scDD synthetic scRNA-seq dataset
收藏资源简介:
scDD是一个基于潜在代码的单细胞RNA测序数据集蒸馏框架,它将基础模型知识和原始数据集信息转移并浓缩到紧凑的潜在空间中,并通过生成器生成合成数据集。该数据集旨在解决单细胞RNA测序数据的高维稀疏性、批次效应噪声、类别不平衡和不断增长的数据规模带来的挑战,适用于多种数据分析任务,如疾病状态分类、发育阶段分析、解剖实体预测等,以实现跨中心知识转移、数据融合和交叉验证。
scDD is a latent code-based dataset distillation framework for single-cell RNA sequencing. It transfers and condenses foundational model knowledge and raw dataset information into a compact latent space, and generates synthetic datasets via its built-in generator. This framework addresses the challenges posed by high-dimensional sparsity, batch effect noise, class imbalance and the ever-growing data scale in single-cell RNA sequencing data, and is applicable to various data analysis tasks including disease status classification, developmental stage analysis, anatomical entity prediction and others, enabling cross-center knowledge transfer, data fusion and cross-validation.




