug conditions and 90 cytokine conditions. The dataset structure is divided into two subdirectories for drugs and cytokines, each containing corresponding .h5ad files. The dataset is intended for uses such as drug response prediction, cytokine signaling studies, and hypothesis generation.

搜集汇总

背景与挑战

背景概述

Perturb Sapiens是由Arc Institute发布的首个人类全器官扰动细胞图谱，包含超过1.03亿个细胞、28个人体组织和40个细胞类别，覆盖111种药物和90种细胞因子条件。该数据集基于Stack基础模型生成，用于预测单细胞扰动响应，支持药物响应预测、细胞因子信号研究和治疗靶点假设生成等生物医学研究。